A developer signup is not a verdict, activation can be

Developer tools are sold to people who read the documentation before they have an account. That single fact
breaks the number most devtool marketing plans are built on, and it is worth being precise about why.

I work as a marketing engineer. The devtool version of the job starts with one question: which event in this
funnel happens often enough that a channel could be judged on it?

Why the signup cannot be that event

A number can serve as a verdict only if it moves when the thing being tested changes. A developer signup moves for
a release, a thread, a conference talk and a docs page — for people who signed up to read a quickstart and will
never call the API. It is cheap to produce and almost impossible to interpret.

An SDK install is worse: it can be an install in a Docker image, a mirror or a CI job. The event that carries
information is the first successful call — the moment the tool did the thing it is sold for.

The floor is not a detail

Every unit has a floor. At a 3% base rate and a +20% relative lift, a strict two-variant test needs 13,914 events
per arm — 27,828 in total. At 500 contacts a day that is 56 days for one test.

Now put the devtool funnel on top of that arithmetic: if activation happens 300 times a month, each arm sees 150
events a month, and the floor of 13,914 is 93 months away. That is the honest number, and it is the reason so
many devtool campaigns end the quarter with a dashboard and no verdict.

Which leaves three moves, and only three:

1. One test, not four. At that volume, the quarter can finish one honest test. Name it.
2. A more frequent unit. Read activation as the money number and judge the channel on the step above it, where volume exists.
3. Say it out loud. Write down that the month produces a progress report about an unreadable unit. That is a legitimate choice; pretending otherwise is not.

What I would ask for in a devtool marketing hire

Four artifacts, and none of them is a dashboard:

1. The definition. Which call, on which surface, with which key counts as an activated account — one sentence, agreed by whoever will argue about it next quarter.
2. The owned number, with its volume. One number for the month and the events behind it in the same sentence.
3. The stop rule. Cost per activated account, named before the spend and checked at 48–72 hours.
4. The artifact that survives the month. Something the next person hired can read.

My packages are public so they can be compared this way: $900 Sprint writes the definitions, the tracking and the
stop rules; $1,900/month Engine runs one channel end to end with an owned number each month; $2,900 Full Build
covers the whole loop.

The devtool-specific version of the argument is on the devtools page,
and the free calculators are here: which step of your funnel can fund a verdict,
how many verdicts your volume can finish
and what one verdict costs to finish.
Same arithmetic in a terminal: npx marketing-mindset unit --steps visits:12000,signups:900,activations:270,paid:40.

If your funnel cannot name the unit it is judged on and the volume that unit needs, the next hire will not fix it.
One page of definitions and one date on the calendar will.