Analysts: 6–12 Month Cohort Forecasts for Net Dollar Retention

The most reliable way to forecast net dollar retention is cohort-based modeling that extrapolates observed retention curves and reports base, upside, and downside ranges rather than a single number. This beats flat-multiplier forecasting because it captures how retention decays differently across account ages and segments. Your first move: export the cohort-dollar triangle and validate every billing event behind it before you fit any curve.

---

TL;DR:

• Forecasting NDR with cohort-based modeling provides more accurate ranges by capturing differing retention decay patterns across segments and account ages.
• Building the cohort-dollar triangle from raw billing data and reconciling reactivations ensures reliable retention shape analysis before applying extrapolation methods.
• Scenario planning with base, upside, and downside cases enhances forecast robustness by accounting for retention levers and acquisition pauses.
• Higher ACV businesses tend to have higher NRR, making percentile benchmarks more meaningful than global medians for specific segments and stages.
• Automated tools that generate explainable health scores and update cohort data reduce manual effort and improve forecast accuracy for ongoing planning.

---

Table of Contents

• What net dollar retention is and how to calculate it
• Why forecasting NDR matters for growth planning and valuation
• Required data and cohort construction: the cohort triangle and segmentation
• Forecasting methods: cohort extrapolation, sBG/power-law fits, and practical two-parameter models
• Scenario planning, stress tests and sensitivity: base, upside, downside
• Benchmarks and how to use them correctly, by ACV and stage
• Common pitfalls and audit checks to ensure forecasts are honest
• Quick practical checklist: produce a 6 to 12 month NDR forecast this week
• Customerscore data and how it informs practical forecasts
• How to treat NDR forecasts in planning and cadence
• How Customerscore operationalizes NDR forecasting
• Sources
• FAQ

What net dollar retention is and how to calculate it

Net dollar retention (NDR), also called net revenue retention (NRR), measures how much recurring revenue an existing customer base generates over a period, including expansion and contraction, but excluding new logos. The canonical formula is (Starting ARR + Expansion − Contraction − Churn) / Starting ARR. Say a cohort starts the year at $1,000,000 in ARR, gains $150,000 in upsells, loses $40,000 to downgrades, and churns $60,000. NDR equals ($1,000,000 + $150,000 − $40,000 − $60,000) / $1,000,000, or 105%.

A few conventions matter more than they seem:

• Gross revenue retention (GRR) excludes expansion and caps at 100%, showing pure retention risk.
• NDR can exceed 100% when expansion outweighs churn and contraction.
• Always anchor the calculation to a fixed starting cohort and exclude any revenue from new customers acquired during the period.

Whether you use MRR or ARR, keep the unit consistent across the whole NRR analysis.

Why forecasting NDR matters for growth planning and valuation

NDR compounds the revenue you already have, which means a forecast built on it tells you how much growth you can expect before a single new deal closes. A company holding 110% NDR doubles its existing base roughly every seven years without new sales; a company at 100% needs new logos just to stand still. That gap changes how much you spend on acquisition versus retention.

A swing from 100% to 110% NDR roughly doubles the compounding effect on existing ARR over a multi-year horizon, which is why SaaS Capital's benchmark research treats NRR as a core input to growth-rate expectations.

Boards and acquirers read NDR forecasts as a proxy for durability. A forecast with tight, defensible bands signals a business that understands its own retention economics and can prioritize investment accordingly.

Required data and cohort construction: the cohort triangle and segmentation

A defensible forecast starts with a cohort-dollar triangle: rows are acquisition cohorts by month, columns are calendar months since acquisition, and each cell holds both a retention percentage and a dollar figure. You need both, because percentages alone hide the size of the accounts driving the trend.

1. Export raw billing events (subscriptions, upgrades, downgrades, cancellations, reactivations) rather than pre-aggregated MRR snapshots.
2. Rebuild MRR per cohort per month from those events so expansion and contraction are traceable to specific invoices, following the diligence-grade approach MyDealList recommends.
3. Segment cohorts by ACV tier, billing term (monthly versus annual), acquisition channel, and onboarding completion, since these materially change survival shapes.
4. Reconcile every reactivation and true-up back to its source invoice before it counts as expansion.

Pro Tip: Build the triangle in dollars first, then derive percentages, never the reverse. Rounding errors in percentage-only triangles compound fast across 12 months.

Forecasting methods: cohort extrapolation, sBG/power-law fits, and practical two-parameter models

Once the triangle is clean, you have two broad ways to project it forward. Non-parametric extrapolation interpolates across the existing triangle diagonals, borrowing the shape of older cohorts to estimate where younger ones will land. It works well when you have at least a year of history and stable segments, and SimpleSubscription's guide to cohort forecasting treats this as the honest alternative to flat-multiplier projections.

Parametric fits are more useful when history is short. Two curves dominate practice:

• The shifted-Beta-Geometric (sBG) model separates a heterogeneous "early churn" population from a stable long-term retainer group, producing a floor retention rate.
• Power-law decay fits a smoothly declining hazard rate, useful when churn keeps dropping slowly rather than flattening.

A simpler, more practical version works when you only have M1 and M6 data: derive a post-M1 monthly hazard rate from the drop between those two points, then apply that hazard forward to build a conservative floor. This two-parameter approach is exactly what SimpleSubscription describes for teams without enough history for a full curve fit, and it is often the fastest way to get a usable forecast this quarter.

Non-parametric methods tend to be more accurate short-term when data is deep; parametric fits extend further but carry more assumption risk. Either way, treat 12 months as the practical accuracy horizon and widen your scenario bands substantially beyond it.

