How do you build a weekly growth dashboard?

A weekly growth dashboard is not a vanity metrics wall—it's a forcing function that makes you face one question every Monday: are more people getting real value this week than last? Build it around three to five numbers that reflect user behavior, not activity. Everything else is noise that slows you down.

Start with the metric that precedes revenue

Most early-stage dashboards fail because founders track outputs (signups, page views, MRR) before they understand the input behavior that drives those outputs. The more useful sequence is to ask: what does a user actually do in your product the moment it starts working for them? That moment—call it your activation event—is the single number that should anchor every weekly review.

For a B2B SaaS tool, activation might be 'connected their first data source.' For a marketplace, it might be 'completed a transaction on both sides.' For a developer tool, it might be 'successfully ran a build.' Whatever yours is, put it at the top of the dashboard and track the week-over-week count of new users who hit it. If that number is not growing, nothing downstream—revenue, retention, NPS—will fix itself on its own.

Paul Graham's observation in 'Do Things That Don't Scale' about early-stage startups being fragile is directly relevant here: the difference between momentum and stagnation in the first year is often just 30 days of paying attention to the right signal. A dashboard that obscures that signal with 40 charts is worse than useless—it actively diffuses your focus.

The five rows every early-stage dashboard needs

Keep the dashboard to a single screen. If you need to scroll, you have too much. Five rows is a reasonable ceiling for the first 18 months:

1. **New activated users this week** (vs. prior week, vs. 4-week average). This is your leading indicator of whether acquisition and onboarding are working together. 2. **Retained users from 4 weeks ago** (the cohort who activated four weeks ago—how many came back this week?). Retention is the most honest signal you have about product-market fit. 3. **Top user action by volume** (whatever your core loop action is: messages sent, searches run, exports created). This tells you if existing users are deepening engagement. 4. **Revenue or monetization proxy** (paying users, trial conversions, or a leading indicator like 'users who upgraded plan view'). Even pre-revenue, track what will become revenue. 5. **One qualitative note** — not a chart, a sentence. What did you learn from talking to a user this week that the numbers can't show?

The fifth row is the one founders skip most often and miss most badly. Numbers tell you that retention dropped; a user conversation tells you why. Both are required for a useful weekly review.

How to run the weekly review so it drives decisions

The dashboard is a prop for a weekly conversation, not a report to be filed. The review should last 30 minutes maximum and produce exactly one decision: what changes this week based on what we see?

Structure it as three questions in order. First, what moved? Identify which numbers changed materially (more than 10% week-over-week for early-stage companies counts as material). Second, why did it move? This is where the qualitative note earns its place—connect the number to something you observed in user behavior. Third, what are we doing differently this week as a result? If the answer is 'nothing,' either the dashboard is measuring the wrong things or you're not using the review properly.

A common failure mode is reviewing the dashboard but not updating the action queue. Teams look at falling retention, nod, and then go back to building features they already planned. The review is only valuable if it has the standing authority to reprioritize your week. That means the founder leading the review needs to be willing to drop or delay something when the numbers demand it.

One tactical detail: annotate the dashboard with events. If you shipped a new onboarding flow on Tuesday, mark it. If a competitor launched, mark it. In four weeks you'll have forgotten what changed when, and those annotations become the most useful part of the historical record.

Avoiding the trap of imitating big-company metrics

Paul Graham notes that Garry Tan identified a specific trap early founders fall into: they want to appear big, so they imitate the behaviors of large companies—including measuring what large companies measure. Big companies track brand awareness, net promoter scores across hundreds of thousands of users, and quarter-over-quarter cohort analysis. None of that is useful at 50 active users.

At your stage, 'professional' metrics are often just a way to avoid looking at the uncomfortable truth that only a handful of people care about what you've built. The weekly dashboard should make that truth visible and specific. Which ten users activated this week? Did you talk to them? Do you know their names?

This doesn't mean ignoring structure—it means right-sizing your measurement to your actual situation. A weekly growth dashboard for a startup with 200 users should be simpler and more qualitative than one for a startup with 20,000. The version you build at 200 users that keeps you close to individual behavior will teach you more than any aggregate chart. As you scale, the dashboard evolves; what you're building now is a discipline, not a permanent artifact.

The temptation to build elaborate dashboards with multiple visualization tools, automated Slack reports, and executive summaries is a form of productive procrastination. An hour building the dashboard is an hour not spent talking to users. For the first year, a shared Google Sheet updated manually every Monday is often the right tool—the friction of filling it in manually forces you to actually think about what each number means.

Connecting dashboard signals to fundraising readiness

Paul Graham's fundraising advice makes explicit what most founders learn too late: by the time you're raising a Series A, investors expect to see a trajectory, not a snapshot. The weekly dashboard is what generates that trajectory. Every week you update it, you're building the evidence base that either supports or undermines the story you'll tell in a fundraise.

The specific signal investors care most about in phase 3 fundraising—which Graham defines as the round after your initial experiment—is whether your core metric is on a trajectory toward public-company scale. That trajectory has to be visible over at least six months, ideally twelve. A dashboard you start building at month three is already too late to show a clean trend by month nine.

Practically, this means your dashboard should be structured so that you can export a clean week-by-week chart of your activation metric and your retention cohorts at any time. When an investor asks 'show me your growth over the last six months,' you should be able to pull that chart in under five minutes. If you can't, you don't have a dashboard—you have a collection of disconnected reports.

The secondary benefit of a disciplined weekly dashboard is that it makes you harder to fool by your own optimism. When you see the numbers every week, you can't miss a declining retention trend for two months before noticing it. Early detection of a problem—while you still have runway to fix it—is worth more than any single feature you'll ship this year.

“They want so much to seem big that they imitate even the flaws of big companies, like indifference to individual users.”

— Paul Graham (attributing Garry Tan), source

The one thing to do

This week, identify your single activation event, track how many new users hit it, and put that number somewhere you see it every Monday before you open email.

Frequently asked questions

How many metrics should be on a weekly growth dashboard?

Three to five is the right range for an early-stage startup. More than five metrics means you haven't decided what matters. The goal is one number that immediately tells you whether the week was good or bad, supported by two or three context metrics that explain why.

What tool should I use to build my growth dashboard?

At under 1,000 active users, a manually updated Google Sheet is often the most useful tool because the act of filling it in forces you to think. Automated dashboards in Amplitude, Mixpanel, or Metabase become worth the setup cost once the volume of data makes manual tracking unreliable—typically somewhere between 1,000 and 5,000 weekly active users.

What's the difference between an activation metric and a vanity metric?

An activation metric measures a user doing something that only makes sense if the product is working for them—completing a task, returning after a week, inviting a teammate. A vanity metric measures something that can go up even when the product is failing, like total signups or website visits. If you can make the metric go up without improving user experience, it's vanity.

How do I know if my retention is good enough?

For most SaaS products, a cohort retention curve that flattens above 30% at week 8 is a meaningful signal of product-market fit. For consumer apps the bar is lower, typically above 10-15% at day 30. If your curve is still declining steeply at week 8 with no sign of flattening, no amount of acquisition spend will fix the underlying problem.

Sources

More playbook answers · Growth Prophet home