Product-Market Fit: Red Flags and Real Signals
Learn how to spot weak product-market fit, assess revenue quality and measure retention. Use practical signals to decide whether to refine, pivot or scale.
Product-Market Fit Is More Than Fast Growth
A startup can add customers quickly, report impressive revenue, and still miss product-market fit. That gap is easier to overlook right now, in an AI market where prototypes can be built fast and many potential customers are willing to try them. But a successful pilot isn't the same thing as a product that becomes part of a customer's day-to-day work.
That distinction matters when founders decide what to build next - and when investors try to figure out whether early traction can last. The real questions aren't only how many people signed up or what the revenue run rate looks like. You also need to know whether customers keep using the product, whether the revenue is likely to continue, and what users would do if the product disappeared.
At TechCrunch Disrupt, Iconiq partner Anu Bharadwaj, Sapphire Ventures partner Rajeev Dham and Superhuman Mail founder Rahul Vohra talked through these issues. Their discussion comes down to a practical framework: separate trials from long-term adoption, measure engagement and retention, and treat product-market fit as something companies have to keep earning.
This article covers the warning signs, the metrics worth tracking, and how to make decisions when the evidence is mixed.
Why Product-Market Fit Is Harder to Read
People often describe product-market fit as the moment when a product meets strong demand. That's a helpful definition, but it makes the process sound cleaner than it usually is. Early signals are messy. A customer might agree to a pilot because the product is interesting, because a team has budget to experiment, or because the cost of trying it is low. None of that guarantees they'll pay and keep using it next year.
AI has made this harder. Teams can put together polished demos quickly, and businesses are generally more willing to test tools that promise productivity gains. That means you can get more pilots - and more pilots that never turn into ongoing use.
As Bharadwaj put it at TechCrunch Disrupt, moving from pilot to production is where things get hard. A pilot can show whether a product seems useful in a limited setting. Production adoption tests something broader: whether the product is dependable, valuable, and practical enough to stick in the workflow.
Treat pilots as experiments, not proof
A pilot is useful only when you're clear about what it's supposed to prove. Before you start, align with the customer on:
- The problem the product should solve.
- Who will use it and how often.
- What baseline you'll compare against.
- Which outcome would justify a paid production deployment.
- Who makes the decision, and when it will happen.
Without that, a pilot can generate a lot of activity without producing real evidence. You might see logins, positive feedback, and a strong demo - but still have no answer to the commercial question: will the customer keep paying and using the product?
This isn't an argument against pilots. They're a sensible way to test when a full rollout is expensive. The problem is treating pilot interest as if it were durable demand. Report them separately, and make the path to conversion clear.
Product-market fit isn't something you "reach" once
Even companies with meaningful revenue can lose relevance. Customer needs shift, competitors improve, and products that once felt differentiated can start to look interchangeable. Dham's point was that companies keep re-earning product-market fit - even when revenue is already in the billions.
That should change how founders think about the concept. Product-market fit isn't a finish line after which you simply maintain the product. It's an ongoing relationship between a product and its market. Growth can suggest the relationship is working, but retention and continued value are what show whether it's holding.
For a related framework, see this guide to customer retention metrics. The goal isn't to track every metric you can. It's to have a consistent way to see whether customer value is strengthening or weakening over time.
Red Flags in Revenue and Customer Adoption
Revenue matters, but the top-line number doesn't tell you enough on its own. Founders and investors need to understand what kind of revenue it is, how concentrated it is, and how likely it is to continue.
Bharadwaj highlighted the difference between annual recurring revenue (ARR) and an annualized run rate. People use these terms loosely, but they aren't interchangeable. ARR usually refers to recurring revenue tied to active subscriptions or contracts. A run rate is an estimate of what revenue would look like over a year if the current pace continues. Even if the run rate looks strong, it doesn't guarantee the revenue will actually recur.
A short pilot that brings in a few months of fees shouldn't be presented as if the customer committed to paying for a full year. The number can be annualized, sure - but the durability isn't proven. Good reporting should let readers separate contracted recurring revenue from temporary, usage-based, or pilot revenue.
Look beneath the top-line number
When you assess revenue quality, ask:
- How much is contracted recurring revenue versus one-off or pilot revenue?
- Are customers live in production, or are they still evaluating?
- How concentrated is revenue among your largest customers?
- What share of customers renew or expand their contracts?
- Can customers replace the product easily or switch to a cheaper provider?
- Do the reported numbers include discounts, services, or usage that might not repeat?
Customer concentration isn't automatically a problem for an early company. A small set of large customers can provide useful feedback and fund development. It becomes risky when the business depends on one or two accounts - especially if those customers have real negotiating leverage or if your product is easy to swap out.
Pricing works the same way. If customers stay only because you're temporarily cheaper than an alternative, the revenue may not be as steady as the contract suggests. Retention depends on more than renewal dates. It depends on whether the product keeps delivering enough value to justify cost and the effort required to keep it in place.
Report numbers so they can be trusted
A good investor update shouldn't hide uncertainty. It should explain how the numbers are built: what counts as ARR, how pilots are treated, whether revenue is contracted, and what time period the figures cover. That's in the founder's long-term interest. If a metric needs a complicated explanation just to sound impressive, it's probably not the metric you should lead with.
That doesn't mean every early-stage company needs a fully mature finance team. But it does mean using consistent definitions. Define the metric once, apply it the same way each month, and call out meaningful changes. A clear view of imperfect traction is more useful than a polished number that obscures how the business is actually performing.
Measure Whether Customers Truly Value the Product
Revenue tells you someone paid. Engagement and retention help you understand whether the product is useful enough to keep. The exact measures vary by product, but the principle stays the same: track behavior that signals customer value, not activity that just looks busy.
