Product launch metrics: seven numbers to read at week 1 and week 4

Product launch metrics are the small set of numbers you agree to read at two fixed dates after launch, not a dashboard you open whenever you feel anxious. This chapter uses seven: live listings, referral sessions, sign-up rate, activation rate, week-4 return rate, correct mentions, and hours per live listing. Read the first three at week 1. Read the last four at week 4. Attach one decision to each number before you look.

What product launch metrics can and cannot tell you

Product launch metrics measure work you control and outcomes you only influence. You control whether a listing is live, whether the facts match, how many hours you spent, and which link you used. You influence sessions, sign-ups, activation and return, but a dozen other things also move them. The ledger keeps that line visible so a good week does not get credited to the wrong action.

The mistake most teams make is reading every number at the same time. Week 1 is too early for retention and too early for a stable activation rate. Week 4 is too late to fix a broken listing link that has been sending people to a 404 for a month. Two read dates force the right question at the right moment.

One boundary before the table: this ledger proves that a change happened and that a channel was live. It cannot prove that the launch caused a sign-up, a citation or a sale. If you need causality, you need a controlled test, and a launch week rarely provides one.

Write the denominator before you write the target. "Get more sign-ups" has no denominator, so it can never be reviewed. "Move sign-ups from 4% to 6% of sessions from launch channels, days 1 to 7" can be reviewed even if the answer is no. The second version is also harder to celebrate by accident, which is the point.

The seven product launch metrics, and when to read each one

Each metric needs a numerator, a denominator and a window. Without all three, two people can report different values from the same week and both be honest. The table fixes those definitions before launch, when nobody is defending a result.

Seven is a deliberate cap. A list of twenty metrics gets a dashboard and no decision. If a number does not change what you do next, leave it out of this ledger and put it in a later review.

Do it in this order

The order matters because the first two steps decide whether the later numbers mean anything. Do them before launch day, not after.

1. Freeze the two read dates and the windows. Week 1 covers days 1 to 7. Week 4 covers days 22 to 28. Write both dates in the same document as the product launch plan. Done when someone else can say when the review happens without asking you.
2. Write the numerator, denominator and window for each metric. Use the table above as the starting point, then change any definition that does not fit your product. Done when two people reading the definition would calculate the same number from the same export.
3. Name the source and the reader for each row. A metric with no named source becomes a guess by week 4. Put the export, report or manual check next to the row. Done when every row has a file, a link or a person attached.
4. Read only reach, intent and corrections at week 1. Check that each listing is live, that the link lands on the right page, and that the name and category are spelled correctly. Fix those first. Done when the live-listing count matches the submission tracker.
5. Read activation, return and hours at week 4, then write one decision per number. Repeat one channel, drop one channel, change one onboarding step, or correct one fact. Done when the ledger has a decision column filled in, not just values.

The deliverable: a two-date metrics ledger

Copy this block into the same document as the launch plan. It is short on purpose. A ledger that takes an hour to update will be abandoned by week 2.

LAUNCH METRICS LEDGER
Product:
Week 1 read date (days 1-7):
Week 4 read date (days 22-28):
Category baseline source (if any):

METRIC | VALUE | WINDOW | SOURCE | DECISION
Live listings |  |  |  | 
Referral sessions |  |  |  | 
Sign-up rate |  |  |  | 
Activation rate |  |  |  | 
Week-4 return |  |  |  | 
Correct mentions |  |  |  | 
Hours per listing |  |  |  | 

ONE DECISION PER ROW
Repeat:
Drop:
Change:
Correct:

The live-listing row comes from the directory submission tracker, not from memory. The tracker records what was submitted and when; this ledger records what is actually visible and what you decided because of it. If the two disagree, the tracker is the evidence and the ledger is the interpretation.

For click and conversion numbers after a listing is live, use link click tracking after a listing. That article owns the click-to-conversion question. This ledger owns the choice of seven launch-stage numbers and the two dates on which you read them.

QueryWin keeps the product profile and the submission record in one place so the source column has something real to point at. See how the product profile and launch record connect if you are still rebuilding the same facts in a new document for every review.

What goes wrong

Three failures account for most useless launch reviews. None of them is a tool problem.

1. The review date moves with the mood. A review scheduled for "when things settle" never happens. Fix the date in the launch plan and treat it like a submission deadline.
2. The denominator changes halfway through. If week 1 counts sign-ups from all sources but week 4 counts only one channel, the comparison is fiction. Freeze the denominator with the definition.
3. The number arrives without a decision. A review that ends with "traffic was okay" produces another review next month. End each row with repeat, drop, change or correct.

There is one more failure that is harder to see: reading a citation check as proof of impact. An AI answer that names your product correctly tells you the entity is legible. It does not tell you the answer was shown to a buyer, or that the buyer acted. We do not know your category baseline, and we will not invent one to make the table look complete.

Common questions

How many product launch metrics should a small team track?

Seven is enough for one launch. If you can maintain fewer, keep live listings, sign-up rate and week-4 return. Those three catch a broken channel, a broken promise and a broken product. Add the others only when a decision actually depends on them.

Should week 1 and week 4 use the same traffic source?

Yes, and write it down. If week 1 uses all sessions and week 4 uses referrals only, the comparison changes two things at once. Pick one definition and keep it for both reads.

What if a channel has not approved the listing by week 1?

Count it as not live, not as zero traffic. The distinction matters: zero traffic is a marketing result, and not live is an operational result. Only one of them is worth fixing this week.

Can I compare my numbers with a public benchmark?

Only if the benchmark states its sample, window and definition. Most public launch numbers do not. A benchmark without a denominator is a story, and it will make your own honest number look wrong.

A launch metric earns its place by changing one decision. If it cannot, it belongs in a later review.