# Retention Analysis — Project Management SaaS (Marketing Agencies, 15–200 Staff) ## 1. Data Inventory & Confidence Assessment **Supplied for this analysis:** - A single aggregate **logo churn rate: 2.8%/month**, described as an average over the trailing 12 months (not a month-by-month series). - A single aggregate **revenue churn rate: 1.9%/month**, described as net of expansion revenue ("expansion partially offsets logo loss"). - A **customer base distribution** by company-size tier: 60% at 15–50 staff, 30% at 50–100 staff, 10% at 100–200 staff. - A **qualitative signal**: churn is concentrated in the 15–50 staff tier. - A **reference to a cohort table** (Month 1 through Month 12 retention, for Jan–Dec signup cohorts of last year) — the *existence* of this table was described, but the actual per-cohort, per-month retention percentages were not included in the data provided for this analysis. **Not supplied:** absolute MRR, expansion $, contraction $, or churned $ figures; per-segment churn or retention rates; per-cohort retention percentages; usage logs; support-ticket data; NPS; feature-adoption data; pricing-tier movement data; account-level data of any kind. **What this permits:** - A precise logo retention/churn figure (Section 2). - A net revenue retention figure *derived by inversion* of the supplied net revenue churn rate, with the caveat noted in Section 2 (Section 2). - A qualitative, base-share-weighted view of where churn is said to concentrate (Section 4, partial). - One directly observed churn-concentration pattern and one calculated revenue-dynamics pattern (Section 5, limited). **What this does not permit:** - A real cohort retention curve — no actual cohort percentages were supplied (Section 3 marked `<NEEDS-INPUT>`). - Quantified segment-level churn or retention rates — only base-share percentages and a qualitative concentration note exist (Section 4 marked `<NEEDS-INPUT>` for the quantitative rows). - Any account-level leading indicator — no usage, engagement, or support signal exists in the supplied data (Section 6 marked `<NEEDS-INPUT>`). - Any churn driver beyond segment concentration and the logo/revenue churn differential — no usage, ticket, or adoption signals exist to support additional hypotheses (Section 5). Nothing past this section references a metric type not confirmed available here. ## 2. Core Retention Metric Definitions & Calculations **Logo Churn Rate** = Customers lost in period ÷ Customers at start of period. Supplied trailing-12-month average: **2.8%/month** *(Supplied)*. **Logo Retention Rate** = 100% − Logo Churn Rate = 100% − 2.8% = **97.2%/month** *(Calculated)*. **Net Revenue Churn Rate** = Revenue lost to cancellations and downgrades, net of expansion revenue from existing accounts, ÷ Starting revenue for period. Supplied value: **1.9%/month**, explicitly described as already netting out expansion *(Supplied)*. **Net Revenue Retention (NRR)** = 100% − Net Revenue Churn Rate = 100% − 1.9% = **98.1%/month** *(Calculated)*. *Limitation:* this is derived by inverting the supplied net revenue churn figure, not by applying the full formula NRR = (Starting MRR + Expansion − Contraction − Churned MRR) / Starting MRR. Absolute Starting MRR, Expansion $, Contraction $, and Churned MRR were not supplied, so the components behind the 1.9% figure cannot be independently verified or broken apart. This substitution is named here rather than presented as a full NRR calculation. **Logo–Revenue Churn Differential** = 2.8% − 1.9% = **0.9 percentage points/month** *(Calculated)*. This gap indicates that expansion revenue among retained accounts is offsetting a meaningful share of the revenue that logo losses alone would imply, though the dollar magnitude and the segment(s) generating that expansion are not known from the supplied data. **Illustrative compounded annualized figures** *(Calculated, with explicit projection caveat)*: - Compounding the average monthly logo retention rate across 12 months: 0.972¹² ≈ **71.1%** implied annual logo retention (≈28.9% implied annual logo churn). - Compounding the average monthly NRR across 12 months: 0.981¹² ≈ **79.4%** implied annual net revenue retention. These two figures assume the supplied monthly rate held constant across all 12 months, which the aggregate figure alone cannot confirm — if the