Launch day produces a flood of numbers, and almost all of them are flattering. Impressions climb. Likes arrive. Someone with a large following quote-tweets you and the counter jumps again. By evening the dashboard looks like a success, and nobody can say whether any of it produced a user.

We have run this experiment across 21+ launches, and the pattern holds: the impression total is the least useful number on the page. It counts how many times a post was rendered on a screen, including everyone who scrolled past in under a second. It tells you the algorithm did something. It does not tell you what the launch did for the business.

Here is a plan you can set up before launch day, without a data team.

Why impressions alone mislead

Product Hunt launch, seed-stage AI developer tools: 8.2M impressions, 1,200 signups in the first 24 hours, #1 Product of the Day, and a 340% engagement lift against the account’s baseline.

Only one of those numbers tells you whether to repeat the campaign.

The 8.2M figure is a byproduct. It is what happens when twelve creators post inside a 45-minute window and the algorithm decides the topic deserves distribution. Had the same reach landed on an audience of giveaway accounts, the number would have looked identical and the signup count would have been a rounding error.

Impressions hide failure well. A quote-tweet argument can triple your reach while poisoning the replies. A single large account can supply most of the total, which means your launch is really one person’s audience, borrowed. None of that shows up in the headline metric.

The test we apply to any launch number is simple: what decision does it change? 8.2M impressions changes nothing. 1,200 signups at a known cost per signup changes next quarter’s budget. A 340% engagement lift against a baseline changes how you brief creators the next time.

The three windows that matter

A launch is not one measurement. It is three, taken at different times, each answering a different question.

The first hour asks whether the launch has lift. This is the window the recommendation system reads most closely, and the one where coordinated posting either works or doesn’t. Log the timestamp when the anchor post crosses 1,000, 10,000 and 100,000 impressions. The absolute numbers matter less than the shape: a post that hits 10K in twelve minutes and 100K in ninety has momentum. A post that crawls to 10K over three hours does not, even if it eventually accumulates a respectable total.

The first 48 hours ask whether the conversation is holding. Second-wave quote tweets, founder replies, and the ratio of new followers to new impressions all show up here. Click-through and signup rates also stabilize, which makes this the first honest read on conversion.

Campaign end, usually 7 to 21 days for us, asks what the launch produced. For the Series A beta release that meant 8.5M impressions across three weeks, 3,500 beta signups, and 12% converting to active users. Note the split: 2M of those impressions landed in the first 24 hours. The 24-hour number explained the spike. The three-week number explained the outcome. Reporting either alone tells half a story.

The six metrics worth tracking

1. Impression velocity checkpoints

Three timestamps — time to 1K, to 10K, to 100K — are enough to see whether distribution is compounding or flatlining. They are more diagnostic than an end-of-day total, which cannot distinguish a fast start from a slow trickle.

2. Click-throughs

Track clicks on the launch link rather than profile visits alone. A post that generates argument in the replies but no clicks is a conversation, not a funnel. Compare click-through per creator: a modest account with a high click rate is usually worth more than a large account with a low one.

3. Signups

Count signups in the same windows as the impressions. Signups per 1,000 impressions is the number we argue about internally, because it is the fastest way to see whether reach landed on the right people.

4. Attributed conversions per creator

This is the metric most launches cannot produce, and the one that answers the question founders actually ask: which creator was worth it? It requires the instrumentation below. Without it you have a total, not an attribution.

5. Reply quality

Read a sample of replies and sort them: product questions, technical pushback, spam, silence. We treat technical pushback as a good sign. It means someone with a real problem read the post closely enough to disagree with it.

6. Profile visits

Profile visits measure curiosity that didn’t convert to a click. A spike in profile visits with flat clicks usually means the post was interesting and the profile page failed to close the loop.

Instrumentation before launch day

None of the above survives contact with launch day unless it is wired in beforehand. Three pieces of setup do most of the work.

Per-creator UTM parameters. Every creator gets a distinct link: a shared utm_campaign for the launch, and utm_content set to their handle. We also issue a separate short link per creator, because creators post from mobile and sometimes paste the URL into a reply. The short link keeps the path traceable even when a UTM gets stripped somewhere in the chain.

A landing page that records source. The page should capture UTM values on first touch, store them, and write them onto the signup record. If your signup table has no source column, no amount of campaign tagging will help you three weeks later.

A baseline captured a week before launch. Record daily impressions, engagement rate, profile visits and signup rate for the account and the product across the seven days before launch. Without it, lift is unmeasurable. The 340% engagement lift on the Product Hunt launch is a real number because there was a baseline to divide by. The same campaign without one would have produced a chart with no axis.

One more piece: a named owner. Someone has to watch the first-hour checkpoints while the window is open, because a launch stalling at minute forty can sometimes be rescued with a second wave. A dashboard nobody reads is a post-mortem.

The seed campaign’s honest lesson

The seed round amplification was, by the numbers we published, a strong campaign: 15.6M impressions, 2.1M unique reach, 89K engagements, 14 creators posting 14 threads, and trending in the AI community within 30 minutes.

We could not tell the client which creator drove which signup.

The links went out with inconsistent tagging. Two creators used the homepage. One used a blog post. Several posted screenshots of the product with no link at all, which is a perfectly good post and completely invisible to analytics. When the client asked for a per-creator breakdown, we reconstructed it after the fact: matching signup timestamps against post times, reading replies for mentions, and asking creators what they had actually published. It took days and produced an estimate, not a measurement.

The fix is unglamorous. Attribution is a setup task, not a reporting task. Every creator brief now includes the exact link, a line telling them not to substitute it, and a request to reply with their post URL once it goes live. It costs nothing and decides whether you end up with a report or a reconstruction.

What the report should say

A launch report that runs twenty pages is usually hiding something. Ours fits on one: a table of the three windows with impressions, clicks, signups and attributed conversions; a creator table with the same columns; a short qualitative read on reply quality; and a paragraph on what we would change next time.

Two rules keep it honest. Report what moved and what didn’t, including the creators who underperformed. And do not claim causality you cannot support — a launch that coincides with a Product Hunt front page and a newsletter mention has three plausible causes for every signup, and a good report says so out loud.

The definitions we use for each metric, including what we explicitly do not claim, live in our measurement methodology. The case studies apply those metrics to three real launches. And if your buyers are technical and skeptical, the B2B SaaS creator approach is the closest analogue to what a measured launch looks like when pipeline, not reach, is the point.

The purpose is not measurement for its own sake. A launch you can measure is a launch you can repeat, and one you can improve. Impressions tell you that something happened. The six metrics above tell you what to do next quarter.