How do you diagnose growth at the seed stage?

The first thing I do before anything else is get access to the data. An advisor without dashboards is just guessing. The most effective advisory arrangement is a recurring weekly meeting with leadership where the agenda flexes to the most pressing problem, but that only works if I can see the underlying metrics myself.

From there the diagnostic runs in sequence, and step one is the three-body problem. Consumer networks lack the established playbooks that enterprise or product-led models have, because they require you to solve for three things at once: the right target customer, the right value proposition, and the minimum network size needed to deliver that value. Most founders who come to me have misdiagnosed their growth problem as “we need more users.” Usually the actual problem is one of the other two legs.

I measure that quantitatively. Pull cohort retention curves by segment. If the curve trends toward zero for every segment you look at, you don't have a retention problem, you have a value proposition problem, and no acquisition strategy fixes that. If the curve flattens for a specific segment, you've found your Oklahoma — now the question is how to get to Missouri.

For network products a flat curve is harder to read than it is for a pure SaaS tool. Retention might be low because the value proposition is weak, or because the network isn't dense enough yet to deliver the value, and those require completely different responses. If you fix density for the wrong value proposition you've just scaled a broken product. So I look for bright spots: are there pockets, geographies, cohorts or use cases where retention is meaningfully higher? If yes, that tells you what the value proposition actually is and who it's for. If no bright spot exists anywhere, the value proposition needs rethinking before you touch acquisition.

Then I audit the acquisition model for kindle versus fire. Kindle strategies are bespoke and unsustainable but useful for early traction; fire strategies are scalable and long-term. At seed almost everything is kindle, and that's fine — but only if the kindle is sequencing you toward a fire. The fire options for consumer networks are limited: virality, SEO, paid, content and sales. I want to know which of those the product is structurally set up to support, and whether the team has even thought about it.

Most seed-stage teams think in funnels. Funnels create linear growth and force you to keep pouring into the top just to stay flat. I push them to model the business as a loop instead — a closed system where the output of one cycle becomes the input of the next. For a consumer network the question is whether retention at the bottom is actually feeding acquisition at the top. If it isn't, you're running a leaky bucket and no acquisition spend fixes that structurally.

The last check is magic tricks versus moats. Especially for AI-native consumer networks right now, a lot of what looks like growth is novelty. Viral growth is not a moat — once the trick is explained or cloned, the product is a commodity. I look at whether defensibility is structural, built on network effects or data network effects, or whether it's a feature incumbents can copy. The most durable models have cross-side network effects.

On the 1/3/6 structure: month 1 is entirely diagnostic and I'm not recommending tactics yet. By month 3 I want a clear answer on which segment has the flattest retention curve and why, what fire strategy the product can realistically support, and what the loop looks like modelled properly. By month 6 the focus shifts from broad diagnosis to deep work with whoever runs growth. If they're still in “let's try stuff” mode at month 6, that's a signal the strategy isn't clear enough yet.

The order matters, because you can't prioritise what to fix until you understand what's actually broken, and at seed stage most founders haven't separated those two questions yet.

So the first thing I'd do is map the actual use cases the product serves. Most seed-stage consumer and network teams have built around a persona — “25-34 year old urban professional” — when the real strategic unit is the use case: what specific job is someone hiring this product to do, at what frequency, in what context? Use cases define your retention curve shape, your acquisition loop structure and your monetization ceiling. Personas don't. I want to know how many distinct use cases are live, which has the most traction, and whether the team has consciously chosen one. Trying to serve several at seed is almost always a mistake.

Then retention, before I look at any acquisition number. The company with the best retention in a category ends up winning, because retention interacts with everything else — it shortens payback, which lets you reinvest in acquisition sooner, which compounds the loop faster. I pull the curves and ask three things. Is there a stable floor, or does the curve go to zero? Does the measurement window match the natural usage frequency of the use case — a weekly product measured on daily retention looks terrible for the wrong reason? And are retention and engagement being conflated? Retention is binary: is the user still active this period. Engagement is depth within the period. Mixing them sends teams in the wrong direction.

Next I map the growth model qualitatively, before any quantitative model. The question is just: how does this product actually grow? Ask every person on the team to draw it. You'll almost always get a different answer from each one, which tells you the team has no shared mental model of how the machine works.

For a consumer or network product I'm looking for three things in that drawing. What is the primary acquisition loop — viral, content/UGC, or paid? How does retention feed back into acquisition, because a strong network product should have retention that strengthens the loop through more content, more connections, more invites? And where does monetization sit, does it help or hurt the other loops?

With the qualitative model in hand I run a Four Fits diagnostic — Market-Product, Product-Channel, Channel-Model, Model-Market. At seed the common failure modes are the first two: Market-Product is weak because the team hasn't narrowed to a specific enough use case, and Product-Channel is undefined because they're “trying every channel” rather than identifying which channel properties match the product's natural distribution mechanics. Products are built to fit channels; channels don't mold to products.

Then I force a prioritisation. There are really only three constraints: the product hasn't found its use case and retention goes to zero; it has a use case but can't acquire efficiently; or it acquires and retains but can't build a sustainable model. At seed it's almost always the first, which means the advisory time should go into product decisions, not growth tactics. Traction driven entirely by word of mouth is one of the strongest indicators of genuine product-market fit — if that organic signal isn't there, layering paid acquisition on a broken retention curve wastes time and money.

Months 2-3 validate the constraint and run focused experiments against it, and the discipline there is concentration: experiments across multiple channels or use cases at once produce noise, not signal. By month 6 the team should have an answer to “how does this product grow?” that every person articulates the same way. The advisory relationship fails when it becomes a list of tactics.

