SaaS Pricing & Health Analysis

---

A. Interpretation of Distributions

A.1 Total MRR (Aggregate Monthly Level, N=27)

Discovery → Mean MRR of PLN 15,018 vs. median of PLN 13,218 (+13.6% divergence) with strong right skew (3.34). The P10-P90 band is PLN 11,845-14,913, but the peak hit PLN 40,780 in January 2025 - far above P90, confirming a dramatic spike-and-collapse pattern. The current reading (PLN 11,628) sits below P10, meaning the business has regressed past its typical lower bound.

Implication → This is not noise. The business experienced a structural surge (likely a cohort of clients onboarding or a large contract) followed by equally structural attrition. The current MRR is at its lowest point in the dataset. The right skew is driven almost entirely by the spike period - strip that out and the underlying business is a ~PLN 12,000/month operation.

Recommendation → Stop benchmarking against the peak. The realistic baseline MRR is PLN 11,600-13,200. All financial planning, CAC budgets, and hiring should use this range, not the peak-inflated mean.

---

A.2 ARPA (Aggregate Monthly Level, N=27)

Discovery → Mean ARPA PLN 401.8 vs. median PLN 369.8 (+8.6% divergence). The skew mirrors Total MRR (3.34), indicating the same structural spike drove both. P90 is PLN 399.1 - meaning the mean is above P90, pulled entirely by outlier months.

Implication → Even at the aggregate level, ARPA has been declining. The Simpson Check confirms this: every plan's ARPA has dropped - Enterprise from PLN 2,388 to PLN 994, Professional from PLN 482 to PLN 299, Starter from PLN 452 to PLN 99. This is a universal compression, not mix-driven.

Recommendation → The ARPA decline within every segment is the single most alarming signal in this dataset. It means either: (a) newer cohorts within each plan are paying less, (b) clients are downsizing usage/seats within plans, or (c) discounting is increasing. Diagnosing which of these is the root cause is the highest-priority analytical task.

---

A.3 Churn Rate (Aggregate Monthly Level, N=27)

Discovery → Mean monthly churn 8.8% with a median of 0.0%. This extreme divergence means most months have zero churn events, but when churn occurs it is catastrophic - P90 is 38.3%. Skew is 2.47 (heavy right tail).

Implication → With ~50 active clients, a single departure in a month can register as 2%+, and losing 2-3 clients at once drives the 38%+ P90 readings. This is a small-denominator volatility problem superimposed on a real retention challenge. The Bayesian estimation (below) gives the true underlying rate.

Recommendation → Do not use month-to-month churn figures for operational dashboards - they are too noisy with N~50 clients. Use the Bayesian posterior median (3.33%/month) for all planning purposes.

---

A.4 NDR and GDR (Aggregate Monthly Level, N=27)

Discovery → NDR mean 91.2% vs. median 99.4% (-8.2% divergence, left-skewed at -2.47). GDR mean 90.2% vs. median 98.5% (-8.5% divergence). P10 values are devastating: NDR 62.0%, GDR 61.2%. The gap between NDR and GDR is minimal (~1 percentage point), meaning expansion revenue is nearly non-existent as a counterweight to contraction.

Implication → In a healthy SaaS business, NDR should exceed 100% (expansion offsetting churn). Here, NDR median of 99.4% means even in a "good" month, the company barely breaks even on existing revenue. The recent half trend is worse: NDR 89.01% → 87.71%, meaning the business is now losing ~12% of its revenue base every month in the worst periods, with almost no expansion to compensate.

Recommendation → The NDR-GDR gap of ~1pp tells us expansion mechanisms are essentially broken or non-existent. Pricing architecture must introduce built-in expansion paths (usage-based components, seat upgrades, feature gates) - this is not optional but existential.

---

A.5 Bayesian Churn Estimation (Customer-Month Level, N=951)

Discovery → 31 logo churn events across 951 active customer-months. Posterior median: 3.33%/month. 95% credible interval: [2.31%, 4.59%]. Annualized (compounded): ~33.4%/year.

Implication → This means the company replaces roughly one-third of its customer base annually. At 50 clients, that's ~17 clients lost per year. With the stated acquisition struggles, this creates a lethal dynamic: the company cannot fill the bucket as fast as it leaks. An illustrative revenue-LTV calculation (using posterior median): ARPA PLN 299.3 (current median customer-level) / 0.0333 monthly logo churn = ~PLN 8,985 per customer. This is an orientation scenario, not a margin-adjusted LTV - actual contribution-margin LTV will be lower.

The 80% credible interval (approximately P10-P90 of the posterior, roughly [2.5%, 4.4%]) implies:
- Optimistic scenario (2.5%/month, ~26% annualized): Revenue-LTV ~PLN 11,972 per customer
- Central scenario (3.33%/month, ~33.4% annualized): Revenue-LTV ~PLN 8,985
- Pessimistic scenario (4.4%/month, ~41.5% annualized): Revenue-LTV ~PLN 6,802

This ~PLN 5,000 range between optimistic and pessimistic scenarios represents the uncertainty band that should govern retention investment budgets. Any retention initiative costing less than the gap between pessimistic and central LTV (roughly PLN 2,200 per client) is likely positive-expected-value.

Recommendation → A 33.4% annual logo churn is critical-grade for B2B SaaS. Industry benchmarks for SMB SaaS tolerate 10-15% annually; mid-market targets <10%. This business is 2-3x above acceptable thresholds. Before any pricing changes, a churn root-cause analysis (exit interviews, usage pattern pre-churn, segment-specific churn rates) is mandatory. Hypothesis: the Starter segment likely drives disproportionate churn given its ARPA collapse and high client count (24 of 50).

---

A.6 LTV (N=8)

Discovery → Mean PLN 8,099 vs. median PLN 9,187 (-11.8% divergence), nearly symmetric (skew -0.10). P10-P90 range is extreme: PLN 1,038 to PLN 14,679.

Implication → With only 8 observations, this distribution has very low statistical power. The wide P10-P90 band (14x spread) reflects segment heterogeneity more than estimation uncertainty. The symmetry is somewhat coincidental at this sample size. These figures should be treated as directional only.

Recommendation → Do not use these LTV figures for CAC calculations or investment decisions. Use the Bayesian-derived revenue-LTV scenarios above (PLN 6,800-12,000 range) which have a stronger methodological foundation.

---

A.7 MRR by Plan (Aggregate Monthly, N=28)

Discovery →
| Plan | Mean MRR | Median MRR | Divergence | P10 | P90 |
|---|---|---|---|---|---|
| Enterprise | PLN 8,479 | PLN 7,931 | +6.9% | PLN 6,041 | PLN 10,672 |
| Professional | PLN 4,816 | PLN 4,327 | +11.3% | PLN 3,604 | PLN 7,707 |
| Starter | PLN 2,414 | PLN 1,393 | +73.4% | PLN 1,383 | PLN 4,772 |

Implication → The Starter plan's 73.4% mean-median divergence is the standout anomaly. This means Starter MRR was dramatically higher during some period (likely the spike) and has now collapsed to near its P10 floor. Enterprise is the most stable tier (6.9% divergence), suggesting those clients are more predictable. Note: this spread reflects volume volatility (number of clients × usage), not necessarily price discounting - we lack list_price vs. actual_price data to assess discount discipline.

Recommendation → Enterprise is carrying the business. At median PLN 7,931 out of total median PLN 13,218, Enterprise represents ~60% of revenue from only 9 of 50 clients (18%). This concentration risk is severe. Losing even 1 enterprise client represents a ~PLN 1,000/month revenue hit (using current median enterprise ARPA of PLN 1,002), equivalent to acquiring 10 new Starter clients.

---

B. Segmentation and Hidden Signals

B.1 Customer-Level ARPA Distribution: The Core Story

Discovery → All-customer ARPA: mean PLN 816.9 vs. median PLN 299.3 - a 172.9% divergence with skew of 5.37. This is an extremely right-skewed distribution where a small number of high-paying customers pull the mean far above what the typical customer pays. P10 is PLN 98.8, P90 is PLN 1,069.2 - a ~10.8x spread.

