Favors.dev
Product launch metricsare the numbers that tell you whether a launch changed anything durable, as opposed to how much attention it collected on the day. There are five worth a solo founder's time: qualified visitors per surface, activation rate of the launch cohort, day-7 retention of that cohort, permanent assets earned, and replies that became relationships. Everything else on the standard launch scorecard is applause.
The measurement guides you find on this topic are written for product marketing teams. They list thirteen or nineteen KPIs across market share, win rate, share of voice, NPS, and feature adoption. All of them are real metrics, and all of them need a sales team, a survey panel, and an analytics hire to populate. Hand that list to one person who just launched on Product Hunt on a Tuesday and the honest outcome is that they track none of them, conclude measurement is for bigger companies, and judge the launch by the upvote counter like everyone else.
So here is the version scoped to one person with a spreadsheet and an afternoon. Five numbers, three checkpoints, one page.
Why don't upvotes tell you whether a launch worked?
Because upvotes measure the size and responsiveness of the audience you had before launch day. That is a real asset and worth building, but it is a fact about your network, not about your product. A founder with a 4,000-person mailing list and a mediocre product will out-rank a founder with a great product and no list, every single time, and both of them will read the ranking as a verdict on the thing they built.
The same logic disqualifies most of the launch-day dashboard. Impressions measure that a platform rendered your name near someone's thumb. Total signups measure how low your signup friction is. Ranking measures who else happened to launch that morning. None of them survive contact with the question that actually matters: did the people who arrived do anything?
| The number you watch | What it actually tells you | Track this instead |
|---|---|---|
| Upvotes, points, badges | How many people your existing network could mobilize in 24 hours | Qualified visitors, the ones who went past the landing page |
| Launch-day traffic | How wide the top of the funnel was, for exactly one day | Activation rate of that same traffic |
| Impressions and reach | That a platform rendered your name near someone's thumb | Day-7 return rate of the launch cohort |
| Total signups | How low your signup friction is, not how good the product is | Signups who reached first value |
| Ranking on the day | How you compared to whoever else happened to launch that Tuesday | Permanent assets earned: links, listings, indexed threads |
| Comment count | Volume of conversation, including the one-word ones | Replies that turned into an ongoing relationship |
There is one legitimate use for the vanity column: as a mechanism, not a result. Ranking higher on a directory puts you in front of more people, which shows up downstream as qualified visitors. So watch the ranking on the day if you like, then judge the launch a week later on what those visitors did. If the ranking went up and activation stayed flat, you successfully mobilized a crowd and converted none of it. That is a specific, fixable problem, and one you cannot even see from the leaderboard.
What are the five launch metrics that actually matter?
Each of these earns its place by changing what you would do next. If a number moves and your behavior does not, it is not a metric, it is a mood. Read the trap under each one. The traps are where founders lose the signal, not the definitions.
Visitors arriving from a launch source who took one deliberate action beyond landing: opened a second page, started the demo, or scrolled to pricing.
How to measure it: A UTM on every link you control, plus one event fired on that second action. Segment by utm_source so each surface gets its own number.
The trap: Counting sessions. A launch always produces a traffic spike; the spike is the least informative thing it makes.
The share of launch-week signups who reached first value, the single step where your product does the thing it promised.
How to measure it: Tag every account created during the launch window with the launch name, then divide activated by total for that tag alone.
The trap: Using your all-time activation rate. Launch traffic is curious rather than motivated, so it almost always activates below baseline. Compare it to your own previous launch, not to your steady state.
The share of that tagged cohort who came back and did something seven days after signing up, once the attention is gone.
How to measure it: Same cohort tag, one returning-session query run on day 8. It is a single number and it is the most honest one you will collect.
The trap: Checking on day 2. Everything looks alive on day 2. Day 7 is where curiosity and need separate.
The things the launch left behind that keep working: live directory listings, editorial backlinks, indexed discussion threads, reviews, and citations inside AI answers.
