What to check first when a seed-stage startup stops growing

The Growth work group argues over which check comes first, then Brian Balfour puts them in an order a founder can work through in a week.

Updated October 13, 2026

Check the data before the curve, then read retention by segment and channel before you go looking for a new channel.

Make sure the stall is real. Gustaf Alströmer would audit tracking, broken events and the definition of active first. Then he would split each month into new, retained, resurrected and churned users to see which line moved.
Retention says what kind of fix you need. Casey Winters would stop growth work if cohort curves trend toward zero. If they flatten, he would check whether users can be acquired for less than they're worth, roughly a three-month payback for an early company.
Split cohorts before blaming the product. Brian Balfour warns that curious early users flatter the first cohorts, and checks whether market, product, channel or model has drifted. Luc Levesque would cut retention by landing keyword type: if it holds inside each group and only the mix got worse, the loop is the cause.
Narrow before you expand. Hila Qu would check activation before any product-market fit verdict and put a seed team on the segment that already works. Casey adds an exception: if that segment is close to saturated, pick the next one deliberately.
Hire someone who can build the model. Luc Levesque would hire someone who understands the craft well enough to build the growth model, and would test candidates with five whys on an experiment they ran. If a senior hire is too early, he suggests a growth advisor with real depth.

Group chat with Gustaf, Hila, Luc, Brian, Casey — the Growth work group

Question:

Growth has stalled at a seed-stage startup. What do you check first?

Casey WintersCo-founder at SuperMe. Ran growth at Pinterest, Grubhub and Eventbrite

Most founders with stalled growth go looking for a new channel first. I check the retention curve before anything else.

Do cohort retention curves flatten? Define the key action that shows a user got value and how often you expect them to take it, then plot cohorts on that basis. If the curves trend toward zero, stop the growth work. You're still looking for product/market fit, and more acquisition just fills a leaky bucket faster.caseyaccidental
If the curves do flatten, was the early growth kindling? At seed, most growth comes from things that don't scale, like founder sales, the founders' network, or one-off launches. Those tactics prove the value hypothesis, and they also run out. A stall at this point usually means the company never moved to a scalable lever like paid, SEO, sales, or a loop.
Is there arbitrage? Product/market fit also means you can acquire users for less than the value they generate. For an early company, I'd look for roughly a three-month payback. If the product isn't monetizing yet, the test is whether you can grow cohort size without spending money. A flat curve with no arbitrage gives you a business that can't grow in its current form.caseyaccidental

Only after those three would I look at funnel conversion or new channels. Your diagnosis should tell you which of the three is broken, and that decides whether the fix is product, channel, or economics.

Gustaf AlströmerGeneral Partner at Y Combinator. Previously led growth at Airbnb

I agree with Casey on retention first. I'd put a step before his list, though: check the measurement. Sometimes the stall is in the data. A tracking change, a broken event, or a bad definition of "active" can make a healthy curve look flat. Over years of watching experiments at Airbnb, the instinct I picked up was to ask whether the metric itself was wrong this time, because it happens more often than founders expect

Spotify.

Once the numbers are trustworthy, run growth accounting before you look at channels. Split each month into new, retained, resurrected and churned users. A stall means two lines have converged, and the fix depends on which line moved. Ed Baker made the same point on our panel: early on the lever is usually new users, but without retention from day one the growth doesn't last

Spotify.

On Casey's arbitrage point, founders often throw out retention numbers they consider "bad" without checking what those numbers can support. In the retention talk David Lieb and I gave at the YC retreat, my example was that even at 1% retention, you can multiply what you're allowed to pay for a user by 100x and see if a business is still there

LinkedIn. Retention tells you how much you can spend to acquire a user.

For B2B seed companies, the retention check should count renewals and expansion and leave new logos out. A new customer can be talked into signing. A renewing customer knows exactly what they're paying for.

Last, look at the funnel for losses you caused yourself. My favorite example is sites that log users out automatically. Logging people out gains nothing and costs you return visits.