The seed stage is where most growth mistakes get baked in permanently, so the diagnostic has to be sequential. You can't set a good roadmap until you understand what you actually have.

Month one is understanding that. The first question is retention shape — pull the D1/D7/D30 curves and look at whether they flatten. A curve that keeps dropping to zero means you're pouring water into a leaky bucket, and acquisition spend just accelerates the bleed. A curve that flattens, even at a low number, tells you there's a real use case for some segment.

The second question is who stayed, and why. Data tells you how bad the problem is; it doesn't tell you what to do about it. So I go watch real users, and specifically I want the marginal user: the person who signed up, came back a few times, and then stopped. Sitting with them, watching them use it, listening to what they were hoping to get out of it — that's where the actual diagnosis lives. The data points you to the drop-off point in the funnel. The user tells you why.

The third question is specific to network products: is the connection dynamic broken? For a network startup the core question is whether users are finding enough value from other users. At Facebook the unlock was getting new users to 10 friends in 14 days, because that was the threshold where the feed became genuinely interesting. The specific number mattered less than the galvanizing clarity it gave the team. I'd want to find the equivalent for this product — what does a user need to experience in the first session or first week for the network to feel alive?

By the end of month one I want one question answered cleanly: is this a retention problem, an activation problem, or an acquisition problem? The order matters. Retention first, then activation, then acquisition. Fixing acquisition before retention is the most common and most expensive mistake at seed stage.

Months 2-3 are about picking the right lever and building the right portfolio. I'd help the team set a single north star that's a behaviour, not a vanity number — DAU is a behaviour, downloads is not — with a time-bound target specific enough to be galvanizing even if the exact number is somewhat arbitrary. Zuck's “10 friends in 14 days” worked because it stopped the team debating and started them building.

Then the portfolio. The failure mode I see most is teams running a bunch of small experiments that individually look reasonable but don't add up to anything. A healthy portfolio has a few cannonballs — big, high-conviction bets that could move the north star meaningfully — and a set of smaller bullets that test assumptions faster. At seed I'd put about 80% of engineering capacity on the cannonballs. The constraint of having to choose forces better prioritisation.

Months 4-6 are about instrumenting and building the muscle. The team should have a working experiment process: a clear hypothesis before each test, a defined success metric, and a decision framework for ship-versus-kill. The absence of that is what leads teams to run experiments and then argue about the results for weeks. I'd also push for at least one structured research session per sprint, because the risk at this stage is that the team starts trusting the metrics more than the users. By month 6 the engagement should have produced a clear north star with a baseline, a portfolio with at least one cannonball in flight, and an experiment process the team runs without being reminded. If those three are in place, the team can grow without me.

The first thing I do is resist the urge to give advice. The biggest mistake an advisor can make is walking in with a playbook from a previous company and trying to apply it. Every company is different, and the channels and loops that worked at Shopify or Facebook may be completely wrong here. So the first 30 days are pure diagnostic.

The gate is product-market fit. If the product isn't satisfying a real need, growing it does more damage than good — when somebody tries something and has a bad experience, they're unlikely to try it again. You're burning your best asset, which is first impressions. So I look at retention curves, user interviews and NPS-style signals, and if PMF is shaky that's where the advisory time goes first, because everything else is downstream of it.

Then the north star metric, and I push hard on this. You want the one metric with the highest impact for the company. For Facebook it was daily active people engaged on the product; for Shopify it was getting as many merchants as possible to be successful. For a consumer or network startup it's usually engaged users or a network density signal. If the team can't articulate this clearly, that's a red flag and an early focus area.

Alongside it I map the loop. Most companies that reach scale do it off one really strong loop or channel that they dominate. I'd map their current acquisition, activation and retention loop on paper, then pressure-test whether it's actually working in the data — full funnel visibility, acquisition sources, activation rates, where users drop off, retention by cohort, and any early signal of word of mouth.

By the end of month one I want to be able to say: here is the single biggest constraint on growth right now. Is it that users aren't activating? That the loop has no viral component? That they're acquiring the wrong users entirely? That diagnosis determines everything that follows.

Months 2-3 I focus there almost exclusively. The tried-and-true principles haven't changed: get users to a delightful moment as fast as possible, optimise onboarding and exit flows, and find the one loop worth dominating. For a consumer or network startup specifically I'm asking what the mechanism is by which one user brings in another — if that mechanism doesn't exist or is weak, building it is the highest-leverage thing we can do. I'm also helping the team build the habit of real experimental rigour. A good growth person understands the levers deeply enough to launch experiments they know with 80% confidence will work, because they truly understand the mechanics.

By month four, if the core loop is working, I shift to two things: how do we make the loop faster and tighter, and what is the primary acquisition channel that will feed the top of it at scale? Most companies get to meaningful scale off one channel they've dominated, so I'd rather go deep on the one with the highest ceiling than spread thin across five. I'd also be checking the org questions by then. The CEO and whoever runs growth need a very tight relationship, ideally weekly — if that alignment isn't there the experiments don't get resourced and the insights don't get implemented, which means the advisory time produces activity but no impact.

The whole diagnostic is in service of one thing: finding the needle in the haystack. When you get it right you can see hundreds of percentage points of lift, sometimes over a thousand. That's what makes growth advisory different from other disciplines — the right insight at the right time can change the company's trajectory. But you only find it by doing the diagnostic work first. The biggest trap is spreading advisory time across too many things. Impact comes from focus.