rate varied month to month (e.g., worsened or improved over the year), the real annual outcome would differ from this projection. This is precisely the ambiguity the actual cohort table (Section 3) would resolve, and its absence is the single largest source of uncertainty in this report. BUSINESS_MODEL is subscription, so no usage-decay metric is required in this section. ## 3. Cohort Retention Curve Analysis `<NEEDS-INPUT: cohort-retention-percentages>` A cohort table (Month 1 through Month 12 retention, for Jan–Dec signup cohorts) was referenced as existing, but the actual retention percentages for that table were not included in the data provided for this analysis. Building a Month 1 / Month 3 / Month 6 / Month 12 curve would require inventing plausible-looking numbers, which this analysis does not do. No curve, trend statement, or flattening/declining judgment can be made until the real per-cohort figures are supplied. ## 4. Segment-Level Retention Breakdown | Segment | % of Customer Base | Retention Rate | Trend Direction | Notable Pattern | |---|---|---|---|---| | 15–50 staff | 60% *(Supplied)* | `<NEEDS-INPUT: segment-level-churn-rates>` | `<NEEDS-INPUT: segment-level-trend-data>` | Flagged as the churn concentration point *(Supplied, qualitative)* | | 50–100 staff | 30% *(Supplied)* | `<NEEDS-INPUT: segment-level-churn-rates>` | `<NEEDS-INPUT: segment-level-trend-data>` | Not flagged as a concentration point | | 100–200 staff | 10% *(Supplied)* | `<NEEDS-INPUT: segment-level-churn-rates>` | `<NEEDS-INPUT: segment-level-trend-data>` | Not flagged as a concentration point | Only the base-share distribution and a qualitative concentration signal were supplied. No segment carries an actual churn or retention percentage, and no per-segment trend data exists. All blended figures in Sections 1–2 (2.8% logo churn, 1.9% net revenue churn) are averages across the full base and may be masking a materially worse or better rate within any one tier — most plausibly the 15–50 tier, given the flagged concentration, but this cannot be quantified from what was supplied. ## 5. Churn Driver Diagnosis **Driver A — Small-Tier Concentration** Churn is reported as concentrated in the 15–50 staff tier, which also represents 60% of the customer base *(Supplied, directly observed pattern)*. Evidence strength: *directly observed pattern* — the concentration itself is stated in the data, though its magnitude relative to the tier's base share is not quantified (see Section 4 gap). **Driver B — Expansion Offset Imbalance** The 0.9-point gap between logo churn (2.8%) and net revenue churn (1.9%) shows that expansion revenue from retained accounts is offsetting part of the revenue impact of lost logos *(Calculated, from Section 2)*. Evidence strength: *correlated but not confirmed causal* — it is not established from the supplied data which segment(s) are generating this expansion, so the inference that it originates disproportionately from the larger, more stable tiers (50–200 staff) is a reasonable read of the pattern but not a confirmed one. No additional churn driver is included here. Common SaaS-industry explanations for smaller-account churn — per-seat pricing sensitivity at lower headcounts, weaker onboarding follow-through, competitive displacement, budget-cycle turnover — may well apply, but none has a traceable signal in the supplied data. These are carried to Section 9 as questions to investigate, not asserted as findings. ## 6. Leading Indicators of At-Risk Accounts `<NEEDS-INPUT: account-level-engagement-data>` No account-level usage, login-frequency, feature-adoption, or support-ticket data was supplied. With only two blended monthly rates and a coarse base-share segment split, no leading indicator can be defined with any data-derived threshold. Stating "low engagement" or "reduced usage" as an at-risk signal without a threshold computed from this business's own data would not meet the bar this analysis holds itself to, so none is offered. ## 7. Benchmark Context (Generic, Not Company-Specific) *The following are widely cited, general industry reference points. They are not derived from this company's data and should be read only as orientation, given the wide variation across pricing tiers, customer sizes, and go-to-market motions.