What should the first 90 days of an AI creative retainer look like, and what does your own team have to put in? This guide sets out an AI creative onboarding 30 60 90 plan for established D2C and consumer brands that already test ads on Meta, TikTok and YouTube. It is written by AI Vidia, a performance creative studio, and it describes how we would run the first three months of our Performance Retainer. The short answer: month one fixes the product, the brand rules and the review routine; month two turns test results into better briefs; month three is where you decide, with evidence from your own ad account, whether the cadence is worth the fee.
The plan uses only the published terms of the retainer: EUR 8,500 a month for 40 finished videos, each up to 30 seconds, delivered in weekly 9:16 batches with a weekly testing log. It does not promise a result. Production is something we control. Performance is something you measure.
Before day one: what to have ready
Most slow starts are input problems, not production problems. A studio can only keep your product accurate if it sees the real product, and it can only write useful variations if it knows what you have already tested. Collect these before the kickoff:
- Product references. Clean photos of every product in scope, including packaging, labels, scale and the details that must never change. Note anything that is easy to get wrong, such as a cap colour or a printed claim.
- Approved claims. The product statements legal or compliance have already cleared, and the ones you cannot make.
- Your current best ad. The ad your media buyer would call the control today, with the metric they judge it on.
- Recent test history. Which angles, hooks and formats you have tried, and what you concluded. A rough list is enough.
- One named reviewer and one backup. The person who can approve or reject a batch without a committee.
- Handoff details. Where finished files go: your ad account, your DAM or a shared folder, and who on the media side receives them.
If you do not have a clear control ad or any test history yet, that is useful information. It usually means a smaller first step is the better choice. More on that below.
Days 1 to 30: lock the product and the review routine
The first week follows our published weekly process: a brief on day one covering brand, audience and the metric the test is judged on; style references and a shot list for you to approve on day two; the first variants for review on day three; your feedback applied on day five; and delivery into your ad accounts on day seven. After that the cycle repeats every week.
The goal of month one is not volume. It is agreement on three things you will rely on for the rest of the engagement:
- What the product must look like. A short list of accepted and rejected examples. If a generated bottle changes shape when a hand picks it up, that clip is rejected, and the reason is written down so it does not recur.
- What the brand must sound and look like. A style system tuned from your own hero imagery, so later batches start from approved references instead of a new interpretation each week.
- How review works. One reviewer, one consolidated round of feedback per batch, and explicit criteria. "More energy" is not a criterion. "Show the product in the first two seconds" is.
Expect the first batches to need the most feedback. That is the point of them. Every rejection with a clear reason makes the next brief shorter.
Checkpoint at day 30: Are the finished videos accurate enough to run without a second internal edit? Is your reviewer spending less time per batch than in week one? If product accuracy is still in question, fix that before adding more concepts.
Days 31 to 60: turn test results into the next brief
By the second month your media buyer should have results on the first batches from your own account. The retainer includes a weekly testing log and a sheet of the variants that performed best. The data comes from your account; our job is to turn it into the next brief.
This is where the difference between a concept and a cosmetic variation matters. Changing the background colour of a winning ad is a variation. Testing a new reason to buy, a different problem the product solves or a different kind of opening is a concept. A useful month two contains both, but it is the concepts that tell you something new. Our guide to building a creative testing matrix explains how to split a batch between them.
Practical habits for this phase:
- Write one sentence per concept stating what it is meant to test, before it is produced.
- Retire an angle after it has had a fair test, instead of producing more versions of it.
- Keep a running list of product and brand rejections, and add it to the brief so it is not repeated.
- Agree with your media buyer how long each batch runs before it is judged, so feedback arrives in time for the next brief.
Our brief template for AI ad production has a structure you can reuse each week.
Checkpoint at day 60: Can you name at least one thing you have learned about your customers that you did not know on day one? Are the briefs shorter and the rejections rarer? If every batch is still a new interpretation of the brand, the style system needs work before month three.