Cohort retention seed: the 3 chart shapes VCs ask for
The three cohort retention shapes seed investors pattern-match in seconds, and how much data you actually need to show.
Cohort retention seed: the 3 chart shapes VCs ask for
Cohort retention at seed comes down to three shapes investors pattern-match in seconds: the smile (older cohorts curve back up, the strongest PMF signal), the flat plateau (curve flattens above zero, sticky enough to fund), and the bleed (curve heads to zero, you are not raising on this chart). Show three months minimum, six if you have it.
Most seed decks lead with logo count and MRR. The chart that actually moves a partner meeting is the cohort retention curve, because it is the one slide where lying is hard and the shape tells the entire product story in under five seconds.
This guide is the tactical breakdown: the three cohort shapes investors look for, how much data is enough at seed, what cadence to plot, and the slicing investors will ask for in the second meeting.
The 3 cohort retention shapes seed investors pattern-match
Every seed partner reading your retention chart is doing one thing: shape recognition. The numbers matter less than which of these three buckets your curve falls into.
- The smile. Each cohort drops, flattens, then curves back up over later months. This is canonical product-market fit and the rarest shape at seed. Sequoia calls it out as the signal of strong PMF and expansion potential, where older cohorts increase in retention or engagement over time (Sequoia, Retention).
- The flat plateau. Cohorts decay for the first one to three periods, then flatten at a non-zero percentage and stay there. This is the most common "fundable" shape at seed. The plateau height is what gets debated, not whether the shape is acceptable.
- The bleed. Cohorts decline toward zero with no flattening visible. Sequoia's framework lists "declining to zero" as one of the three primary archetypes, and it is the one that ends conversations (Sequoia, Measuring Product Health). If your curve looks like this, do not lead with it. Fix the product, then come back.
- The fake plateau. Looks flat but only because you have two months of data. Investors discount this immediately. Three months minimum, six months ideal.
- The mixed bag. Newer cohorts retain better than older ones, suggesting product improvement. Show this with newer cohorts on top, annotated with what changed.
How much cohort retention data is enough at seed
Three months is the floor, six months is the answer that closes the question. The minimum exists because partners need to see the curve start bending toward a plateau, and one or two data points cannot prove a bend.
Y Combinator's guidance on cohort retention recommends matching measurement windows to product cadence, with D1, D7, D28, D84, and D364 as the standard checkpoints for predicting long-term health (Y Combinator, How To Improve Cohort Retention). For most seed B2B SaaS, that translates to monthly cohorts shown over three to six months. For consumer apps with daily usage patterns, weekly cohorts over eight to twelve weeks reads stronger.
The trap to avoid: showing cohort data with the wrong cadence to inflate the number. Y Combinator is explicit that picking a cadence misaligned with natural usage frequency artificially inflates retention results (Y Combinator, How To Improve Cohort Retention). A weekly-usage product showing daily retention is a tell, and partners who have seen a thousand decks will catch it.
What cadence to plot for your monthly retention cohort
Pick the cadence that matches how your product is actually used, then defend it in one sentence on the slide.
| Product type | Cadence | Window to show |
|---|---|---|
| B2B SaaS, weekly+ usage | Monthly cohorts | 3 to 6 months |
| Consumer app, daily usage | Weekly cohorts | 8 to 12 weeks |
| Marketplace, transactional | Monthly cohorts on first transaction | 3 to 6 months |
| Vertical SaaS, low-frequency workflow | Quarterly cohorts | 2 to 4 quarters |
| Dev tools, sporadic but sticky | Monthly active cohorts | 6 months |
Whatever you pick, define your cohort action precisely. For revenue businesses, Y Combinator recommends defining the action as both paying and active, not just paying, because billed-but-dormant accounts lie about engagement (Y Combinator, How To Improve Cohort Retention).
What the cohort chart investor actually wants on the slide
The chart itself is half the work. The other half is the framing around it.
- Triangle table plus line chart, side by side. The triangle (the raw cohort percentages) shows you have the data. The line chart shows the shape. Don't make the partner squint at numbers.
- New cohorts visually distinct from old. Color-code recent cohorts brighter or thicker. Y Combinator notes investors prefer seeing the trajectory of new versus older cohorts rather than a point-in-time stat (Y Combinator, How To Improve Cohort Retention).
- One annotated takeaway. A single sentence under the chart: "Cohort 6 plateaus at 42% by month 3, up from 28% in cohort 1 after the onboarding redesign in March." This is the line that gets quoted in the partner meeting memo.
- Slice by acquisition channel or geo if you have signal. OpenVC's metrics guidance flags that slicing cohorts by acquisition channel or geography is essential for explaining what you learned from your best and worst groups (OpenVC, Metrics Primer). If paid-acquired cohorts retain at half the rate of organic, show it. Hiding it costs you trust when they find it in diligence.
- Don't lead with NRR if it is below 100%. Net revenue retention is a Series-A headline metric. At seed, lead with logo retention or gross retention if NRR doesn't tell a clean story.
If your cohort chart shows a flat plateau above 30% by month three, you have enough to raise. If it shows a smile, you have enough to choose your investor.
How to present cohort retention to seed investors
Three rules that separate decks that get a second meeting from decks that get a polite pass.
Show the chart, not the conclusion. Founders who write "we have great retention" without the chart get asked for the chart. Founders who lead with the chart and let the shape speak skip a step.
Annotate inflection points. When a cohort visibly improved after a product change, label it on the chart. "v2 onboarding shipped" with an arrow to the improving cohort is the kind of detail that reads as operator-grade.
Pre-empt the slicing question. The second meeting will include the question "how does this look by channel / geo / plan tier?" Have the cut ready as a backup slide. Bringing it out unprompted compresses two meetings into one.
If you are pulling cohort data manually from Stripe, Mixpanel, or your warehouse every time you update the deck, tools like Causo keep the cohort cuts current so the slide doesn't go stale between investor meetings.
FAQ
How many months of cohort data do investors want at seed? A minimum of three months, with six being the answer that closes the question. Three months lets a partner see whether your curve is bending toward a plateau. Six months proves the plateau is real, not a delayed bleed. Anything under three months reads as "too early to tell" and pushes the diligence to your next update.
What does a good cohort retention chart look like for seed-stage startups? It bends, then flattens, ideally above zero. Each new cohort sits on or above the older ones, showing your product is getting stickier as you ship. The shape investors love most is the smile, where older cohorts re-engage and curve back up, but a clean flat plateau at a non-trivial percentage also passes.
How do I calculate cohort retention for marketplaces? Group users by the month of their first completed transaction, not signup. Then track the percentage of each cohort that transacts again in months 1, 2, 3, and so on. Run the analysis on both sides of the marketplace separately, supply and demand, because their retention shapes almost always diverge and investors will ask.
What is net revenue retention and what should it be at seed? Net revenue retention (NRR) measures revenue from a cohort over time, including expansion, contraction, and churn. At seed it is rarely the headline number because cohorts are small and noisy. If you do have it and it sits above 100%, lead with it. Below 100%, show logo retention and gross retention instead and explain the expansion plan.
What is a smile in user retention cohorts? The smile is when a cohort's retention curve drops, flattens, then ticks back up over time as users return or expand usage. Sequoia treats it as one of the strongest signals of product-market fit because it implies the product compounds value rather than decays. Smiles are rare at seed and earn outsized credit when present.
Run this playbook inside Causo.
Match to the best-fit partner at 1,000+ funds, draft a hyper-specific email, and send from your email — in one place.