How to measure it: A manual list, plus Search Console's external links report a fortnight later. Count only what resolves and stays live.
The trap: Assuming a submitted listing is a live listing. Roughly a third of directory submissions never publish, and nobody emails to tell you.
The count of named humans you are still in conversation with two weeks after launch day: testers, reviewers, people who filed a real bug.
How to measure it: A list of names in a text file. There is no tool for this and there does not need to be one.
The trap: Treating it as soft. This number is the input to your next launch, which means it compounds while every other number on this page decays.
Metric four deserves an argument, because it is the one that has quietly become more valuable and is still the one founders skip. The durable output of a launch is not the traffic, it is the material the launch leaves lying around the internet: a listing on a directory that ranks, a thread under your Show HN, a review someone wrote, a write-up on your own domain. That material is what gets read back to people later, increasingly by machines.
The scale of that shift is easy to over- and under-state, so here is the sober version. AI assistants sent 1.13 billion referrals to the top 1,000 websites in June 2025, up 357% year over year, according to Similarweb data reported by TechCrunch, with ChatGPT accounting for more than 80% of them. Google Search sent 191 billion over the same period, so this is a fast-growing sliver rather than a replacement. But the answer, not the click, is what is being competed for now, and the engines differ in what they feed on. Similarweb's citation analysis and the 5W State of AI Citations 2026 both find ChatGPT leaning heavily on Wikipedia and Reddit, Perplexity citing far more sources per answer and rewarding attributed, named-author pages, and Google's AI Overviews overlapping substantially with the top-20 organic results. Reviews, community threads, and linked write-ups are exactly that diet. Your launch produces them. Count them.
How do you instrument a launch in one afternoon?
Three things, and none of them require a new tool. Add UTM parameters to every link you control, fire one event when a visitor does something deliberate, and stamp a cohort tag on every account created during the launch window. That is the whole setup, and it takes about two hours the week before you launch.
One: tag the links.UTM parameters are the standard way campaign traffic gets attributed, and they cost nothing but discipline. Google's documentation on collecting campaign data with custom URLs is the reference; you only need four of the nine parameters it defines. Decide the values now and write them down, because the one thing that ruins this is inventing a new spelling of your source halfway through launch day.
| Parameter | Values to use | Why it earns its place |
|---|---|---|
| utm_source | producthunt · hackernews · indiehackers · newsletter | Which room the person walked out of. This is the field you segment every other metric by. |
| utm_medium | launch | Keep it constant across the whole launch so one filter isolates launch traffic from everything else forever. |
| utm_campaign | v2-launch-aug26 | The launch's name. Reuse it as the cohort tag on the account record so traffic and users join up. |
| utm_content | maker-comment · gallery-link · bio-link | Which link inside a surface did the work. Cheap to add, and it settles arguments later. |
One caveat worth knowing before you trust the numbers: you can only tag links you write yourself. The maker comment, your gallery links, your bio, the newsletter. Those are yours. The link a stranger copies into a Slack is not, and neither is the one an aggregator republishes. Expect a meaningful chunk of direct traffic during launch week that is really launch traffic wearing a disguise, and resist the urge to explain it away.
Two: pick one qualifying event. Not five. One action that a curious visitor would not take by accident: starting the demo, opening a second page, hitting the pricing anchor. It is the difference between a visitor and a qualified visitor, and it makes every surface comparable on the same scale.
Three: stamp the cohort. When an account is created during the launch window, write the campaign name onto the account record. One column. This is the piece almost nobody does, and it is the piece that makes the day-8 read possible. Without it you have aggregate signups and no way to isolate the people who came from the launch. Reuse the same string as your utm_campaign so traffic and users join up cleanly.
For the fourth metric, one more free tool covers most of it: Search Console's external links report lists the sites linking to you, grouped by root domain. Check it two weeks after launch, not two days. Links need to be crawled before they appear, and the report is explicitly a sample rather than a complete index. Pair it with your own list of submitted listings, because a directory link that never published will never show up in any report at all.