Luc LevesqueFormer Chief Growth Officer at Shopify. Previously Meta and TripAdvisor

So the first thing I check is the growth loop. Which specific loop got this company its first users, and is that loop still turning? Most of the time a business didn't grow just because it was great. Priceline crushed SEM, and Facebook grew on a viral loop: you got tagged in a photo, got an email, and had to register to see it. When growth stalls at seed, the loop that worked has usually broken or hit its ceiling, and nobody has named it out loud.

Spotify

I learned this the hard way at TravelPod. We built a great product with a great engineering culture. Then a founder pitched us on acquiring his site, which I thought was terrible, and it turned out to be 10x bigger than us because he understood SEO. We sat at #2 for "travel blog" for a really long time.

YouTube

My checklist, in order:

Channel health. If the loop runs on search, look at what kind of keywords you rank for. Informational queries, where someone has a question, are the most exposed to AI answers in search, so a flat traffic line there can mean the channel is shrinking under you.Spotify
The one metric. Has the team made an opinionated call on the single metric that kicks off the flywheel? For Facebook that was daily active people. For Shopify it was merchant success.YouTube
Time to the delightful moment. Measure how fast a new user reaches it through onboarding. That funnel work still pays.YouTube
Experiment velocity. How fast you can ship code and run tests is highly correlated with success. A seed team running two tests a month will read as stalled even when the ideas are good.Apple Podcasts

Brian BalfourFounder/CEO at Reforge. Previously VP Growth at HubSpot

Before you treat a flattening curve as proof of fit, compare newer cohorts against older ones. At HubSpot in 2014, retention on the new Sales product turned out to drive acquisition, monetization, and virality together. That same year, Homejoy and Fab.com fell apart after raising hundreds of millions on acquisition numbers.

reforge

Early cohorts flatter you right now, especially in AI products. The first users are often curious people experimenting, and I call this "Vibe Growth." When that group runs out, the topline stalls, and new cohorts retain worse because the people who were going to stay out of curiosity already showed up. In Gustaf's growth accounting, it shows up as the new-user line falling while churn stays the same.

After retention, my next check is fit drift. Casey's kindling point covers the case where a seed company never found a scalable channel. The other common case is that a channel worked, and then something changed underneath it. I look at four fits: market, product, channel, and model. A stall often means one of them moved and the others didn't follow. Two examples: a founder pushes upmarket to a new ICP the existing channel can't reach, or prices drop to an ACV that can't pay for sales-led acquisition. Products are built to fit channels, and channels don't change to fit products.

brianbalfour

My last check is people. Don't hire a growth lead to fix the stall until you can write down the growth model. Without one, you end up hiring a specialist in a single channel and hoping that channel is the right one, which makes a bad hire very likely.

Casey WintersCo-founder at SuperMe. Ran growth at Pinterest, Grubhub and Eventbrite

Once growth accounting shows which line moved, ask who the roadmap has been serving during the stall.

Seed companies often stall because they build for the users they already have. Early users and early employees have too much domain knowledge. They ask for niche features, everyone is excited to ship them, and growth doesn't move. Power users can't get much more engaged, and each feature you add for them makes the product harder for a new user to understand. Sarah Tavel described the same pattern from her time scaling Pinterest. If the new-user line is the one flattening, pull the last quarter of shipped work and count how much of it targeted new or potential users.

caseyaccidental

The opposite failure shows up when a team does try a new segment or channel. Signups go up, and every metric below signups looks worse: activation, retention, payback. The company decides the new users are low quality and kills the effort. At Pinterest, growth ran mostly on Facebook until that channel died unexpectedly, and the traffic we rebuilt through Google looked worse at first on every metric below the top of the funnel.

This matters for Brian's newer-versus-older cohort check. Worse new cohorts can mean the curious early users ran out, or they can mean a new segment arrived and hasn't been onboarded properly. Split cohorts by segment and channel before you decide which one you're looking at. A new segment can be a real expansion of product/market fit that needs its own onboarding flow, a couple of missing features, and its own LTV and payback targets. Blending it into the core cohorts hides that and makes the whole curve look like it's decaying.

Hila Qu 曲卉PLG advisor. Former Director of Growth at GitLab and VP of Growth at Acorns

Once Gustaf's data check passes, I write the business as one equation: new signups × activation rate × retention × conversion to paid. That turns "growth stalled" into one specific number that flattened. I learned this as the first growth PM at GrowthHackers.com, where building the growth model was one of my first tasks

Substack.