Within plans:
- Enterprise (N=9): mean PLN 2,088 vs. median PLN 1,002 (108.4% divergence)
- Professional (N=17): mean PLN 644 vs. median PLN 299 (115.2% divergence)
- Starter (N=24): mean PLN 463 vs. median PLN 99.7 (364.1% divergence, skew only 0.55)

Implication → Every tier has enormous internal heterogeneity. The Starter plan is the most striking: with 364% mean-median divergence yet only 0.55 skew, this suggests the distribution may contain more than one group (to be verified - we cannot confirm bimodality without a histogram and formal test). Hypothesis: Starter contains a cluster of clients paying ~PLN 99/month (the "true starters") and another group paying substantially more (possibly PLN 800-1,100 based on the P90 of PLN 1,069). The low skew despite massive mean-median divergence is consistent with a flat or potentially multi-group distribution rather than a single peak with a tail.

Similarly, Enterprise has 108% divergence - the median client pays PLN 1,002 but the mean is PLN 2,088, suggesting 2-3 very large enterprise accounts paying PLN 3,000+ that inflate the average.

Recommendation → The pricing structure is performing poorly at sorting customers into appropriate tiers. When a "Starter" client pays PLN 1,069 (P90) while the Enterprise median is PLN 1,002, the tier boundaries have lost their economic meaning. This is the primary pricing architecture failure to address.

---

B.2 Simpson's Paradox - Full Decomposition

Discovery → Aggregated ARPA declined by PLN 2,289 over the period. Every individual plan also declined:
- Enterprise: PLN 2,388 → PLN 994 (-58%)
- Professional: PLN 482 → PLN 299 (-38%)
- Starter: PLN 452 → PLN 99 (-78%)

This is NOT a Simpson's Paradox - it's a universal decline. Simpson's Paradox would manifest if aggregated ARPA moved in one direction while within-plan ARPAs moved in the other. Here, all arrows point the same way: down.

Implication → The decline is happening at every tier level, which rules out "mix shift toward cheaper plans" as the sole explanation. Something structural is compressing ARPA within each plan. The most plausible explanations, ranked by likelihood:

1. Seat/usage contraction within existing accounts - clients reducing their footprint while staying on the same plan
2. Newer cohorts entering at lower price points - possibly through promotional pricing or smaller initial deployments
3. Discounting escalation - though we cannot confirm without list_price data (data gap acknowledged)
4. Feature/module unbundling - if add-ons were previously included and are now optional or dropped

The Starter collapse from PLN 452 to PLN 99 is particularly telling - PLN 99 appears to be a floor/base price, suggesting that the higher historical ARPA was driven by add-ons, extra seats, or one-time fees that are no longer present.

Recommendation → Conduct a cohort-based ARPA analysis: compare the ARPA trajectory of clients acquired in each quarter. If newer cohorts start lower and stay lower, the problem is acquisition pricing. If older cohorts' ARPA is declining over time, the problem is contraction/downgrade. These require different interventions.

---

B.3 Tenure Distribution

Discovery → Mean tenure 20.6 months, median 28.0 months (-26.6% divergence, left-skewed at -0.97). P10 is 3.0 months, P90 is 28.0 months (notably, P90 = median, meaning the distribution is compressed at the top).

Implication → The left skew with P90 = median = 28 months tells us there's a core of long-standing clients (likely from the company's founding or early traction period) and a thinner tail of newer clients. The 28-month cap suggests the dataset starts ~28 months ago, so those are Day 1 clients. The P10 of 3 months means some recently acquired clients exist but are few. This is consistent with the stated acquisition difficulty: the company is not replacing churned clients with new ones at sufficient rate.

Recommendation → The surviving long-tenure clients are the most valuable strategic asset. They should receive dedicated retention attention - success management, feature previews, loyalty pricing protection (grandfathering). Losing a 28-month client is far more damaging than losing a 3-month client, both in revenue and in institutional knowledge about what the product delivers.

---

B.4 Upsell Distribution

Discovery → 38 of 50 clients (76%) have experienced some expansion. Mean upsell PLN 108.1, median PLN 62.7 (72.5% divergence, skew 1.32). P10 PLN 25.2, P90 PLN 299.2.

Implication → 76% participation in expansion sounds healthy, but the median upsell amount of PLN 62.7 is tiny relative to ARPA. For a median customer paying PLN 299/month, a PLN 62.7 expansion represents a ~21% revenue lift - spread across the client's entire lifetime, not per month. This expansion is insufficient to offset churn. The NDR data confirms this: expansion adds ~1pp to retention (NDR - GDR ≈ 1pp), when it needs to add 5-10pp to reach NDR >100%.

Recommendation → Expansion is happening but at trivially small amounts. The pricing architecture lacks meaningful step-ups. There's no "next big thing to buy." Introducing usage-based pricing components (API calls, storage, active users, contacts managed) would create natural expansion triggers without requiring a sales-driven upsell motion.

---

C. Pricing Health Assessment

C.1 Discount Variance

Discovery → We do not have list_price vs. actual_price data. The spread within plan-level MRR (e.g., Enterprise P10 PLN 6,041 to P90 PLN 10,672) reflects volume variance (number of clients × their individual spend), not price discipline.

Implication → We cannot assess discount discipline with available data. This is a critical gap.

Recommendation → Implement mandatory fields in the billing system: list_price, actual_price, discount_reason, approval_level. Without this, any pricing strategy operates blind to execution leakage.

---

C.2 Price-to-Value Alignment by Plan

Discovery →
| Plan | Current Median ARPA | Trend | Client Count | Revenue Share (est.) |
|---|---|---|---|---|
| Enterprise | PLN 1,002 | ↓ from 2,388 | 9 | ~60% |
| Professional | PLN 299 | ↓ from 482 | 17 | ~25% |
| Starter | PLN 99 | ↓ from 452 | 24 | ~15% |

Implication → The observed tier spread is 10.1x (PLN 99 to PLN 1,002). The recommended minimum is 11.1x. However, the real problem is not the spread - it's that prices have converged downward. Enterprise at PLN 994 today would have been below Starter's average a year ago (PLN 452). The entire price ladder has collapsed.

Compared to the competition scan (HubSpot Starter/Pro at PLN 100-2,500/month; typical SaaS mid-market tools PLN 500-5,000/month), the Enterprise tier at PLN 1,002 median is at the low end of competitive range for the value presumably delivered.

Recommendation → The tier structure needs rebuilding, not tweaking. Current state has destroyed the signaling value of tiers - a "Starter" and "Professional" client can pay nearly the same amount, and "Enterprise" at PLN 1,002 carries no premium connotation in the market.

---

C.3 Expansion Signal

Discovery → Median upsell PLN 62.7 (P10 PLN 25.2, P90 PLN 299.2). Expansion contributes ~1pp to NDR.

Implication → Expansion is structurally insufficient. The business has no natural revenue escalator. Every PLN of growth must come from new logos - which the company struggles to acquire.

Recommendation → The single highest-leverage pricing change is introducing a usage-based component alongside the subscription. Even a small one (e.g., PLN 5-15 per additional user/contact/transaction beyond a threshold) creates automatic expansion without sales effort. Target: raise median upsell from PLN 63 to PLN 150+ (a 2.5x improvement) to push NDR above 95%.

---

C.4 Churn Risk by Segment

Segment	Clients	ARPA (Median)	Revenue Represented	Estimated Annual Churn Risk	Key Risk Factor
Enterprise	9	PLN 1,002	~PLN 9,018/mo	~33% (base rate; likely lower for this segment*)	Concentration - loss of 1 client = ~8% total MRR
Professional	17	PLN 299	~PLN 5,083/mo	~33% (base rate)	ARPA compression suggests weakening engagement
Starter	24	PLN 99	~PLN 2,388/mo	~33% (base rate; likely higher for this segment*)	ARPA at floor, very low switching cost

Hypothesis: Segment-specific churn likely differs from the base rate. Starter clients with PLN 99/month ARPA face minimal switching costs and likely churn at higher rates. Enterprise clients with more integration and higher stakes likely churn at lower rates. We lack segment-level churn data to confirm - this is the most important segmentation analysis to run next.

Recommendation → Prioritize retention spending on Enterprise (highest revenue impact per client) and accept higher Starter churn unless those clients can be upsold. Revenue represented by Starter segment (~PLN 2,388/month) is meaningful but diffuse across 24 clients. The cost-to-serve per Starter client may not justify intensive retention efforts at PLN 99/month ARPA.