Scenario planning, stress tests and sensitivity: base, upside, downside

A single NDR number is a guess dressed up as a fact. Build three scenarios instead: base case uses your fitted hazard rate and current acquisition pace, upside assumes retention levers (onboarding fixes, CS playbooks) reduce hazard by a defined amount, and downside assumes flat or slightly worsening hazard with no new acquisition for six months.

1. Freeze new-logo acquisition in the downside case and recompute MRR at month 6 using only existing cohort decay.
2. Apply a ±10% shift to the monthly hazard rate in base case and recompute 12-month MRR and NDR for both directions.
3. Present all three cases side by side with the specific assumption driving each.

Leadership needs the range and the assumption behind each edge, not a false point estimate. This is the single most useful check SimpleSubscription recommends before anyone signs off on a plan built around retention.

Benchmarks and how to use them correctly, by ACV and stage

Global medians flatter or scare people depending on where they sit relative to the number. Private B2B SaaS companies report a median NRR of roughly 101% to 103%, with median GRR around 91%, according to SaaS Capital's survey, and the same research finds higher ACV correlated with higher retention.

That correlation is the reason a single global median misleads:

• A low-ACV, self-serve product should expect NDR closer to the lower end of that range, since smaller accounts churn more easily and expand less.
• An enterprise-ACV business with dedicated CS coverage should expect to sit above the median, and a forecast landing below it deserves scrutiny.
• Use the ACV-matched percentile band, not the global figure, when anchoring your downside case.

Pulling the wrong peer group into a board deck is how "good" numbers get misread as warning signs, or the reverse.

Common pitfalls and audit checks to ensure forecasts are honest

Retention metrics are easy to inflate without anyone intending to mislead. The most common problem is counting reactivated ARR or one-time usage true-ups as recurring expansion, which flatters NDR for a quarter and then collapses the following one.

1. Reconcile every expansion dollar in the forecast back to an actual invoice line.
2. Flag any period where a handful of accounts drive a disproportionate share of total expansion.
3. Compare the GRR-to-NRR spread across cohorts; a sudden widening usually means expansion is masking churn, not offsetting it.

These checks come straight from the manipulation risks SaaS Capital documents in its benchmarking work.

Pro Tip: Run the GRR-NRR spread check every month, not just at quarter close. A spread that widens quietly for two months is far easier to fix than one discovered at board prep time.

Quick practical checklist: produce a 6 to 12 month NDR forecast this week

You do not need a data science team to get a usable forecast on the table.

1. Export the cohort-dollar triangle from raw billing events and reconcile reactivations and true-ups.
2. Fit a two-parameter hazard model from M1 and M6, or interpolate the triangle directly if you have a full year of history.
3. Build base, upside, and downside scenarios, including the flat-acquisition downside MRR check at six months.
4. Package the result as one slide: the range, the assumptions behind each edge, and two recommended retention actions tied to the gap between base and upside.

That last step is what turns a spreadsheet exercise into something a CFO or board actually uses.

Customerscore data and how it informs practical forecasts

Real cohort shapes rarely follow a clean textbook curve, which is why grounding assumptions in actual usage data matters. Customerscore's retention study across 44,000 users surfaces practical cohort shapes and shows where health scoring correlates with hazard reductions.

• Product usage signals often move ahead of billing signals, giving earlier warning than a triangle built on invoices alone.
• Explainable health scores can be fed directly into a hazard model as a leading covariate rather than a lagging one.

— Patrik

How to treat NDR forecasts in planning and cadence

Treat the cohort-based forecast with scenario bands as a living input, not a quarterly ritual. Review it monthly internally, present it quarterly to the board, and use the gap between your base and upside case to decide which single retention experiment to run next.

— Patrik

How Customerscore operationalizes NDR forecasting

Building and maintaining a cohort triangle by hand is the part most teams abandon after the first quarter. Some platforms automate the pieces that make that triangle honest: pulling billing, product usage, CRM, and support data into one place, generating explainable health scores, and predicting churn risk before it shows up as a canceled invoice.

• Automated cohort-dollar exports remove the manual reconciliation step described earlier in this guide.
• Explainable health scores feed directly into your hazard estimates as a leading signal.
• Integrations with popular tools can keep the triangle current without a manual export cycle.

If you want to see how automated cohort projections work and what they would mean for your cohorts, book a demo or check pricing, a flat platform fee tiered by your client ARR rather than per seat.

Sources

The figures and frameworks in this guide draw on a small set of primary sources worth reading directly.

• Private SaaS company growth-rate benchmarks (SaaS Capital)
• Subscription revenue forecasts: cohort models (SimpleSubscription)
• SaaS churn cohort analysis & forecasting (MyDealList)

FAQ

What is a good net dollar retention rate?

A rate above 100% means expansion is outpacing churn and contraction within your existing customer base. Private B2B SaaS companies report a median NRR of roughly 101% to 103%, though the right target depends heavily on your ACV tier.

What does net dollar retention mean?

Net dollar retention measures how much recurring revenue your existing customers generate over a period, counting expansion and subtracting contraction and churn, while excluding any revenue from new customers. It isolates how well you grow and keep the base you already have, as explained in Customerscore's breakdown of NRR.

Can net retention rate be over 100 percent?

Yes, and it should be for a healthy expansion-driven business. NDR exceeds 100% when upsell and cross-sell revenue from existing accounts outweighs the revenue lost to downgrades and cancellations.

What is the rule of 40 in SaaS?

The rule of 40 states that a healthy SaaS company's growth rate plus its profit margin should add up to roughly 40% or more. It is a separate efficiency benchmark from NDR, though strong retention makes the growth side of that equation easier to sustain.

Recommended

• 8 SaaS Retention Metrics Beyond Churn
• SaaS Retention: 44,000-User Study
• Net Revenue Retention (NRR) in SaaS
• Gross Revenue Retention (GRR) Guide