A Shopify brand we took over had been working with their media buyer for 4 months. ROAS had moved from 2.7 to 2.9 over 6 weeks of activity — A/B tests on button colors, interest targeting adjustments, bid cap tweaks. Technically busy. Strategically stuck. When we reviewed the brief they'd given the buyer, the problem was clear: 2 pages covering brand voice, audience demographics, and campaign objectives, with "maximize ROAS" as the north star. We rewrote it to 4 specific directives. ROAS hit 4.1 in the following 4 weeks. Same buyer. Same account. Different brief.
What a vague brief actually costs in ROAS points
A brief that doesn't name a specific problem leaves a media buyer optimizing for activity rather than outcomes. They run tests — because testing is what media buyers do. They report on metrics — because metrics are what clients expect. But without a specific, measurable problem statement, the tests and metrics are chosen by the buyer's judgment, not the business's actual priorities.
For this brand, the binding constraint was a 1.6% landing page CVR from paid traffic. That was the ceiling on ROAS. No bidding adjustment was going to overcome a broken post-click experience. But "maximize ROAS" was interpreted as "optimize targeting and bids" — which is reasonable given what the brief said. Six weeks of bid strategy work on a broken landing page produced 0.2 ROAS points. Fixing the landing page produced 1.4.
The brief didn't give the buyer the right problem to solve.
The 4 directives we use to replace a standard brief
When we write briefs for accounts we manage, we skip the brand overview and go straight to constraints and rules. Four directives cover what matters:
Directive 1: The primary constraint — the single metric that, if improved, would most move ROAS. Name it with its current value and target. "Landing page CVR is 1.6% from paid traffic. Target is 3.0%. This is the primary focus for weeks 1–3."
Directive 2: What not to touch — the things that are working and should be left alone. We list specific campaign and creative IDs that are performing and should not be paused or restructured without a direct conversation first.
Directive 3: The test format — what types of tests to run, at what budget, and how long before a decision. "All creative tests run at $40/day for 5 days. Winner declared on landing page CVR from paid traffic, not CTR. No test gets additional budget before 5 days are complete."
Directive 4: The one reporting metric — the single number we want in every weekly update. "Report on landing page CVR from paid traffic and ROAS from Shopify UTM data. No other metrics unless you're flagging an anomaly."
What the rewritten brief looked like
Original: 2 pages, brand voice, audience personas, campaign goals, preferred platforms.
Rewritten: 1 page, 4 sections, 400 words.
Section 1: "The problem. Landing page CVR from Meta traffic is 1.6%. This is the ROAS ceiling. Target is 3.0% CVR before we increase budget. Everything else is secondary."
Section 2: "What's working. Campaigns X and Y are stable at 3.1 ROAS. Do not restructure. Creative IDs A, B, C are not to be paused without discussion."
Section 3: "Test format. Creative tests at $40/day for 5 days per variant. We test landing page angles by running identical ads to two different landing page URLs. Decision criterion is 5-day landing page CVR via Shopify UTMs."
Section 4: "Weekly report: landing page CVR from Meta traffic this week vs last week. Add ROAS from Shopify if it moved more than 0.3 in either direction."
ROAS at week 4: 4.1. The buyer knew exactly what to do.
Rewrite your brief using this structure
Take whatever document you currently give your media buyer. Replace it with 4 sections: the constraint (one metric, current value, target), what not to touch (specific campaigns and creative IDs), the test format (budget per test, duration, decision metric), and the one weekly number.
If you can't name the primary constraint, the first step is identifying which metric is the binding constraint on ROAS. That diagnostic comes before the brief — and it's where most accounts are actually stuck.