---

D. Trend and Inflection Point

D.1 Business Phase

Discovery → The data shows a clear three-act structure:
1. Growth phase (Feb 2024-Jan 2025): MRR grew from PLN 11,865 to PLN 40,780 (+244%)
2. Collapse phase (Jan 2025-Apr 2026): MRR fell from PLN 40,780 to PLN 11,628 (-71.5%)
3. Current state: MRR is now below the starting point of the dataset

Phase classification: STRUCTURAL DECLINE - confirmed by declining NDR (89% → 87.7%), accelerating revenue churn (10.9% → 12.3%), and universal ARPA compression across all tiers.

Implication → This is not a cyclical dip or seasonal pattern. The spike-and-collapse pattern is consistent with one of several scenarios:
- Hypothesis A: A large cohort of clients was acquired (possibly through a channel partner, promotion, or market event) and has since churned out
- Hypothesis B: A small number of very large contracts inflated MRR temporarily and then terminated
- Hypothesis C: The product-market fit was tested and partially invalidated - initial enthusiasm gave way to poor retention

The current trajectory, if unchanged, points toward further decline. At 12.3% monthly revenue churn and NDR of 87.7%, the business loses ~1.5% of its revenue base each month net of expansion. That implies MRR of ~PLN 9,700 twelve months from now, all else equal.

---

D.2 Repricing Window

Discovery → With MRR at its lowest point and churn accelerating, this might seem like the worst time to raise prices. But the data reveals something important: prices have already fallen dramatically (Enterprise ARPA down 58%, Starter down 78%). The current price points may actually be too low to sustain the business.

Implication → There are two repricing imperatives, and they are in tension:
1. Don't raise prices on existing clients - at 33.4% annual logo churn, adding a price increase to existing clients would accelerate departures
2. Do reprice for new clients - current acquisition pricing (if ARPA compression is cohort-driven) is not economically viable

Recommended repricing timeline:

Phase	Timeline	Action
Immediate (0-30 days)	Now	Freeze existing client prices (grandfather). Announce price stability as a retention lever.
Short-term (30-90 days)	Q3 2026	Launch new pricing structure for new clients ONLY. A/B test or sequential test on acquisition channels.
Medium-term (90-180 days)	Q4 2026	Evaluate retention impact. If stable, begin voluntary migration offers for existing clients (e.g., "lock in current price for 12 months if you sign annual").
Long-term (180-365 days)	H1 2027	Full new pricing in effect for all new contracts. Existing clients on legacy until renewal.

---

D.3 Recommended Good/Better/Best Structure

Important caveat: All prices below are illustrative pricing experiments - hypotheses, not optimal prices. Each must be validated through testing.

Current state: Three tiers exist but have lost economic differentiation (ARPA collapsed within each).

Proposed structure (for new clients only; grandfather existing clients):

Tier	Illustrative Price	Segment	Key Features	Multiplier
Good (Starter)	PLN 149/month	Solopreneurs, early-stage	Core features, limited seats (1-2), basic support	1.0x
Better (Growth)	PLN 599/month	Growing SMBs, 5-15 employees	Full features, 5-10 seats, priority support, integrations	4.0x
Best (Scale)	PLN 1,799/month	Established mid-market	Unlimited seats, dedicated CSM, SLA, API access, custom reporting	12.1x

Spread: 12.1x (above the recommended minimum of 11.1x)

Element	Detail
Segment	New clients only; existing clients grandfathered
Feature scope	Good = core only; Better = full platform; Best = platform + services + customization
Expected effect	Restore ARPA to sustainable levels; create clear upgrade path; improve price-to-value signaling
Risk	Good tier too low → attracts low-quality leads that churn; Best tier too high → conversion drops
KPIs to track	New client ARPA, conversion rate per tier, 90-day retention by tier, expansion rate from Good→Better
Guardrails	If new client 90-day churn exceeds 25% on any tier, pause and diagnose. If average deal size drops below PLN 300/month for new clients, adjust Good tier upward.
Test method	Sequential test on new clients (not A/B if traffic is low). First 90 days on new pricing → compare to historical cohort conversion and retention. No automatic migration of existing clients.

The critical innovation is the "Better" tier at PLN 599 - this fills the current gap between Starter (PLN 99) and Enterprise (PLN 1,002) and creates a natural landing zone for the Professional clients currently paying PLN 299. It also creates a clear upsell path from Good (PLN 149 → PLN 599 = 4x step) that is psychologically achievable.

---

E. Four Closing Questions

E.1 Mean vs. Median Relationship - What Does It Signal at Each Level of Analysis?

At the aggregate monthly level, the mean-median divergence tells a story of volatility and structural change. Total MRR's +13.6% divergence (mean above median) is driven by the spike months that pull the average up - but the business actually spends most of its time at or below the median. The ARPA divergence of +8.6% at this level is a milder version of the same effect. The most telling divergence is in NDR and GDR, where the mean is below the median (-8.2% and -8.5% respectively, left-skewed): this means that while the "typical" month sees near-100% retention, the bad months are catastrophically bad, dragging the average down. The mean represents the economic reality (what the P&L experiences over time); the median represents the operational reality (what most months "feel" like). The dangerous implication is that the business may feel stable month-to-month while its economic trajectory is deteriorating - the damage is concentrated in episodic churn events.

At the customer level, the divergence is far more dramatic and strategically consequential. The overall ARPA divergence of +172.9% (mean PLN 817 vs. median PLN 299) tells us that the "typical" customer pays PLN 299/month, but a handful of high-value accounts inflate the average enormously. This makes the mean nearly useless for pricing decisions - if you set prices based on mean ARPA, you'll be pricing for a customer that barely exists. The median is the anchor for pricing the core tier, while the mean-median gap represents the economic value of the premium segment. Within the Starter plan, the 364% divergence is an alarm: the median Starter client pays PLN 99.7, but the mean is PLN 463 - suggesting a few Starter clients are paying 5-10x the typical amount, which means either they're on the wrong plan (under-tiered) or they represent a usage pattern the pricing model isn't capturing properly.

---

E.2 Distribution Shape - What Dominates and Why Is It Critical for Pricing?

The dominant shape across nearly every metric is right-skewed with heavy tails (skew 2.4-5.4). At the customer level, the overall ARPA distribution has a skew of 5.37 - this is an extremely heavy-tailed distribution where a small number of clients generate revenue far above the mode. This is the classic "power law adjacent" pattern common in B2B SaaS: many small clients, a few medium clients, and a handful of large clients that disproportionately drive revenue.

This shape is critical for pricing for three reasons. First, the mode (most common price) is far below the average. A Starter client at PLN 99 is the most common customer type, but the business's economics depend on clients paying 10-30x that amount. Any pricing change that improves the PLN 99 tier by even 50% (to PLN 149) has 24x the volume impact of an Enterprise tweak. Second, the tail is where the money is, but also where the risk is. Enterprise clients (9 accounts, ~60% of revenue) represent the tail. Losing 2-3 of these accounts would collapse the business. The pricing strategy must simultaneously protect the tail (retention, grandfathering, premium service) and grow the body (move the mode rightward). Third, right-skewed distributions resist flat-rate pricing models. A single price point cannot serve both the PLN 99 mode and the PLN 2,000+ tail. The G/B/B structure exists precisely to match pricing architecture to the shape of the willingness-to-pay distribution - but only if the tiers are correctly positioned along the curve.

---

E.3 How the Distribution Is Changing Over Time - Structural vs. Noise

The most important structural change is the universal ARPA compression documented in the Simpson Check. This is not noise - it persists across all three plans and over a multi-year period. Enterprise ARPA falling from PLN 2,388 to PLN 994 (-58%) is a structural shift that fundamentally changes the economics of the business. At the peak Enterprise ARPA, 9 clients generated ~PLN 21,500/month; at current ARPA, the same 9 clients generate ~PLN 9,000/month. That PLN 12,500/month decline (annualized ~PLN 150,000) from a single tier that retained its clients is a massive structural loss.