For a consumer product, you might look at how often users come back, which features they use, and how long they keep their subscription. For business software, you might track active teams, workflow completion, renewal rates, seat expansion, or use of a core feature. A login isn't automatically proof of value. A user could sign in once to finish an administrative task and never return.
Vohra's example from Superhuman is helpful because it connects a standard financial metric to real customer behavior. The company tracked ARR, but it also built an internal metric called "delighted ARR" - revenue from customers who were highly engaged. The idea wasn't only to ask how much revenue there was. It was to see how much came from customers actively getting value.
Build a metric around the behavior that matters
A useful engagement metric should tie back to the product's job. Start with a short chain of questions:
- What meaningful outcome does the product help a customer achieve?
- What repeated behavior indicates the customer is achieving it?
- What evidence connects that behavior to renewal or expansion?
- How can the team measure it consistently?
For a collaboration tool, the signal might be repeated team use - not a big count of individual accounts created. For a developer product, it might be successful use in a real production workflow, not just a one-time API call. For an AI assistant, it might be recurring use on tasks where the output is reviewed and acted on, not just prompts sent during an initial trial.
Pick metrics that match how your product is actually used. A daily-use target doesn't make much sense for software built for a monthly task. And if your product is meant for daily operational work, relying too heavily on monthly active users can hide a real problem.
Set benchmarks that fit your category and stage. Vohra urged founders to understand what "good," "great," and "excellent" look like in their space. That's more useful than borrowing a benchmark from a different product and treating it like a universal rule.
Pair engagement with dollar retention
Engagement tells you whether customers use the product; dollar retention tells you whether the commercial relationship holds. Track both. A customer might stay active in name while cutting seats or spend. Another might use the product less often but renew at the same level because the product supports an important, periodic task.
For subscription businesses, look at gross revenue retention and net revenue retention alongside customer-level behavior. These metrics answer different questions. Gross retention shows how much recurring revenue stays before expansion. Net retention includes expansion - and it can mask churn when growth from a small number of accounts offsets losses elsewhere.
Segment the data when you can. New and long-standing customers often behave differently. Averages can hide a pattern where a small group loves the product and most customers barely use it. You don't need a dashboard full of charts. You need enough visibility into the underlying customer behavior to guide product decisions.
Decide When to Improve, Pivot or Keep Building
Metrics are inputs to a decision, not the decision itself. Founders should combine behavioral data with real conversations. Ask customers what they value, what they'd miss if the product went away, and what they'd use instead. Then compare those answers with actual usage and renewal behavior.
Vohra also discussed a simple customer question: how would users feel if they could no longer use the product? He suggested that if only 13%, 14%, or 15% of customers say they'd be very disappointed, a founder should consider a major pivot. Around 20%, he argued, there may be room to optimize toward a stronger outcome.
Those numbers are a helpful reference, not a law. Survey answers depend on who you ask, how you phrase the question, and whether your sample reflects your target users. Someone saying they'd miss the product isn't the same as renewing a contract. Treat the survey as one signal - and then check it against retention, engagement, and willingness to pay.
Use a decision framework, not a slogan
When the evidence is weak, separate three common situations:
- Your best customers get real value, but onboarding or distribution is weak. Improve the path to value, onboarding, positioning, or sales process before you change the product's core purpose.
- Customers like the idea, but they don't use the product repeatedly. Check whether the problem is frequent and important enough, and whether the product fits into the actual workflow. A better demo won't fix a weak use case.
- A small group uses the product deeply, but the broader audience doesn't. Consider focusing on that segment. Strong product-market fit often starts with a specific customer and use case, not a sweeping claim about the entire market.
A pivot isn't automatically a failure. At the seed stage, Bharadwaj said she pays less attention to whether the company has already found product-market fit and more attention to whether the founder has a plausible thesis and can adapt as conditions change. Dham also pointed out that founders who have pulled off hard pivots often show solid judgment in a shifting market.
But flexibility isn't an excuse to change direction every time a metric moves. A pivot should follow clear evidence: a problem that isn't urgent, a buyer who won't pay, a workflow that can't support the product, or a segment that consistently shows stronger value than your original target. Write down what the team believes, what evidence would challenge that belief, and when you'll review it.
Improve the right thing before scaling
If a segment shows strong retention and engagement, improving the product for those customers can be a better move than chasing more leads. Vohra described focusing on Superhuman users who were highly engaged. The logic is straightforward: learn from the customers most likely to stay, then figure out what creates their value.
That focus has a trade-off. Serving a narrow group may mean passing on opportunities that look good in the short term. But trying to satisfy everyone can lead to a product that's too generic to be indispensable to anyone. Be explicit about which customers the team is optimizing for - and make that visible in your product priorities.
Before increasing sales or marketing spend, ask whether the product can retain customers at a reasonable rate and whether new customers look like the ones who already get value. Scaling acquisition before validating retention can make the top of the funnel look healthier than the underlying business really is.
For a practical companion, see this guide to validating an AI product idea. It's especially relevant when building a working demo is easy, but the repeat-use case still feels uncertain.
Conclusion
Fast growth can be a useful signal, but it doesn't settle the product-market fit question. Founders need to separate pilots from production adoption, explain revenue clearly, and track the behaviors that show customers are getting ongoing value. The combination of retention, engagement, customer conversations, and honest revenue definitions gives a more reliable picture than any single headline metric.
Use that evidence to decide whether to improve the product for an engaged segment, fix the path to value, or reconsider the underlying thesis. Product-market fit isn't a badge a startup earns once - it's a relationship that changes as the market changes.
Start by reviewing your current pilots and how you define revenue. Mark what's recurring, what's temporary, and which customers actually use the product deeply. That quick check often makes your next move clearer: scale, refine, or pivot.