How Much Budget Should Go to Retargeting? A Working Method

Sooner or later every performance account hits the same question: how much money should actually go into retargeting? 'It depends' is the honest answer and the useless one. What is more useful is a method — a way of producing a number from your own funnel data that you can defend in a budget meeting, rather than a percentage copied from someone else's slide.

The method below works from three inputs most teams already have. It produces a required spend first, and a percentage only afterwards. The order matters, because a percentage-first approach is how accounts end up paying premium rates to chase an audience that was never large enough to justify the allocation.

Retargeting can only reach people who have already been to you. Count your non-converting visitors over a month, then decide how long a visitor stays worth re-reaching — the intent window. For a low-ticket store that might be seven days; for a lending application, two weeks or more.

The pool is the number of non-converters still inside that window at any moment, not the monthly total. If you generate ninety-eight thousand non-converting visits in a month and your window is fourteen days, roughly half of them are live at any given time. That figure — not the monthly total — is what your budget has to cover.

One impression rarely changes behaviour. Decide, per segment, how many times a user needs to see you before the message lands — commonly a handful of times a week, and fewer for audiences you have already reached deep inside a form.

This is also where the frequency cap belongs. Beyond a certain number of exposures, extra touches stop producing conversions and start producing complaints. Set the cap first, then let it define the impression volume you need to buy.

Your network or platform can give you an indicative cost per thousand for the inventory you intend to use. Multiply pool size by touches per user by weeks, divide by a thousand, and you have the monthly spend the plan requires. Nothing in that calculation is a convention or an industry benchmark; it is arithmetic on your own funnel.

Assume a month of one hundred thousand paid visits converting at two percent. That is two thousand conversions and ninety-eight thousand non-converters. With a fourteen-day intent window, roughly forty-nine thousand unique users are live at any moment.

Six impressions per user per week across that pool is about two hundred and ninety-four thousand impressions a week, or roughly 1.27 million a month. At an indicative four dollars per thousand, the plan costs about five thousand dollars a month.

Now the percentage appears — as an output, not an input. Against a fifty-thousand-dollar media budget, that is ten percent. Against a fifteen-thousand-dollar budget, the same plan is thirty-four percent. Identical pool, identical plan, different allocation, because the right share of spend depends on how much prospecting is feeding the pool.

Long consideration cycles and document-heavy funnels — lending, insurance, high-value gaming — justify a larger share, because the distance between abandoning and returning is short and a recovered user is worth a lot. Impulse purchases and low-ticket carts justify a smaller one.

Demand conditions tilt it too. When prospecting auctions get expensive, your existing pool is comparatively cheap to work, so a temporary tilt toward retargeting can be the efficient choice. In the quiet weeks after, cheap prospecting inventory is the moment to rebuild the pool you are about to spend against. The reasoning behind treating the two budgets as one system, rather than as competitors, is covered in our piece on splitting a performance budget.

One caution. A plan that only ever looks efficient is probably collecting credit for conversions that would have happened anyway. The only clean check is a holdout — suppress a random slice of the audience and compare the two groups. Run one before you defend a large increase.

Once the total is set, divide it by segment rather than by placement. Recent, deep-funnel abandoners get the first claim on budget; shallow bouncers get the remainder and a low cap. Inside a single platform the same rule applies — the audience that started your form is worth more per impression than the audience that read one page, and bidding both the same rate is the quietest way to waste a retargeting budget.

Three things decide whether the allocation was right: coverage (what share of each segment actually got reached before its window closed), incremental conversions (what the exposed group produced that the suppressed group did not), and blended cost per acquisition across prospecting and retargeting together.

Review coverage weekly, reallocate monthly, and re-run a holdout quarterly. If coverage on your highest-intent segment is incomplete and blended cost is holding, the answer to 'how much budget' is simply: more. If frequency is climbing while incremental conversions are flat, it is less — and the pool, not the percentage, is the reason.