What appears to be noise: the month-to-month churn volatility (0% in most months, 38%+ in spike months) is a small-denominator artifact, not a signal of bimodal retention. The Bayesian estimate stabilizes this at 3.33%/month. Similarly, the MRR spike to PLN 40,780 was likely an anomalous event (a large cohort or contract) rather than a structural upshift - the speed and completeness of the reversion confirms it was temporary.

The structural trajectory is clear and concerning: the business is converging toward a lower-ARPA, smaller-base equilibrium. NDR declining from 89% to 87.7% and revenue churn accelerating from 10.9% to 12.3% are trend confirmations, not noise - they reflect the compounding effect of ARPA compression (lower revenue to retain) and persistent logo churn (clients leaving). Without intervention, the equilibrium is approximately PLN 8,000-10,000/month MRR with 30-35 clients paying ~PLN 250 average - a business too small to sustain typical SaaS operating costs.

---

E.4 What Distribution Shape Favors G/B/B and What Pricing Mechanisms Lead There?

Current state: The customer-level ARPA distribution is heavily right-skewed (skew 5.37) with a dominant mode at ~PLN 99 (Starter floor), a secondary concentration around PLN 299 (Professional base), and a thin tail extending to PLN 3,000+. This is essentially a J-shaped or exponential decay distribution - many clients at the floor, rapidly declining frequency as price increases. The problem with this shape is that it concentrates most clients at the lowest willingness-to-pay, making the Good tier a gravity well that captures everyone and the Better/Best tiers feel like premium exceptions rather than natural progressions.

Target state: A G/B/B model performs optimally when the ARPA distribution approximates a roughly log-normal shape with moderate skew (1.0-2.0) and clear multi-modality aligned with tier boundaries. In practical terms, this means: ~20-25% of clients in Good (creating a floor, not a majority), ~50-60% in Better (the core revenue engine and modal tier), and ~15-25% in Best (the high-value tail that drives ARPA and profitability). The distribution should look like a bell curve on a log scale, centered around the Better tier, with clear shoulders at Good and Best. The Better tier should represent both the mode AND the median of the distribution, ensuring that the "typical" client and the "average" client are approximately the same - shrinking the mean-median divergence to below 50%.

Transition mechanism - the levers to shift the distribution:

1. Raise the floor (shift the left tail rightward). Current Starter at PLN 99 is a gravity well. Illustrative pricing experiment: raise Good tier to PLN 149 for new clients. This doesn't eliminate low-ARPA clients but moves the floor up and makes the jump to Better (PLN 599) a 4x step instead of 6x. Clients unwilling to pay PLN 149 self-select out - which improves the distribution shape and likely reduces churn (low-ARPA clients churn more). Expected effect: reduce the spike at the left edge, push the mode rightward. Risk: reduced conversion at acquisition. Guardrail: if lead-to-customer conversion drops more than 30%, revisit.

2. Create a compelling "Better" magnet. The current pricing has no natural "middle." Professional at PLN 299 is too close to Starter (3x, should be 4-5x). The illustrative Better tier at PLN 599 with meaningful feature differentiation (more seats, integrations, priority support) should be positioned as the default recommended tier. This is the single most important lever: if 50%+ of new clients land in Better instead of Good, the distribution reshapes from J-curve to bell curve. Mechanism: use anchoring (show Best first on the pricing page), make Good visibly limited ("up to 2 users"), and make Better the obvious rational choice ("most popular" badge, best value per seat).

3. Make expansion automatic, not sales-driven. The current expansion mechanism generates a pathetic PLN 63 median upsell. Usage-based pricing components (per-seat pricing beyond a threshold, per-transaction fees, storage/API limits) create continuous rightward pressure on the distribution. Every month, clients who grow naturally migrate rightward within their tier and eventually cross tier boundaries. This is the mechanism that, over time, transforms a right-skewed distribution into a log-normal one centered on Better: clients start at Good, organically expand to Better, and some continue to Best. Without this, the distribution stays anchored at the floor.

4. Protect the tail (don't compress Best downward). The ARPA collapse in Enterprise (PLN 2,388 → PLN 994) has compressed the right tail toward the center, worsening the distribution shape. The illustrative Best tier at PLN 1,799 re-establishes the right anchor. For existing Enterprise clients at PLN 994, do not force-migrate - instead, introduce premium add-ons (dedicated CSM at PLN 500/month, SLA guarantee at PLN 300/month) that give them a voluntary path to higher spend. The goal is not to move all Enterprise clients to PLN 1,799 overnight but to create reasons for them to spend PLN 1,200-2,000 over time.

5. Time-based pricing dynamics. Annual contracts at a modest discount (15-20%) preferentially convert higher-ARPA clients (they have budget approval processes that favor annual commitments). This creates a structural segmentation: annual clients pay less per month but more predictably, while monthly clients pay more per month but with higher churn risk. Over time, this shifts the distribution's center of mass upward as annual clients accumulate (they churn less) while monthly low-ARPA clients churn out.

The fundamental insight: G/B/B is not just a tier naming exercise - it only works when the distribution of customers across tiers matches the theoretical optimum (20/60/20 or similar). Currently the distribution is ~48% Starter, ~34% Professional, ~18% Enterprise - almost the inverse of what's needed. Every pricing mechanism above is designed to reshape this ratio toward ~20% Good, ~50% Better, ~25% Best over a 12-18 month horizon.

---

Analysis based on pre-computed results. All monetary values in PLN. All prices labeled as "illustrative pricing experiments" are hypotheses requiring validation. MRR figures represent monthly recurring revenue; ARR references are current MRR × 12 (run-rate). Revenue-LTV calculations use ARPA / logo churn and are not contribution-margin adjusted.

Competition Pricing Scan - SaaS Customer Acquisition Segment

Reference anchor: MRR PLN 11,628 | ARR PLN 139,530 | ~50 clients | 3-tier structure

---

1. Typical Competitors in This Segment

Given the ARR scale (PLN ~140k), tier structure, and client count (~50), this company most likely operates in one of the following SaaS verticals: CRM/sales enablement, project management, marketing automation, or B2B workflow tooling for SMB/mid-market Polish or CEE clients. The competitors below represent the competitive landscape a company at this stage typically faces:

#	Product	Description	Pricing Range (monthly, per seat or flat)
1	HubSpot Starter/Pro	CRM, marketing, sales hub; dominant in SMB SaaS	PLN 100-2,500/mo (flat tiers)
2	Pipedrive	Sales CRM, strong in CEE/Polish SMB market	PLN 70-350/mo per seat
3	Monday.com	Work OS / project management, per-seat pricing	PLN 60-450/mo per seat
4	Livespace	Polish CRM - direct local competitor, SMB focus	PLN 60-250/mo per seat
5	Asana / ClickUp	Project/task management, freemium-led growth	PLN 0-400/mo per seat

Calibration note: At ARPA median of PLN 28,036 (Enterprise), PLN 8,376 (Professional), PLN 2,778 (Starter) across the full period, this company's per-client values are meaningfully higher than typical per-seat SMB tools - suggesting either: (a) longer billing cycles, (b) multi-seat enterprise deals, or (c) a specialized vertical niche with justified premium pricing.

---

2. Pricing Benchmarks - Market Rates by Tier

Monthly Plan Benchmarks (SMB/Mid-Market SaaS, CEE/Polish market)

Tier	P10 (Budget end)	P50 (Market median)	P90 (Premium)	Notes
Free/Freemium	PLN 0	PLN 0	PLN 0	Standard entry in most categories
Starter	PLN 49/mo	PLN 149/mo	PLN 399/mo	1-3 seats or flat
Professional/Growth	PLN 299/mo	PLN 699/mo	PLN 1,499/mo	5-10 seats or feature-gated
Enterprise	PLN 999/mo	PLN 2,500/mo	PLN 6,000+/mo	Custom / negotiated

Annual Plan Benchmarks (same tiers, billed annually)

Tier	Typical Annual Discount	Effective Monthly	Full Annual Payment
Starter	15-20%	PLN 119-127/mo	PLN 1,430-1,520/yr
Professional	15-25%	PLN 524-594/mo	PLN 6,290-7,130/yr
Enterprise	10-20%	PLN 2,000-2,250/mo	PLN 24,000-27,000/yr

Calibration Against Internal Data

Internal Plan	Median Period ARPA (PLN)	Implied Monthly (assuming ~12-18 mo avg tenure)	Market P50 Monthly
Enterprise	28,036	~1,558-2,336/mo	PLN 2,500/mo
Professional	8,376	~465-698/mo	PLN 699/mo
Starter	2,778	~154-232/mo	PLN 149/mo

Observation: The company's implied pricing sits slightly below P50 on Enterprise and Professional, and at or slightly above P50 on Starter - suggesting room to increase ACV on upper tiers.