How do you read the launch cohort a week later?
On day 8, you compare four numbers against each other rather than against anyone else's benchmark. There is no industry-standard activation rate that means anything for your product, and hunting for one is a way of avoiding the read. What matters is the shape: which of visitors, activation, and retention is high and which is low. Each combination points at a different problem, and they need opposite responses.
High visitors, low activation
A positioning problem, not a distribution one
The wrong people came, or the right people misread the promise. Fix the first screen before you launch anywhere else.
Low visitors, high activation
A distribution problem, not a product one
The product works on the people who find it. Do the same launch again on three more surfaces.
High activation, low day-7
A first-week value problem
First value landed, second value never did. Find what the returning few did on day 3 and make it the default path.
Low on everything, good assets
A slow success
The launch failed as an event and worked as an investment. The listings and links keep compounding. Do not conclude the product is dead.
The most common misread is the second one. Low traffic feels like failure, so founders conclude the product is wrong and start rebuilding, when the cohort is quietly telling them the product is fine and almost nobody saw it. If activation and day-7 held up on a small cohort, the correct response is not a redesign, it is four more launches. That is the argument the first 100 users playbook makes at length, and the cohort numbers are how you know it applies to you rather than merely sounding reassuring.
The second misread is impatience with metric four. Two weeks out, most of what your launch earned has not surfaced yet: directory listings sit in moderation queues, backlinks wait to be crawled, reviews get written by people who used the thing for a fortnight first. A launch that looks flat on day 8 and rich on day 21 is completely normal, and the launch week playbook is built around exactly that asymmetry: the spike is the smallest thing the week produces.
The one-page launch dashboard, column by column
One sheet, one row per surface, eight columns. Not one row per day, because days blur together and tell you nothing, while surfaces are the unit you actually make decisions about. Build it before launch day with the rows filled in and the numbers blank, because a dashboard you create afterwards gets built to flatter whatever happened.
| Column | What goes in it | Where it comes from |
|---|---|---|
| Surface | Product Hunt, Show HN, each directory, each newsletter. One row per surface, not one row per day | Your launch plan |
| Visitors | Sessions attributed to that utm_source | Analytics, filtered to utm_medium=launch |
| Qualified | Of those, how many fired the second-action event | Same filter, plus the event |
| Signups | Accounts created carrying that source | Your own database, cohort tag |
| Activated | Of those signups, how many reached first value | Your own database, cohort tag |
| Day 7 | Of those signups, how many returned on or after day 7 | Run on day 8, never earlier |
| Assets | Links, listings and threads this surface left behind, with URLs | Manual, re-checked at two weeks |
| Names | People from this surface you are still talking to | Manual, and worth more than the six columns before it |
Three checkpoints, then stop
- Launch day + 24h: visitors and qualified visitors, by surface. Nothing else is readable yet, and staring at the dashboard on the day is a way of not answering comments.
- Day 8: activation and day-7 retention for the cohort. This is the real verdict.
- Day 14: permanent assets and names. Re-check every listing, then close the sheet.
After two weeks, stop attributing anything to the launch. Traffic still arriving is just traffic, and counting it as launch traffic flatters you into repeating whatever you did.
If you run more than one launch, and you should, keep every sheet in one file, one tab per launch, identical columns. The comparison across launches is worth more than any single launch's numbers, because it is the only place you can see whether you are getting better at this. On Favors.dev the launch calendar gives you the same view for the crowd side: which rallies drew real help, and who showed up for them.
The launch retro template
Run this within 48 hours of the day-8 read, alone, in writing, in under thirty minutes. Six questions. The value is entirely in doing it while the details are still available and the disappointment is still specific.
What did I expect each surface to produce, written before launch day?
Without the prediction, every result becomes retroactively unsurprising and you learn nothing.