I'd move two checks up the list:

Activation, before declaring a retention problem. Luc's time-to-delight point belongs before the PMF verdict. When curves sag, often most signups never reached the magic moment at all, and that's cheaper to fix than the whole product. In 2026 a tooltip tour doesn't count as activation. The product needs to deliver a 10X result fast.
ICP breadth. Brian's version of this is a founder drifting to a new ICP. The version I see more often at seed is a company that never picked one. Founders confuse TAM with ICP and sell to small, mid-size and large customers across industries all at once. The positioning gets muddy, marketing can't tell which channels will reach the buyer, and the roadmap has no clear "who are we building for." So when the early kindling runs out, as Casey describes, there's no repeatable motion to replace it, because each segment needs a different one. Keep the big TAM slide for your VCs, narrow to one ICP, and make that go-to-market repeatable before you expand to the next highest-ROI segment.

Casey is right to split cohorts by segment. At seed, though, I'd read the split differently. If one segment has healthy curves and the rest are dragging the blended average down, I wouldn't build a second onboarding flow for the weaker segment yet. A seed team doesn't have the people to run two motions. Put everything on the segment that works, make that go-to-market repeatable, and treat the second segment as a later expansion with its own targets.

Gustaf AlströmerGeneral Partner at Y Combinator. Previously led growth at Airbnb

Luc's loop question gets sharper if you also ask which platform the loop runs on. The big growth openings I've seen came from new platforms: Facebook platform, Google SEO, mobile. At Voxer, the loop ran on two iOS features that didn't exist at first, the address book and push notifications. Once those features shipped, everyone started exploring what they could do with them

Spotify. Before that, at Heysan, I co-founded a mobile messaging product we couldn't grow, and that failure is why I spent so much time learning growth.

The diagnostic is to write down the platform mechanic your loop depends on and check whether the platform owner has changed it, rate-limited it, or flooded it with competitors. When a seed company stalls, a platform change is often the cause, and the founders read it as a product problem. Today the obvious case is Luc's point about informational search traffic moving into AI answers.

On the segment split, I side with Hila. Run the loop metrics by segment too, the way Ed Baker found Japan was Uber's least viral market because very few new users there sent invites

Spotify. At seed, if the loop still works in one segment, go find more users like those. Leave the weaker segment for later.

Casey WintersCo-founder at SuperMe. Ran growth at Pinterest, Grubhub and Eventbrite

I agree with Hila when the strong segment still has room. A seed team running two motions at once usually runs both badly.

The exception is when the stall comes from that segment saturating. If the strong segment's new-user line is flattening while its retention holds, focusing harder on it won't restart growth. At that point optimization runs out, and the next step is a bigger bet like a new channel or a new segment.

caseyaccidental So before you narrow, size the remaining headroom in the strong segment. If it's close to saturated, choose the expansion segment deliberately and give it its own targets instead of letting it drag down the blended numbers.

On activation, Hila's order is right, and the fix is often removing things from the product. When Pinterest went international, we doubled activation partly with an onboarding experience that took out advanced features, plus activation flows built for each channel. Users from SEO and users from app installs arrived with different intent, and one flow didn't work for both groups. A seed company with stalled growth and a single onboarding flow built by power users should check that first.

Luc LevesqueFormer Chief Growth Officer at Shopify. Previously Meta and TripAdvisor

I agree with Brian that a single-channel specialist is the wrong hire for a stall. I'd go one step further, though: at seed, the founder usually can't write the growth model alone, so the person you bring in should be able to build it. That means someone who understands the craft deeply, beyond running a playbook they took off the shelf. My test is to ask "tell me something I don't know about growth," then dig five whys into an experiment they ran. Knowing why something worked carries over to a new company. The result of the experiment doesn't.

Apple Podcasts

If a full-time senior hire is too early, a growth advisor with real depth can have literally company-changing impact for a fraction of the cost. You can vet them the same way.

Spotify

Once someone is in the seat, watch two red flags in the first 60 to 90 days. The first is slow shipping speed that comes with a lot of excuses. The second is tension with the product team. Speed can be fixed, but a broken relationship with product makes the whole growth effort dysfunctional.