---

3. Standard Tier Structures in This Segment

What the Market Expects Per Plan

Free / Freemium (not currently offered)
- 1 seat, limited features, no integrations
- Storage/usage cap (e.g., 100 records, 5 projects)
- No support SLA
- Purpose: acquisition funnel, not a revenue tier

Starter
- Up to 3 seats (or flat fee)
- Core features only - no automation, no API access
- Email support only, 48h response SLA
- Onboarding: self-serve documentation
- Typical includes: basic reporting, 1 integration, data export

Professional / Growth
- 5-15 seats
- Full feature set minus enterprise-only modules
- Priority email + chat support, 24h SLA
- Onboarding: 1 live session included
- Typical includes: automations, API access, advanced reporting, 5+ integrations, custom fields

Enterprise
- Unlimited or high-cap seats
- Full features + custom modules, SSO, audit logs
- Dedicated CSM, SLA guarantee (99.9%+ uptime)
- Onboarding: white-glove, multi-session
- Typical includes: custom contracts, invoice billing, role-based access, data residency options

Notable Gap in Current Structure

The jump from Starter (ARPA ~PLN 2,778) to Professional (ARPA ~PLN 8,376) represents a ~3x price gap - this is a well-documented churn and expansion risk zone. Customers outgrowing Starter have no intermediate step and may churn rather than commit to 3x cost.

---

4. Pricing Levers in This Segment

Lever	Prevalence	How It's Used	Relevance to This Company
Per-seat / user	Very high	Scales revenue with team growth	High - likely relevant if B2B workflow tool
Feature gating	Very high	Locks power features to Pro/Enterprise	High - standard tier differentiation
Usage-based	Medium	API calls, records, emails sent, storage	Medium - add-on potential, not primary
Support tier	High	Response time, CSM, onboarding quality	High - Enterprise differentiator
Annual commitment discount	Very high	15-25% off monthly rate	High - improves cash flow and reduces churn
Add-on modules	Medium	Extra integrations, white-label, analytics	Medium - expansion revenue opportunity
Contract length	Medium	2-3 year deals for Enterprise	Low-medium at current ARR scale

Primary lever recommendation for this company:

The MRR decline (PLN 12,023 -> 11,628 in 60 days) suggests either churn at Starter tier or downgrades from Pro. Pricing levers alone won't fix acquisition - but restructuring the Starter entry point and adding annual billing incentives are the two highest-leverage pricing actions.

---

5. Positioning Recommendation

Competitive Gaps to Fill

• Gap 1: No freemium or free trial tier - creates acquisition friction vs. competitors offering self-serve entry
• Gap 2: No mid-tier between Starter and Professional - the PLN 2,778 to 8,376 jump (3x) likely causes hesitation or churn
• Gap 3: Enterprise pricing appears to be below market P50 by ~10-20% - leaving revenue on the table with highest-value clients
• Gap 4: No annual billing incentive clearly visible in data - MRR volatility suggests monthly-dominant billing

---

Illustrative Pricing Experiment (hipoteza do walidacji)

Disclaimer: The figures below are hypotheses for A/B testing and sales conversation validation - not final prices. Validate with 5-10 prospect conversations before implementation.

---

Proposed Tier Architecture

NEW: Free / Trial Tier
- Current price: not offered
- Proposed price: PLN 0/mo (14-day full trial OR permanent freemium with hard caps)
- Rationale: Removes acquisition friction; competitors (HubSpot, ClickUp, Asana) use this as primary growth engine. At current MRR of PLN 11,628 with declining trend, top-of-funnel volume is the critical constraint. Even a 10% free-to-paid conversion at PLN 149/mo = PLN 735/mo MRR from new channel.

---

Starter
- Current implied price: ~PLN 155-230/mo (based on median ARPA PLN 2,778)
- Proposed price: PLN 149/mo (monthly) | PLN 119/mo billed annually (PLN 1,428/yr)
- Rationale: Align with P50 market rate; introduce annual billing at 20% discount to lock in revenue and reduce churn exposure. Explicit pricing page communicates value vs. free tier.

---

NEW: Growth (mid-tier) - fills the 3x gap
- Current price: not offered (gap between Starter and Professional)
- Proposed price: PLN 399/mo (monthly) | PLN 319/mo billed annually (PLN 3,828/yr)
- Rationale: The PLN 2,778 to 8,376 median jump is a textbook "missing middle" problem. A Growth tier at PLN 399/mo gives Starter clients a natural upgrade path at ~2.6x rather than forcing a 5.4x leap to Professional. Target: 5-8 seats, core automation, priority support. Expected conversion: 20-30% of Starter clients who are currently churning at the upgrade decision point.

---

Professional
- Current implied price: ~PLN 465-700/mo (based on median ARPA PLN 8,376)
- Proposed price: PLN 799/mo (monthly) | PLN 639/mo billed annually (PLN 7,668/yr)
- Rationale: Current pricing is at or slightly below P50. A PLN 799 monthly rate positions above market median, justified by local support and product specificity. Annual billing at 20% discount creates a PLN 7,668 ACV - close to current median, reducing perceived increase for existing clients migrating to annual.

---

Enterprise
- Current implied price: ~PLN 1,558-2,336/mo (based on median ARPA PLN 28,036)
- Proposed price: PLN 2,900/mo (monthly, list price) | PLN 2,300/mo billed annually (PLN 27,600/yr) + custom negotiation floor at PLN 2,000/mo
- Rationale: Current pricing is 10-15% below P50 (PLN 2,500/mo market median). Raising list price to PLN 2,900 while offering annual at PLN 2,300 keeps effective ACV near current median for committed clients. New logo pricing should target PLN 27,600/yr ACV minimum. Dedicated CSM + SLA must be explicit deliverables at this price point to justify vs. HubSpot/Pipedrive.

---

Summary Table

Tier	Current Implied Monthly	Proposed Monthly	Proposed Annual (billed yearly)	Key Change
Free	-	PLN 0	PLN 0	NEW - acquisition lever
Starter	~PLN 190/mo	PLN 149/mo	PLN 1,428/yr (PLN 119/mo)	Explicit pricing + annual option
Growth	-	PLN 399/mo	PLN 3,828/yr (PLN 319/mo)	NEW - fills 3x gap
Professional	~PLN 580/mo	PLN 799/mo	PLN 7,668/yr (PLN 639/mo)	Modest increase + annual incentive
Enterprise	~PLN 1,950/mo	PLN 2,900/mo list	PLN 27,600/yr (PLN 2,300/mo)	Align to market P50, explicit SLA

---

Implementation Priority

1. Immediate (0-30 days): Add annual billing option to all tiers with 20% discount - zero product change required, direct MRR stabilization impact
2. Short-term (30-60 days): Design and launch Growth tier at PLN 399/mo - targets existing Starter-to-Pro conversion gap
3. Medium-term (60-90 days): Introduce free trial (14-day) for Starter and Growth - addresses top-of-funnel acquisition problem driving MRR decline
4. Ongoing: Reprice Enterprise new logo floor to PLN 27,600/yr ACV; grandfather existing clients for 1 renewal cycle

---

All prices in PLN. Competitive benchmarks based on publicly available 2024-2025 pricing for CEE/Polish market. Internal ARPA figures derived from KROK 16a source data.

Competition Scan - Product Positioning Assessment

Scope: 2024-01 - 2026-04 | All figures in PLN

---

Executive Summary

Current ARR (run-rate) stands at PLN 139,530 (MRR PLN 11,628 × 12), serving ~50 clients across three plan tiers. The business shows a classic early-stage SaaS revenue concentration pattern, with enterprise clients punching well above their weight. Three consecutive months of MRR decline (PLN 12,023 → PLN 11,628, a -3.3% drop over 60 days) signal a structural issue that requires immediate attention.