* - For SMB-oriented, per-seat B2B SaaS, monthly logo churn in roughly the **1–3%** range is commonly discussed as healthy, with **3%+/month** more typical of lower-priced or lower-touch segments *(Generic benchmark)*. - Net Revenue Retention above **100%** is commonly cited as a target for enterprise-weighted SaaS businesses; **90–100%** is often described as an acceptable range for SMB-weighted pricing tiers *(Generic benchmark)*. - These ranges vary enormously by ICP, price point, and contract length, and are offered here purely as generic context, not as a target derived from this company's own performance. ## 8. Retention Improvement Recommendations **Recommendation 1 — Segment-targeted retention motion for the 15–50 staff tier** *(tied to Driver A — Small-Tier Concentration, Sections 4–5)* Because this tier holds 60% of the base and is the reported concentration point for churn, a dedicated onboarding check-in cadence and win-back workflow aimed specifically at 15–50 staff accounts should reduce this tier's disproportionate contribution to overall logo loss. *Falsifiable within one month:* begin tagging cancellations by segment now; at the next monthly cycle, check whether the 15–50 tier's share of churned logos is at, above, or below its 60% base-share. **Recommendation 2 — Instrument segment-level and cohort-level churn tracking** *(tied to the Section 3 and Section 4 data gaps)* The inability to see per-segment or per-cohort retention percentages currently blocks confirmation of Driver A's magnitude and blocks any cohort curve analysis. Building this instrumentation is a prerequisite to knowing whether the concentration is a durable structural pattern or an artifact of the blended view. *Falsifiable within one month:* whether a segment-tagged and cohort-tagged retention dataset exists and is reviewable at the next reporting cycle. **Recommendation 3 — Confirm and, if validated, deepen the expansion motion behind Driver B** *(tied to Driver B — Expansion Offset Imbalance, Section 5)* If expansion revenue is confirmed to concentrate in the 50–200 staff tiers, a structured upsell/seat-expansion push into those tiers could widen the offset further while the 15–50 tier's root-cause issues are addressed separately. *Falsifiable within one month:* whether expansion MRR can be broken out by segment, and whether a defined upsell campaign is launched into the 50–200 tier. **Recommendation 4 — Deploy baseline account-level engagement tracking** *(tied to the Section 6 data gap)* Without usage or support-ticket signals, at-risk accounts cannot be flagged before cancellation, and Driver A cannot be explained beyond "it happens in this tier." Instrumenting login frequency and feature usage per account is a prerequisite for any leading-indicator model. *Falsifiable within one month:* whether a usage-decay report exists and can be cross-referenced against next month's cancellation list. ## 9. Open Questions & Data Requests Ranked by how much answering would change the Section 5 diagnosis: 1. **Per-segment (per-tier) logo and revenue churn rates** `<NEEDS-INPUT: segment-level-churn-rates>` — would confirm or overturn the magnitude of Driver A, and could reveal whether the 50–100 or 100–200 tiers have a hidden problem currently masked by the blended 2.8%/1.9% aggregate figures. 2. **Actual Month 1–Month 12 retention percentages for the Jan–Dec cohorts** `<NEEDS-INPUT: cohort-retention-percentages>` — would show whether the concentration pattern is stable, worsening, or improving across the year, directly affecting whether Driver A is treated as an ongoing structural issue or a recent shift. 3. **Segment attribution of expansion MRR** `<NEEDS-INPUT: expansion-source-by-segment>` — would confirm or reject Driver B as currently framed (that expansion is concentrated in the larger, more stable tiers). 4. **Account-level engagement and support-ticket data** `<NEEDS-INPUT: account-level-engagement-data>` — needed to test *why* the 15–50 tier churns more (under-adoption, price sensitivity, support friction, or otherwise), and to build any leading indicator at all. 5. **Absolute MRR, Expansion $, Contraction $, and Churned MRR figures** `<NEEDS-INPUT: absolute-mrr-figures>` — would allow Net Revenue Retention to be computed directly from its full formula rather than derived by inverting the supplied net revenue churn rate.