Which surface produced the highest qualified-visitor rate, not the highest volume?
That is the surface to lead with next time, regardless of how it felt on the day.
What did the people who activated have in common?
This is your positioning, discovered rather than guessed. Rewrite the headline with their words.
What is still live two weeks later, and what quietly died?
Separates the permanent assets from the confetti. Re-submit anything that never published.
Who helped, and have I paid it back yet?
The single strongest predictor of how the next launch goes, and the one no dashboard tracks for you.
What would I refuse to do again?
One line, written while it still stings. It is the only part of a retro you will actually reread.
Question five is the one that quietly determines the next launch. The people who showed up, tested the thing, and wrote an honest review did it because of a relationship, and relationships have a balance you can run down. The launch-day checklist treats assembling that crowd as a pre-launch task, which it is. But paying it back is a post-launch task, and it is the one that never makes it onto anybody's dashboard.
The metric that compounds
Four of these five numbers decay. Traffic ends, cohorts age, even links lose relevance. The names do not. Every founder who tested your product, filed a real bug, or left an honest review is still reachable for the next launch, and they arrive warmer than they did the first time. That is the mechanic Favors.dev makes explicit: founders trade verified marketing help through a points economy, so the help is metered and nobody spends what they have not earned. Your dashboard tracks what a launch produced. The leaderboard tracks what you have banked for the next one.
Frequently asked questions
What are product launch metrics?
Product launch metrics are the measurements that tell you whether a launch changed anything durable about your product's position, as opposed to how much attention it attracted on the day. For a solo founder that reduces to five: qualified visitors from each launch surface, activation rate of the launch cohort, day-7 retention of that same cohort, permanent assets earned (backlinks, live listings, indexed threads, reviews), and the number of replies that turned into ongoing relationships. Enterprise launch scorecards add market share, win rate, NPS and share of voice, but those need a sales team and a survey panel to populate, so they measure nothing useful at a one-person scale.
How do you measure the success of a product launch?
Measure the behavior of the people who arrived, not the size of the crowd. Tag every launch link with UTM parameters and every account created during the launch window with a cohort tag, then wait. On day 8 you look at three numbers for that cohort: how many took a real action on the site, how many reached first value, and how many came back a week later. Two weeks out you count what the launch left behind: live listings, links, threads, reviews. A launch that produced twelve returning users and four permanent links beat a launch that produced six hundred upvotes and silence, and the two are easy to tell apart once the cohort is tagged.
Are Product Hunt upvotes a good measure of launch success?
No. Upvotes measure how many people your existing network could mobilize within a single 24-hour window, which is a real thing but not a thing about your product. They correlate with the size and responsiveness of the audience you had before launch day, not with whether strangers found the product worth using. Treat ranking as a distribution mechanism. A higher position puts you in front of more people, which shows up in qualified visitors. Then judge the launch on what those visitors did next. If upvotes went up and activation stayed flat, you mobilized a crowd and converted none of it.
How long should you measure a launch for?
Two weeks, with three checkpoints. Launch day plus 24 hours gives you traffic and qualified visitors by surface, and nothing else worth reading. Day 8 gives you the honest ones: activation and day-7 retention of the cohort, once the attention has drained away. Day 14 is when you count permanent assets, because directory listings and backlinks take days to publish and get indexed. After two weeks, stop measuring the launch and start measuring the product. Anything still arriving is ordinary traffic, and continuing to attribute it to the launch will flatter you into repeating whatever you did.
What launch metrics should a solo founder ignore?
Ignore impressions, reach, follower growth, total signups without an activation denominator, and any ranking that resets at midnight. Ignore net promoter score until you have enough users for the arithmetic to mean something. Ignore anything you cannot instrument in an afternoon, because a metric you have to maintain a pipeline for is a metric you will stop collecting by the second launch. The five that survive all share one property: each takes under a minute to read and each one changes what you would do next.