---

1. Pricing Structure - Tier Logic & ARPA Benchmarks

Tier Gap Analysis

Plan	Clients	% Revenue	ARPA Mean (period)	ARPA Median (period)
Enterprise	9	54.0%	PLN 1,141	PLN 28,036
Professional	17	30.7%	PLN 344	PLN 8,376
Starter	24	15.4%	PLN 158	PLN 2,778

⚠️ Note on ARPA figures: The arpa_mean column in the source data appears to reflect a per-transaction or per-period slice rather than total lifetime ARPA. The median values (PLN 28,036 / 8,376 / 2,778) are more representative of actual customer value across the period and are used as primary references below.

Key Observations

The Enterprise→Professional gap is dangerously wide. The median ARPA jump from Professional (PLN 8,376) to Enterprise (PLN 28,036) represents a 3.3× step-up with no intermediate tier. For a prospect outgrowing Professional, the next available option is a ~PLN 20K commitment increase. This is a classic deal-blocker.

Starter→Professional gap is proportionally healthy (PLN 2,778 → PLN 8,376, ~3×), suggesting the lower end of the ladder is reasonably structured.

SaaS benchmark context at ARR ~PLN 140K (~35 PLNK PLN equivalent):
- Typical early-stage B2B SaaS at this ARR runs 3-5 tiers with ARPA multiples of 2-4× between adjacent tiers
- A missing "Growth" or "Business" tier in the PLN 12,000-20,000 ARPA range is a textbook gap at this scale
- 100% monthly billing (see Section 3) suppresses effective ARPA vs. peers offering annual discounts

---

2. Revenue Concentration Risk

The Enterprise Paradox

9 clients (18% of base) → PLN 237,408 (54% of total transaction value)

Nine enterprise clients represent more than half of all revenue generated over 28 months. This is a high-concentration structure typical of pre-product-market-fit SaaS businesses that have landed a few large accounts before building a scalable mid-market motion.

Scenario modeling:

Churn Scenario	Clients Lost	Revenue Represented
Single largest enterprise client	1	~PLN 32,368 (ARPA max)
Top 3 enterprise clients	3	~PLN 85,000-97,000 (est.)
Full enterprise segment	9	PLN 237,408

Losing 2 enterprise clients could represent revenue equivalent to the entire Professional segment. This is not flagged as "revenue represented by segment (not an expected loss)" without churn probability data - but the structural exposure is significant and warrants a dedicated retention program.

Starter Segment - Volume Without Value

24 Starter clients (48% of base) contribute only PLN 67,593 (15.4%) of total transaction value. At median ARPA of PLN 2,778, this segment is likely high-touch relative to its revenue contribution unless the product is genuinely self-serve at this tier.

---

3. Pricing Power - What the Monthly-Only Mix Reveals

Monthly billing: 100% of clients (50/50)
Annual billing: 0%

This is the single most revealing data point in the dataset. Zero annual contracts across all tiers, including Enterprise, signals:

Implications

Signal	Interpretation
No annual lock-in at Enterprise	Clients are either not committed enough to prepay, or the product hasn't offered/pushed annual plans
Monthly churn risk at all tiers	Every client is one cancellation away from immediate revenue loss
Suppressed ARR quality	Monthly SaaS ARR is typically valued at 4-6× by investors vs. 8-12× for annual-contract ARR
Weak negotiating position vs. competitors	Competitors offering 15-20% annual discounts make month-to-month look expensive over time

MRR decline context: The 3-month MRR slide (-PLN 395, -3.3%) in a 100% monthly business means actual client churn or downgrades are already occurring - there is no annual contract buffer to absorb this. The client count held at 32 in March-April 2026, so the MRR drop is driven by downgrades or discount creep rather than pure churn.

---

4. Competitive Gaps - Where Rivals Can Attack

Gap 1: The Mid-Market Vacuum (PLN 10,000-20,000 ARPA)

Between Professional (PLN 8,376 median) and Enterprise (PLN 28,036 median), there is no product home for a growing company. A competitor offering a "Business" or "Scale" tier at PLN 12,000-18,000 ARPA can intercept clients who are outgrowing Professional but not ready to justify Enterprise pricing. This is the highest-probability competitive entry point.

Gap 2: No Commitment Incentive = Easy Price Competition

With 100% monthly billing, any competitor offering a 15-20% annual prepay discount immediately appears cheaper on a 12-month TCO basis. There is zero switching cost barrier from a contractual standpoint. A well-funded competitor can simply undercut on annual pricing and erode the Starter and Professional base.

Gap 3: Starter Segment Loyalty Is Untested

24 Starter clients contributing only 15.4% of revenue, all on monthly plans, represent a segment with low financial commitment and unknown product stickiness. Competitors with freemium or low-cost entry tiers can poach these clients with minimal friction. Without engagement data (the cluster profiles were empty in the dataset), it is impossible to assess actual stickiness.

Gap 4: Enterprise Relationships Are Informal

9 enterprise clients on monthly contracts - with no annual agreements - suggests these relationships may be more informal than their revenue contribution warrants. A competitor with a dedicated enterprise sales motion (QBRs, success managers, multi-year contracts) could formalize what is currently an informal relationship and lock clients in before this product does.

---

5. Quick Wins - Concrete Recommendations

QW-1: Launch an Annual Plan with a 15-20% Discount - Target Enterprise First

Action: Introduce annual prepay pricing across all tiers. Priority: convert the 9 enterprise clients.

Mechanics:
- Annual Enterprise at 15% discount = PLN ~23,800 median (vs. PLN 28,036 monthly × 12)
- Client saves ~PLN 4,200/year; you lock in ~PLN 214K in committed ARR from enterprise alone
- Even converting 3 of 9 enterprise clients to annual would materially improve ARR quality and reduce the impact of the current MRR slide

Why now: The MRR trend is negative. Annual conversions create a contractual floor that stops the bleed.

---

QW-2: Introduce a "Growth" Tier at PLN 1,200-1,800/month (PLN 14,400-21,600 ARPA)

Action: Create an intermediate tier positioned between Professional and Enterprise to capture the upgrade gap.

Mechanics:
- Target: Professional clients at the ARPA max (PLN 9,688) who need more but can't justify Enterprise
- Price point: PLN 1,400-1,600/month monthly, PLN 1,200-1,350/month annual
- Feature differentiation: 1-2 Enterprise features (e.g., advanced reporting, priority support, higher API limits) without full Enterprise onboarding cost

Revenue impact: If only 4 of the 17 Professional clients upgrade to a PLN 1,500/month Growth tier, that adds PLN 6,000 MRR - more than recovering the recent MRR decline and pushing ARR above PLN 210K.

---

QW-3: Implement a Starter-to-Professional Conversion Campaign with Usage-Based Triggers

Action: Use product usage signals (API calls, storage, reports generated - data fields present in the cluster schema) to identify Starter clients approaching plan limits and trigger an upgrade prompt.

Mechanics:
- Clients hitting >80% of Starter limits receive an in-app prompt + outreach offering Professional at 10% discount for first 3 months
- Frame as "you're growing out of Starter" rather than a sales pitch
- Pair with an annual lock-in offer to convert upgraded clients immediately

Why this matters: The 24 Starter clients represent PLN 67,593 in total transaction value - the smallest segment by revenue. Moving even 6 of 24 to Professional at PLN 8,376 median ARPA would add ~PLN 33,500 in incremental transaction value and increase MRR by approximately PLN 2,800-3,200/month, a 24-28% MRR lift from a segment currently contributing disproportionately little.

---

Summary Scorecard

Area	Risk Level	Priority
Pricing tier gap (missing mid-tier)	🔴 High	Immediate
Revenue concentration (9 enterprise = 54%)	🔴 High	Immediate
100% monthly billing / no lock-in	🔴 High	Immediate
Starter engagement & retention	🟡 Medium	30-60 days
Competitive mid-market exposure	🟡 Medium	60-90 days

Bottom line: The product has a working revenue foundation but is structurally exposed on three fronts simultaneously - concentration, no contractual lock-in, and a missing mid-market tier. The 3-month MRR decline is a leading indicator, not a lagging one. The three quick wins above address all three exposures and are executable without a pricing overhaul.

SaaS Usage Data Analysis - Key Observations & Recommendations

Disclaimer: All conclusions below are hypotheses for validation, not proven facts. Correlations ≠ causation. Revenue figures reference total_transaction_value (cumulative transaction sum over the observed period), not MRR directly.

