How to brief an agency so you get what you wanted

Most bad agency work traces back to the brief, not the agency. A team can be talented and still deliver something you did not want, because the input was a paragraph of wishes and a logo file. You get what you specify. Vague in, vague out.

The frustrating part is that a good brief takes maybe two hours to write and saves weeks of revisions. Yet most briefs skip the parts that actually determine the outcome and over-index on the parts that feel productive to write. Here is how to fix that.

Start with the decision, not the deliverable

A lot of briefs open with "we need a new website" or "we want to run paid social." That is a deliverable, and it is usually the wrong place to start because you have already picked the answer before stating the problem.

Write the decision the work is meant to support instead. Something like: "We need to book 30 more consultations a month, and our current site converts 1.2 percent of traffic." Now the agency knows what success looks like and can tell you whether a new site is even the right lever. Maybe your traffic is fine and your intake form is the problem. You would never learn that if the brief only asked for a redesign.

The test: if the agency reads your brief and cannot argue you out of your requested deliverable, you have written a purchase order, not a brief.

The eight things every brief needs

Skip any of these and you introduce a place where the work can drift from what you pictured. In order of how often they get left out:

• The business outcome. What number moves, and by how much, and by when. Not "more awareness." Revenue, leads, bookings, retention.
• The audience, specifically. Not "small business owners." A plumber with two trucks who does not answer the phone during jobs. The more specific, the better the copy and targeting.
• What you have already tried. Including what failed. This saves the agency from proposing something you burned money on last year.
• Constraints that are real. Budget ceiling, legal review requirements, brand rules you cannot break, a founder who insists on approving every headline.
• The one thing this must not do. Often more clarifying than the goals. "Do not make us sound like a startup" tells a writer more than three paragraphs of positioning.
• Who decides. One name. The person with final sign-off, not a committee.
• What "done" looks like. The specific artifacts you expect to receive and in what format.
• The deadline and what it is tied to. A trade show, a fiscal quarter, a product launch. Deadlines with reasons get respected. Arbitrary ones get gamed.

You will notice most of these are not about the creative work at all. They are about context and authority. That is the point.

Give examples, and say why

"Make it clean and modern" is not a brief, it is a horoscope. Everyone thinks their taste is clean and modern. The words carry no information because they mean something different to each reader.

Instead, pull three examples of work you like and three you do not, and write one sentence about each explaining the reason. "I like this homepage because the pricing is visible without scrolling" is worth more than a page of adjectives. "This competitor's ads feel desperate because every one has a discount" tells the agency exactly which ditch to avoid.

The reasons matter more than the examples. When you say why, you hand over the underlying judgment, which the agency can then apply to a hundred decisions you will never see. When you only show examples without reasons, they guess at your logic and guess wrong.

A brief is not a wish list. It is the transfer of your judgment so someone else can make decisions the way you would.

Watch for taste smuggled in as fact

A common failure: a stakeholder dislikes a color, so the feedback comes back as "this doesn't convert." Maybe it does not. But if the real reason is personal preference, say so. Preference is legitimate. Dressing it up as performance data sends the agency chasing a problem that does not exist and erodes trust when the data does not back it up.

Separate the two in your brief. Here are the things we know from data. Here are the things that are our preference and are non-negotiable anyway. Both are fine. Confusing them is not.

Set the feedback rules before the work starts

The brief is not just the kickoff document. It should also define how you will respond to the work, because the response phase is where most projects fall apart.

Decide up front:

• How many rounds of revision are included, and what a round means.
• Who consolidates feedback. One person collects input and resolves internal disagreements before it reaches the agency. Conflicting notes from five people is how you get mush.
• The turnaround on your side. Agencies build momentum. If work sits in your inbox for two weeks, you pay for the restart in quality, not just time.
• What kind of feedback you want at each stage. Early on, you want reactions to direction. Late, you want line edits. Sending line edits on a first concept wastes everyone's effort, because the concept might not survive.

The single most useful sentence you can add: "At this stage, tell me if the direction is wrong, not if the button should be blue." It calibrates the whole review.

Write it down, then read it back

Once the brief exists, do one thing before you send it. Read it and ask whether a competent stranger could execute against it without calling you. If every third line raises a question, the brief is not done.

Then send it and ask the agency to reflect it back in their own words. If their summary matches what you meant, you are aligned. If it does not, you just caught the misalignment before anyone spent a dollar. This five-minute step catches more expensive mistakes than any revision round.

A good brief also protects you when compensation is tied to outcomes rather than hours, which is how we structure our own work. When both sides agree on the number that has to move, there is no argument later about whether the work "counts." The scoreboard was set on day one.

If you want a partner who will push back on the brief when the requested deliverable is the wrong lever, apply for partnership. Or if you would rather start with a hard look at what your site is actually doing, run the free grader and bring the results to your next brief.