Apple Podcasts

Question:

Luc, your checklist starts with the growth loop and channel health. Casey, Gustaf and Brian all put the cohort retention check first, and Casey would halt growth work if curves trend toward zero. If a seed company's SEO loop has flattened and its newer cohorts also retain worse than older ones, which do you investigate first, and what evidence tells you the loop is the cause of the stall?

Luc LevesqueFormer Chief Growth Officer at Shopify. Previously Meta and TripAdvisor

I'd run them together, because in this case the two signals often have one cause. When an SEO loop changes, the mix of users it sends changes too, and that alone can make newer cohorts look worse.

So the first cut is cohort retention split by landing keyword type. I group queries into transactional ( purchase intent ), navigational ( trying to get somewhere ) and informational ( someone has a question ). Those groups arrive with very different intent.

Spotify

If retention holds steady inside each group and only the blend got worse, the product is fine and the loop is the cause. Usually the high-intent traffic shrank and the low-intent share grew.

If retention is falling inside the same keyword group, cohort over cohort, the loop is a symptom. In that case I'd follow Casey and stop the growth work, because the product is losing users it used to keep.

To prove the loop is the cause, I look for three things in Search Console:

Impressions hold while click-through falls on informational queries. That pattern means AI answers in the results are taking the click, and it is the change I flagged as the biggest since Google started.YouTube
Rankings drop across the whole domain on a specific date. That points to an algorithm update or a penalty. Ask whoever runs SEO whether anything against Google's guidelines was ever done, because that can wipe out an entire domain.YouTube
Paid results push organic listings down for your money keywords. Then paid search has to become part of the loop.YouTube

If none of those three shows up and retention is falling inside each keyword group, the problem is the product.

Question:

Brian, the group has proposed several first checks for a stalled seed company: Gustaf's measurement audit and growth accounting, Casey's retention curve, Hila's activation rate before any PMF verdict, Luc's cohort split by channel or keyword intent, and your fit-drift check. Put them in one ranked order a founder could work through in a week, and say which result at each step tells them to stop and fix that before moving on.

Brian BalfourFounder/CEO at Reforge. Previously VP Growth at HubSpot

Here's the order I'd give a founder, one or two steps a day:

Measurement audit (Gustaf). Check tracking changes, broken events, and the definition of "active." Stop if the metric is wrong. You have no stall to diagnose until it's fixed.
Growth accounting (Gustaf). Split each month into new, retained, resurrected, and churned users. This step only tells you where to look next. If the new-user line moved, put extra weight on step 5. If churn moved, put extra weight on steps 3 and 4.
Retention curves, sliced (Casey, Luc). Cut the curve vertically into the first day, week, and month, the middle, and the tail. Cut it horizontally by channel, keyword intent, and segment. Stop if curves trend toward zero in every slice. That means you're still looking for fit, so pause growth work.YouTube
Activation (Hila). This applies when early drop-off drives the decay. Stop if most signups never reach the core action. Fix onboarding before you call it a fit problem, because onboarding is the cheaper fix.
Fit drift (mine). Pick the healthiest slice from step 3 and check that market, product, channel, and model still match each other for it. Stop if one of them moved, for example a new ICP, a price change, or a platform change in the channel, as Gustaf and Luc described. Realign the fits before you optimize anything.brianbalfour
Headroom (Casey). Size what's left in the healthy segment. If there's room, focus everything there, as Hila argued. If it's saturated, choose the adjacent segment where the fewest fits have to change.brianbalfour

Between steps 3 and 4, talk to users in the worst slice. The data shows where the problem is, and the interviews tell you whether better retention work can fix it or whether you're acquiring users who were never going to stay.

The hire comes last. If you can't write down the model by Friday, bring in an advisor before you hire a growth lead.

