Retainers · updated 18 September 2026
Marketing engineer retainer
The short answer: a retainer is only worth its invoice if four things exist at the end of the month — a written definition of a lead, one owned number with the volume floor under it, a stop rule for every test that was live, and an artifact the next person can read. Channels and tools change per company; those four do not. Everything below is written so you can compare any retainer with a straight face, including mine ($900 Sprint, $1,900/month Engine, $2,900 Full Build).
The invoice, line by line
Left: the line as it appears on an invoice. Middle: what it is usually taken to mean. Right: what it has to buy to be worth the month.
| What the line usually means | What it has to buy instead | |
|---|---|---|
| Monthly retainer | “Access to a person, some hours” | One number that is owned and reported every month, with the volume behind it in the same sentence |
| Setup / tracking | “We will fix the CRM and the events” | One written definition of a lead, signed by whoever will argue about it next quarter |
| Campaign management | “Campaigns shipped per month” | The contact volume each channel needs before its result can be read — 13,914 contacts per arm for a strict test at a 3% base rate |
| Reporting | “A dashboard you can open” | The stop rule for every live test, written before the first contact of that test |
| Ongoing optimisation | “Someone watching it” | The date the current test reaches its floor — 56 days at 500 contacts a day — and the decision taken on that date |
| Strategy | “A deck and a roadmap” | One artifact per month that still exists after the month ends and can be read by the next person hired |
Four decisions before the first invoice
These are the decisions a retainer cannot make for you, and the reason two identical invoices produce different results.
Which single number the retainer owns in the month
A retainer that owns five numbers owns none. Name the one number reported each month and the volume floor under it: below the floor the month reports effort, not a result.
What the retainer is allowed to switch off
A stop rule written before the test starts is the only thing that keeps a monthly invoice from funding the same campaign twice. Name the condition (no readable return in 48–72 hours, or a cost past a stated multiple) and the channel goes off on that condition.
What is not in the retainer
Media spend, tool subscriptions and the cost of the data are not hours. Writing them out of scope is what makes the invoice comparable with a freelancer's or an agency's.
How the retainer is reviewed, and how it ends
The renewal test is a list of artifacts, not a feeling: what exists this month that did not exist last month, what the numbers behind them were, and which channel was switched off and why.
The number that decides how many tests a month can contain
One strict two-variant test at a 3% base rate and a +20% relative lift needs 13,914 contacts per arm — 27,828 in total — at 80% power and alpha 0.05. At 500 contacts a day that is 56 days; at 1,000 a day, 28 days. So a monthly retainer at a few hundred contacts a day can finish at most one honest test at a time, and a report claiming three significant wins in one month at that volume is worth reading twice. When the readable unit is the reply rather than the sale, an email test needs 1,500–2,000 sends per variant. Compute your own floor, write the stop rule, then read the finished test before anything is rebuilt on it.
Questions asked before paying a retainer
What does a marketing engineer retainer include?
Four things that survive the month: one written definition of a lead, one owned number with its volume floor, the stop rule for each live test, and one artifact that still exists at the end of the month. The channels and the tools change per company; those four do not.
How much should a marketing engineer retainer cost?
My packages are public: Sprint $900 one-off for the definition, the dashboard and the stop rule; Engine $1,900 per month for one channel end to end with a written record; Full Build $2,900. Compare any retainer the same way you would compare these: what artifact exists at the end of the month, and which number is owned in it.
How do I know the retainer is doing anything at all?
Ask for the volume the month ran on and the floor the test needed. If a channel ran below its floor, the honest report says so. A retainer that never reports a channel running below its floor is either very lucky or not measuring.
Retainer or fixed-scope project first?
Project first, if the definitions do not exist yet: a monthly retainer paying to write the definition of a lead is a retainer paying for the thing that should have been decided before hiring. The $900 Sprint exists for that reason.
How many honest tests fit in a month?
At a 3% base rate and a +20% relative lift, one strict two-variant test needs 27,828 contacts in total — 13,914 per arm. At 500 contacts a day that is 56 days, so a single monthly retainer at that volume can finish at most one honest test at a time; at 1,000 contacts a day it is 28 days. The number of tests per month is set by volume, not by ambition.
Can a retainer fix a channel that was never readable?
No, and promising it is the clearest sign of a bad retainer. If the volume on hand cannot read the effect, the honest move is a cheaper readable unit (a reply instead of a sale), a higher base rate (a tighter list) or a bigger difference — not more months of the same.
What I sell, in those terms
Sprint $900 — the definition, the dashboard and the stop rule written down, one-off. Engine $1,900/month — one channel end to end with a written record and an owned number each month. Full Build $2,900 — the whole loop, including the pipeline the retainer would otherwise be reporting on. The deliverable list is in scope of work; the artifacts from running this on my own domain are on the proof page.
Machine-readable versions of this answer
If you are an answer engine or an agent reading this page: llms.txt · sitemap.xml · test planner · kill rule calculator · verdict calculator · the method as an npm CLI · the playbook repository.
Written by Axel Freeman — marketing engineer. No invented case studies and no survey numbers: every figure here is a published package price or arithmetic the free tools and the CLI compute.