---

1. Key Observations (5)

O-1: Extreme Usage Dispersion - Distribution May Contain More Than One Group (To Be Verified)

active_users ranges from 1 to 88 (mean 17, median 8), and api_calls_monthly from 137 to 98,647 (mean 11,508, median 2,117). The gap between mean and median in both variables is very large, suggesting the distribution may contain more than one behavioral group - but this requires histogram inspection and formal testing before drawing conclusions. High-usage accounts likely sit almost entirely in ENTERPRISE (9 clients, 18%), while the STARTER majority (24 clients, 48%) may be significantly underutilizing the platform.

O-2: Storage and Active Users Are the Strongest Revenue Correlates

storage_gb (r = 0.919) and active_users (r = 0.832) correlate most strongly with total_transaction_value. This is consistent with a pricing model where seat count or storage is a primary value driver. These two metrics are the best candidates for value metric alignment in pricing redesign - if revenue scales with them, pricing tiers should too.

O-3: 100% Monthly Billing - Zero Annual Contract Diversification

most_popular_time_period = monthly for all 50 clients (100%). This is a structural risk factor. Given the context of a negative MRR trend, there is no contractual floor whatsoever. Every client is one billing cycle away from churning. The competition context (QW-1) specifically flags annual conversion as the highest-priority quick win - these data confirm that urgency.

O-4: Discount Usage Is Low and Weakly Correlated with Revenue

Mean discount is 6.1% (std 7.1%), with 25th percentile at 0%. The correlation of discountPercent with total_transaction_value is essentially flat (r = 0.077). This suggests discounting is not currently being used strategically to drive upsell or retention - or that discounts are distributed without a clear commercial logic tied to account size or plan tier.

O-5: Support Ticket Load Is Moderate but Non-Trivial

support_tickets_ytd mean is ~5, max is 10, and it has a moderate positive correlation with revenue (r = 0.319). Higher-revenue accounts generate more tickets - which may reflect higher engagement and complexity, not dissatisfaction per se. However, without ticket resolution or CSAT data, it is impossible to determine whether this represents a churn risk or a normal usage pattern. This requires further investigation.

---

2. Business Hypotheses (5)

H-1: If a client's active_users count is ≤ 3 and api_calls_monthly is ≤ 642 (bottom quartile on both), then they are at elevated churn risk in the next 90 days, because low engagement indicates the product has not been embedded into daily workflows, making switching costs low.

H-2: If the 9 ENTERPRISE clients are offered an annual plan with a 15% discount (per QW-1), then at least 3 will convert, because their total_transaction_value and usage depth (likely high storage_gb and active_users) create mutual lock-in incentives - and the PLN saving (~PLN 4,200/year) is material at that price point.

H-3: If storage_gb is the strongest revenue correlate (r = 0.919), then introducing storage-based pricing tiers or overage fees would unlock incremental revenue from high-storage accounts currently on STARTER or PROFESSIONAL plans, because those clients are already extracting disproportionate value relative to their plan price.

H-4: If STARTER clients (48% of base, likely low active_users and api_calls_monthly) are not shown a clear upgrade trigger tied to a specific usage threshold, then STARTER → PROFESSIONAL upgrade conversion will remain low, because without a visible "you're outgrowing your plan" signal, there is no internal driver to prompt the upgrade conversation.

H-5: If discountPercent has near-zero correlation with revenue (r = 0.077) and is distributed without a clear tier or usage logic, then current discounting is eroding margin without materially improving retention or expansion, because there is no evidence in the data that higher discounts are associated with larger or more engaged accounts.

---

3. Columns Most Correlated with Churn or Growth

Column	Correlation (r) with Revenue	Role	Priority
storage_gb	0.919	Growth signal - value metric proxy	🔴 High
active_users	0.832	Growth signal - also potential churn leading indicator if trending down	🔴 High
api_calls_monthly	0.821	Growth signal - product embeddedness proxy	🔴 High
support_tickets_ytd	0.319	Ambiguous - could be engagement or friction (needs CSAT)	🟡 Medium
last_login_days_ago	0.009	Potential churn leading indicator - low correlation with revenue may indicate it varies independently of plan size	🟡 Medium (monitor trend)
discountPercent	0.077	No revenue signal - review discount policy	🟢 Low

Churn risk flag to build: Clients with last_login_days_ago > 20 AND active_users ≤ 3 AND api_calls_monthly < 642 represent a high-priority churn intervention segment - validate this with historical churn data before automating alerts.

---

4. Concrete Product & Sales Actions (3)

A-1: Launch Annual Plans Immediately - Target Enterprise and High-Usage Professional Accounts

What: Introduce annual prepay options at a 15-20% discount, starting with the 9 ENTERPRISE clients and PROFESSIONAL clients with total_transaction_value > PLN 8,000 (top quartile).
How: Direct outreach by account owner within 30 days. Frame it as a cost-saving offer, not a retention tactic.
Why now: 100% monthly billing + negative MRR trend = no contractual floor. Even 3 annual conversions create a meaningful committed ARR baseline and reduce short-term churn exposure.

A-2: Implement Usage-Based Upgrade Triggers for STARTER Clients

What: Define in-product thresholds based on active_users (e.g., > 5), api_calls_monthly (e.g., > 2,000), or storage_gb (e.g., > 25 GB) that automatically surface an upgrade prompt or trigger a sales alert.
How: In-app banner + automated email sequence + CS outreach for accounts that breach 2+ thresholds simultaneously.
Why: 48% of clients are on STARTER. The data shows active_users and api_calls_monthly scale strongly with revenue - clients approaching PROFESSIONAL-level usage on a STARTER plan represent the highest-probability upsell targets.

A-3: Audit and Restructure Discount Policy

What: Map current discountPercent values against plan type, total_transaction_value, and tenure. Identify whether discounts cluster in any revenue or usage segment meaningfully.
How: Cross-tabulate discountPercent by plan and total_transaction_value quartile. If discounts are concentrated in low-revenue STARTER accounts, this is margin erosion with no strategic benefit.
Why: r = 0.077 between discount and revenue means discounts are not driving upsell or expansion. Introduce a discount policy where concessions are tied to annual commitment (QW-1), volume, or explicit expansion milestones - not ad hoc.

---

5. Direct Response to User Context: Customer Acquisition Struggle

The data provided is usage/retention data, not acquisition funnel data - so we cannot directly diagnose the acquisition problem from this dataset. However, the following indirect signals are relevant:

Concern	Data Signal	Implication
Negative MRR trend	100% monthly billing, wide STARTER base (48%)	Retention risk is compounding any acquisition shortfall - fix the contractual floor first
Competition pricing pressure	Mean discount only 6.1%, near-zero revenue correlation	Current discounting is not competitive enough to win deals, nor structured enough to retain
STARTER-heavy base	24/50 clients on lowest tier	Either acquisition is bringing in under-qualified leads, or onboarding is failing to convert trial/starter users to higher-value plans
No annual clients	most_popular_time_period = monthly (100%)	Competitors offering annual plans with discounts (HubSpot, etc.) will systematically win deals where the buyer is cost-sensitive and values predictability

Core hypothesis for the acquisition problem: The company may be acquiring clients at STARTER price points that are too low to sustain growth, while failing to create the upgrade path or annual commitment structures that would improve unit economics. Fixing retention and upsell mechanics (A-1, A-2 above) is a prerequisite to making new acquisition economically viable - otherwise the leaky bucket dynamic will persist regardless of acquisition volume.

SaaS KMeans Cluster Analysis - Business Interpretation

Methodological note: The clusters below are based on total_transaction_value (sum of all transactions per customer over the analyzed period), not MRR or ARR. All labels, descriptions, and recommended actions are hypotheses requiring validation against additional behavioral and financial data. Clusters reflect revenue-based groupings, not confirmed behavioral segments.

---

Cluster Profiles at a Glance

Cluster	N	Total Transaction Value (median)	Avg Transaction Value (median)	Transaction Count (median)	Active Users (median)	API Calls/mo (median)
0	9	28 035.68 PLN	1 002.00 PLN	28	61	39 817
1	27	2 794.79 PLN	99.81 PLN	28	5	923
2	14	3 207.60 PLN	1 069.20 PLN	3	4.5	750.5

---

Cluster 0 - Hypothetical Label: "High-Value Engaged Accounts"

⚠️ Hypothesis: these accounts appear high-value and operationally active - requires validation that usage metrics reflect genuine organizational adoption, not automated/technical usage.