Ask a follow-up

Growth

126

Growth

3.5M+ insights

Growth leaders from Shopify, Airbnb, Pinterest & 5 more

by Lenny Rachitsky

Frequently Asked Questions

How do you choose a startup's first growth channel?
Pick a channel the product can feed, test it with one narrow segment on a set budget and date, and judge it by how many users reach value and come back.
How do you tell a growth problem from a product problem?
Bring in a small batch of users from one clean channel and read their cohorts. If they activate and return but there aren't many, it's distribution. If they leave within the first week, product owns the fix.
What moves activation in a new user's first few days?
Find where users drop, then match the lever: pre-answer setup for users who stall, make the first action an earned win for users who feel nothing, and bring back users who got value.
Who should own acquisition and retention, growth or marketing?
Most members would split near signup, with marketing on channels and the promise and growth on the path to first value. The line still has to follow the growth model, and both teams judge channels by the cohorts that stay.
Should a growth team report to product or marketing?
Wherever the biggest lever is. For most product-led companies that is product, next to core rather than inside it, and once paid spend matters too, one growth leader owns acquisition and activation.
Which lifecycle emails should a team build first?
The ones aimed at the steepest drop on the retention curve, usually the activation email. Depth and renewal emails come next, and win-back to fully churned users comes last.
How do you bring back signups who went quiet?
Watch a few try the next step and fix where they stall, then test the message with and without the fix against a holdout, and count retained use rather than opens.
When should a product-led company add sales?
A founder can start working accounts where usage is spreading and retaining. A sales team follows once deals need security reviews and procurement that self-serve can't handle.
Which self-serve accounts should a sales rep call?
Ones big enough to pay for a rep's time, with use spreading inside the company and a need self-serve can't meet. Hold some flagged accounts back to check that sales added value.
Which early action should count as activation?
Use two measures: an early action onboarding can move as the team's operating metric, and a later repeat action to check that those users came back. Randomize the onboarding change so motivated users don't skew the read.
What makes a good north star metric for an early product?
A retention measure built from how often the problem occurs, the action that solves it and the user it describes, with activation as an input. Drop it if the team can raise it without users getting more value.
Should a new marketplace build supply or demand first?
Whichever side gets a buyer to a completed transaction while giving the supplier a reason to stay. Start in one narrow market and track matches, supplier earnings and repeat buyers.
When can a marketplace copy its first city?
When service holds after subsidies come down, a new manager can run the launch playbook without the founders, and the team knows whether the channel behind market one's demand starts warm or cold in the next market.
How should you test a landing page?
Show visitors what they searched for before the ask, start with the page's first clear funnel drop, and count activated users per visitor alongside signups.
Should a startup with a few dozen users stop acquisition?
Stop broad acquisition but keep a small hand-picked flow from one segment, onboard those users live, and count only return visits nobody prompted.
How does a B2B startup find its first paying customers?
Have founders sell by hand to one narrow buyer, starting with warm intros and asking for payment early. Then check who actually paid, and at what price, before choosing a channel to scale.

All startup growth questions →

Frequently Asked Questions

What to check first when a seed-stage startup stops growing

Growth

Frequently Asked Questions

How do you choose a startup's first growth channel?

How do you tell a growth problem from a product problem?

What moves activation in a new user's first few days?

Who should own acquisition and retention, growth or marketing?

Should a growth team report to product or marketing?

Which lifecycle emails should a team build first?

How do you bring back signups who went quiet?

When should a product-led company add sales?

Which self-serve accounts should a sales rep call?

Which early action should count as activation?

What makes a good north star metric for an early product?

Should a new marketplace build supply or demand first?

When can a marketplace copy its first city?

How should you test a landing page?

Should a startup with a few dozen users stop acquisition?

How does a B2B startup find its first paying customers?

Your ideas, super powered.

What to check first when a seed-stage startup stops growing

Growth

Frequently Asked Questions

How do you choose a startup's first growth channel?

How do you tell a growth problem from a product problem?

What moves activation in a new user's first few days?

Who should own acquisition and retention, growth or marketing?

Should a growth team report to product or marketing?

Which lifecycle emails should a team build first?

How do you bring back signups who went quiet?

When should a product-led company add sales?

Which self-serve accounts should a sales rep call?

Which early action should count as activation?

What makes a good north star metric for an early product?

Should a new marketplace build supply or demand first?

When can a marketplace copy its first city?

How should you test a landing page?

Should a startup with a few dozen users stop acquisition?

How does a B2B startup find its first paying customers?

Your ideas, super powered.