How should you run a weekly ecommerce creative meeting?

01

What should happen before the weekly creative meeting?

Participants should review performance asynchronously, while test owners prepare compact decision cards for the live meeting.

Each card should show the original hypothesis, variable tested, control and variant, primary success metric, guardrails, evaluation window, audience, offer, placement, launch date, result, measurement caveats, and proposed decision. Google recommends forming an experiment hypothesis around a business objective, selecting one or two metrics, testing one variable at a time, and retaining experiment findings for future planning.

Place the original hypothesis at the top of the card and do not edit it after launch. A new interpretation may be worth testing, but it is a new hypothesis—not retroactive proof that the completed test worked.

Set a submission deadline early enough for participants to read the cards. Items without a hypothesis, evidence window, result, and proposed decision stay out of the live queue. This prevents the meeting from becoming an improvised dashboard review.

02

What is the 45-minute weekly creative meeting agenda?

Reserve the meeting for resolving uncertainty, approving work, and committing resources.

The timing below is an operating recommendation rather than a rule prescribed by an advertising platform. Assign a facilitator to control scope and a recorder who can update decisions and roadmap items in real time.

Do not use a separate meeting document that must later be translated into production work. The decision record, brief, and roadmap should reflect the decision before the group proceeds.

1. 0–5 minutes — Confirm constraints and close old actions. Mention only changes that could affect decisions, such as inventory, promotions, tracking problems, spend conditions, or production capacity. Close, reassign, or reschedule every overdue action.
2. 5–10 minutes — Triage the decision queue. Confirm that each item has its original hypothesis, predetermined metric, evidence window, current result, and proposed decision. Return incomplete items for asynchronous preparation.
3. 10–28 minutes — Decide completed and active tests. Read the hypothesis, compare evidence with the pre-agreed rule, identify the creative learning, choose one queue, and record the next action. Review high-impact or disputed items first.
4. 28–38 minutes — Approve production. Compare proposed briefs by evidence strength, likely commercial importance, production effort, urgency, novelty, and whether the pipeline is overconcentrated in one concept.
5. 38–43 minutes — Reconcile the roadmap. Add approved launches, extend continuing tests, create iteration briefs, archive stopped work, and adjust production order for actual capacity.
6. 43–45 minutes — Read back commitments. State each decision, deliverable, owner, due date, and roadmap status. Resolve missing ownership before ending the call.

03

How do the four creative decision queues work?

The queues define what happens next: launch approved work, continue gathering evidence, iterate a specific element, or stop further distribution and production.

Launch means a new test, derivative, or broader creative rollout is sufficiently defined and approved. Record the locked hypothesis, variable, success metric, guardrails, audience, offer, asset owner, launch date, and review date. A recommendation to scale media should still be executed and monitored in the relevant ad platform.

Continue means the hypothesis is still viable but the evidence is immature. Specify what evidence is missing, why it could change the decision, and the exact review date or threshold. Google advises Performance Max advertisers to account for conversion lag, so a short evaluation window may not tell the full story.

Iterate means preserving an element supported by evidence while deliberately changing another. TikTok’s Video Insights guidance recommends examining engagement peaks, comments, topics, and video elements to identify hooks or themes to reuse or modify. An iteration brief should explicitly state “keep,” “change,” “reason,” and the new hypothesis.

Stop means the test failed its predefined rule, became strategically irrelevant, showed sustained deterioration, or is unlikely to resolve useful uncertainty with more investment. Stopping should not delete the asset or its learning. Preserve the result so future teams do not unknowingly repeat the same test.

04

How should the team evaluate creative evidence?

Separate the media outcome, creative observation, and next-test implication instead of treating one blended metric as a complete explanation.

The media outcome describes what happened in the ad platform: for example, changes in delivery, cost, conversions, or conversion value. The creative observation identifies what viewers encountered: the hook, message, topic, creator, format, proof, offer, or call to action. The implication states what the next brief should preserve or change.

TikTok advises using video-level insights to investigate moments of engagement and individual creative elements. Google says Performance Max asset reporting can help advertisers examine asset and asset-group results and identify themes to prioritize. These diagnostics can generate useful hypotheses, but they do not by themselves prove why an overall business result changed.

Do not change several dimensions and then claim one caused the result. When possible, make the next iteration interpretable by changing one principal variable. Record attribution problems, small samples, conversion lag, promotion changes, or audience differences as caveats rather than hiding them.

05

How do you prevent post-result storytelling?

Judge the result against the test definition that existed before launch, then capture any new explanation as a separate hypothesis.

Start each review by reading the original statement: “We believe changing X for audience Y will affect metric Z because…” Then ask whether the evidence met the agreed rule. Only after answering that question should the group discuss additional observations.

If someone proposes an explanation that was not tested—such as the creator, season, offer, or audience causing the result—record it in a hypothesis backlog. Do not use it to redefine the completed experiment. This distinction protects the learning record from hindsight bias and gives the proposed explanation a fair future test.

For consequential or ambiguous outcomes, schedule a deeper postmortem rather than consuming the entire weekly meeting. The weekly forum should decide whether deeper analysis is warranted and assign its owner and deadline.

06

What must be recorded before a discussion is closed?

A decision is incomplete until it creates an owned deliverable and changes the relevant roadmap or production status.

Use one durable record: “Test or concept; original hypothesis; evidence reviewed; decision; reason; learning to retain; next deliverable; owner; due date; roadmap updated; next review trigger.” Atlassian’s meeting guidance similarly recommends time-boxed agenda items, advance materials, recorded decisions, and action items that specify responsibility and deadlines.

The deliverable must be concrete. “Explore new hooks” is not sufficient; “Write three problem-led opening hooks that retain the current creator and proof sequence” is actionable. The owner should be one person, even when several people contribute.

Roadmap updates should happen live. Add approved production, create iteration briefs, extend continuing tests, and archive stopped items while the rationale is fresh. This makes capacity conflicts visible and reduces the chance that meeting conclusions disappear into notes.

07

Where does ATIYO fit in this workflow?

ATIYO can hold the creative context, decisions, briefs, iterations, assets, roadmap changes, and reusable learnings produced by the meeting.

Media performance remains in the ad platform, and ATIYO preserves the creative context and learnings around that performance. A user must record the relevant performance evidence; ATIYO does not connect to ad accounts, buy media, calculate ROAS, or automatically know campaign results.

Teams can use ATIYO to keep the hypothesis with the creative, document why an item was launched, continued, iterated, or stopped, turn an approved decision into production work, and retain the lineage between an original asset and its iterations. That makes it a creative-strategy operating system rather than a replacement for platform reporting.

Regardless of the software used, apply the same operating rule: no item leaves the meeting without a decision, rationale, owner, due date, and updated production status.