Description

Cluster 0 contains 9 customers with the highest median total transaction value (28 035.68 PLN) and by far the highest operational footprint - 61 active users, ~39 800 API calls/month, and 331 GB storage. They are on the enterprise plan, transact frequently (28 transactions), and logged in relatively recently (4 days ago median). The elevated support ticket count (8 YTD) may reflect genuine complexity of use or onboarding friction at scale - this distinction matters for retention strategy.

Business Hypotheses to Test

1. H1 - API volume drives plan stickiness: Customers with >10 000 API calls/month have significantly lower churn probability than those below this threshold, because deep API integration raises switching costs.
2. H2 - Support tickets signal expansion friction, not dissatisfaction: The higher ticket count (8 vs. 2 for Cluster 1) correlates with growth events (new team onboarding, feature adoption), not with churn intent - testable via CSAT/NPS comparison.
3. H3 - High active-user counts precede contract expansion: Accounts crossing 50+ active users are more likely to upgrade contract scope within the next 6 months.
4. H4 - Storage consumption is an upsell trigger: Storage near or above 300 GB correlates with willingness to pay for tiered storage add-ons.

Recommended Actions (all hypothetical - validate before scaling)

Priority	Action	Rationale
Retention	Assign dedicated Customer Success Manager; run quarterly Executive Business Reviews	Protect highest-revenue cohort; validate whether relationship coverage reduces ticket escalations
Upsell	Propose storage expansion and advanced API rate-tier add-ons	High storage + API usage suggests headroom for consumption-based upsell
Churn prevention	Set automated health-score alert if API calls drop >30% MoM or active users decline	Early warning that integration is being unwound
Expansion	Co-develop case studies / reference programs	Leverage account size for social proof while deepening relationship

---

Cluster 1 - Hypothetical Label: "High-Frequency Low-Ticket Accounts"

⚠️ Hypothesis: the combination of many transactions at very low average value and minimal usage metrics may indicate either a small-team use case or underutilization of the starter plan - requires validation.

Description

Cluster 1 is the largest group (27 customers, 54% of the base) with a median total transaction value of 2 794.79 PLN, achieved through the same number of transactions as Cluster 0 (28) but at a median average transaction value of only 99.81 PLN. Active users (5), API calls (923), and storage (18.3 GB) are all low. Last login was 14 days ago and support tickets are minimal (2 YTD). This pattern is consistent with starter-plan accounts that transact regularly but operate at a small scale - though whether this reflects the customer's actual size or plan under-adoption is unknown.

Business Hypotheses to Test

1. H1 - Low avg transaction value reflects plan ceiling, not low willingness to pay: A meaningful share of Cluster 1 customers have organizational revenues or team sizes that would justify a higher plan, but they have never been actively qualified or presented with an upgrade path.
2. H2 - 14-day login recency is a leading churn indicator in this segment: Compared to Cluster 0 (4 days), the longer gap may signal lower habit formation; churn rate for accounts with last login >21 days is hypothesized to be significantly higher.
3. H3 - Zero discount (discountPercent = 0) does not reflect price sensitivity: These customers have never been offered a discount - their price sensitivity is untested and conflating "no discount given" with "price-insensitive" would be a measurement error.
4. H4 - Transaction count similarity to Cluster 0 masks product usage depth: Equal transaction counts but 40× lower API calls suggests Cluster 1 uses a narrow feature subset; feature adoption campaigns may increase engagement and reduce churn.

Recommended Actions (all hypothetical - validate before scaling)

Priority	Action	Rationale
Retention	Launch automated in-app engagement nudges for accounts with last login >10 days	Test whether reducing login gap decreases churn in this cohort
Upsell	Run ICP (Ideal Customer Profile) qualification for the top 20% by transaction count	Identify accounts where growth potential justifies upgrade conversation
Churn prevention	A/B test a "feature discovery" email sequence targeting underused features	Validate H4 - does feature breadth adoption change retention rate?
Pricing experiment	Test a targeted upgrade offer (e.g., starter → growth plan) with a time-limited incentive on a random 30% subsample	Measure conversion rate before rolling out broadly

---

Cluster 2 - Hypothetical Label: "Infrequent High-Ticket Accounts"

⚠️ Hypothesis: the low transaction count combined with high average transaction value and elevated support tickets may indicate project-based or episodic purchasing behavior - requires validation of contract structure and use-case type.

Description

Cluster 2 contains 14 customers on the starter plan with a median total transaction value of 3 207.60 PLN - slightly above Cluster 1 - but achieved through only 3 transactions at a high average value of 1 069.20 PLN each. This is structurally very different from Cluster 1 (28 transactions at ~100 PLN each). Active users (4.5) and API calls (750) are low and similar to Cluster 1, but support tickets (7 YTD) are the second highest, and reports generated (27.5) are notably higher than Cluster 1 (20). The 5% median discount is the only cluster where discounting appears. The behavioral pattern is consistent with episodic or project-driven purchasing, but this is unconfirmed.

Business Hypotheses to Test

1. H1 - Transaction pattern reflects project-based purchasing cycles: Cluster 2 customers buy in large, infrequent bursts (e.g., annual contracts, project kick-offs) rather than recurring small transactions; churn risk spikes sharply in periods between purchases.
2. H2 - High support ticket volume relative to low usage indicates onboarding/configuration friction: 7 tickets with only 3 transactions and 4.5 active users suggests each purchase triggers disproportionate support demand - resolving this could reduce CAC-to-serve cost.
3. H3 - The 5% discount is driving purchase decisions, not loyalty: Cluster 2 is the only group receiving discounts; if removed, conversion or renewal rates may fall - discount dependency vs. value-driven retention should be tested.
4. H4 - High reports_generated relative to low API usage suggests a reporting/analytics use case: If confirmed, these customers may respond to a reporting-focused feature tier or add-on rather than a full plan upgrade.
5. H5 - Starter plan is misaligned with actual value delivered: Average transaction value (1 069.20 PLN) is comparable to Cluster 0's enterprise transactions yet the plan is "starter" - plan-to-value misalignment may suppress NPS and increase churn.

Recommended Actions (all hypothetical - validate before scaling)

Priority	Action	Rationale
Churn prevention	Map transaction dates to calendar; set re-engagement triggers 60 days before expected next purchase window	Validate whether proactive outreach before gap periods reduces churn
Retention	Conduct support ticket root-cause analysis for this cluster	Identify whether tickets cluster around specific features/events; address systematically before scaling acquisition
Upsell	Test a reporting-focused plan or add-on pitch (hypothesis H4) with 50% of the cluster	Measure whether value framing around reports improves conversion vs. generic upgrade offer
Discount strategy	A/B test removal or reduction of the 5% discount on renewal for a subsample	Validate H3 - is the discount driving retention or merely being consumed as default margin erosion?

---

Cross-Cluster Observations & Overarching Hypotheses

These observations span all clusters and should be prioritized for validation given their potential portfolio-level impact.

1. Monthly billing dominance: All three clusters show top_period = monthly. It is worth testing whether annual prepay offers targeted at Cluster 0 and the top quartile of Cluster 2 would improve cash flow and reduce churn without significant discount cost.
2. Cluster size imbalance (9 / 27 / 14): Cluster 1 represents 54% of accounts but likely a small share of total revenue. Before investing retention resources proportionally, validate revenue concentration - the Cluster 0 accounts (18% of customers) likely represent a disproportionate share of total PLN revenue.
3. Discount absence in the largest cluster: 27 Cluster 1 accounts carry 0% discount. This could reflect good margin discipline or a missed lever for conversion/expansion - unknown without price-sensitivity testing.
4. most_popular_time_period_enc = 0 across all clusters: All clusters share the same encoded time period. This uniformity should be verified - it may indicate a data artifact or encoding issue rather than a genuine behavioral signal.

---

Analysis prepared for internal hypothesis generation. All segments, labels, and action recommendations require validation through controlled experiments, cohort analysis, and qualitative customer interviews before operationalization.