Most retargeting budgets get set the same way: someone points at the abandoned-cart number, someone else quotes a legendary ROAS figure, and a line item appears in the plan for the next quarter. Then it underperforms, gets defended anyway because the platform dashboard says it worked, and quietly grows. Priced honestly, retargeting is a useful channel with a narrow window of profitability. Priced sloppily, it is a tax on conversions that would have happened for free.
This piece walks through what a retargeting line item actually costs when you fund it: the media, the tools around it, the creative refresh, the measurement work, and the two levers that decide whether any of it pays back.
What size does the audience have to be before retargeting is even worth budgeting?
The Media Line Is Only Part of the Bill
The media invoice is the visible cost. It is also the smallest surprise. In 2026, the average Meta Ads CPM sits at $13.48 across industries, roughly 20% above 2025, with categories ranging from Food & Beverage near $2.82 to Finance around $29.16. Retargeting inventory typically clears above the blended average, because the audience is smaller and other advertisers are bidding on the same warm pool. Retargeting CPCs on Meta usually land under $1, which flatters the reports even when incremental value is thin.
Around the media, four cost lines almost always show up and rarely make it into the first budget draft:
- Tooling. Pixel and server-side event forwarding, tag management, an audience-sync layer, and a place to see the numbers side by side. Some of this is free; the pieces that are not tend to bill per event or per tracked visitor.
- Creative refresh. Warm audiences see the same ads repeatedly. Two to four fresh variants per active segment, per month, is a realistic floor.
- Measurement. Platform ROAS is not enough on its own. A quarterly incrementality read (holdout or geo test) has real cost in foregone revenue.
- Human time. An hour a week to prune audiences, swap creative, and check frequency is the minimum sustainable cadence.
A useful mental split for a new program: expect the media to be about 60-70% of the total, with tooling, creative, and measurement absorbing the rest.
Audience Size Is the First Gate
Before the CPM conversation matters, the audience has to be big enough to serve. This is where most small programs quietly fail. Meta's Custom Audience minimum is 100 people, but 1,000 or more performs meaningfully better. Google reduced its remarketing minimum to 100 active users across display, search, and YouTube in late 2025, down from a 1,000-user Search floor. LinkedIn requires 300 members in a segment before a campaign will serve.
Minimums are the floor, not the target. There is a second problem underneath them: the pool you can actually address is smaller than your traffic. Meta's Conversions API typically recovers only 20-40% of the conversions and audience signal lost to iOS and browser tracking restrictions, so a site with 50,000 monthly visitors may have a matched retargeting pool a fraction of that size once identity resolution is done.
The practical implication: if the monthly warm pool at your standard 30-day window sits below a few thousand matched users per platform, retargeting spend is going to bounce off frequency caps before it can produce a meaningful lift. Extend the window, stack sources (website plus engagement plus customer list), or accept that a real retargeting program starts later. Building and sizing those pools cleanly is a job in itself; a good audience builder does most of the plumbing so the spend decision is about strategy rather than tag debugging.
Frequency Is the Second Gate, and It Is Cheap
Once the audience is large enough, the next lever is how often each person sees an ad. This is the most under-priced control in the whole program. The 2026 sweet spots are 5-7 impressions per user per week for prospecting and 3-5 for retargeting; beyond those bands, incremental CTR collapses and brand sentiment measurably erodes.
Cost stacks up quickly here. A 25,000-person retargeting audience served four impressions per week at a $15 CPM is roughly $1,500 a week in media at the recommended ceiling, or about $6,500 a month. Push frequency to eight and the media doubles, the audience gets annoyed, and the incremental CTR is worse than at four. Frequency discipline is functionally free; ignoring it is what turns a $6,500 line item into a $13,000 one that performs worse.
The Real Cost Is Cannibalization
The most expensive line in retargeting never appears on any invoice. It is the conversions the platform reports that would have happened anyway.
The mechanic is simple. A user who has already visited a site has shown intent, so the retargeting ad may not meaningfully change their decision, yet click-based attribution can make the ad look highly effective. Baymard's meta-analysis puts the average cart abandonment rate at 70.22% across 50 studies, and a share of those abandoners return on their own. Retargeting takes credit for many of them.
How much? Incrementality studies suggest roughly 15-30% of platform-reported retargeting conversions are non-incremental. If cannibalization is ignored, retargeting CAC will look lower than it really is, and the actual ROI can be much lower or even negative. That gap between reported and real is the number that decides whether the line item earns its budget.
Two habits close the gap. First, exclude recent purchasers and, for shorter cycles, exclude anyone who visited the confirmation page within the same session. Second, run a holdout at least once a quarter: withhold retargeting from a random 10-20% of the eligible audience, and read the conversion delta between held-out and served groups. A neutral view across platforms makes this readable in a day rather than a week; a proper cross-channel view is the difference between a decision and an argument.
A Working Budget Shape
The numbers below are illustrative shapes, not promises. They exist to give the line items proportion.
For a mid-market DTC or B2B program with a matched retargeting pool of about 25,000 people across web and engagement sources:
- Media: roughly $6,000-$9,000 a month at a compliant 3-5x weekly frequency and a blended $13-$18 retargeting CPM.
- Creative refresh: $1,500-$3,000 a month to keep 6-10 fresh variants live across segments. AI-assisted production has bent this line down materially; the trade-off is brand control, covered in more depth in this piece on AI creative.
- Tooling and measurement: $500-$2,000 a month depending on whether audience sync, server-side events, and cross-channel reporting are consolidated or bought separately.
- Incrementality holdouts: budget 10-20% of retargeting media as foregone revenue, once a quarter.
Two program design choices matter more than any of these line items. The first is what you count as a conversion in the first place; getting that wrong makes every downstream number lie, which is the topic of this guide on conversion goals. The second is the intent signal driving audience membership: a page-view retargeting pool and a pricing-page-plus-return-visit pool behave nothing alike, which is why visitor intent scoring tends to lift retargeting ROAS more than any bid tweak.
What a Payback Read Actually Looks Like
Payback is not the ROAS in the ads dashboard. It is contribution margin on incremental orders, minus all four cost lines above, divided by the total spend including foregone revenue during holdouts. If that number is above one, expand. If it is between zero and one, tighten frequency, shrink the pool to higher-intent segments, and rerun. If it is negative, the honest answer is to spend less and let organic return traffic close the sales it was going to close anyway.
Retargeting is a lever, not a growth engine. It compounds a good top-of-funnel program and papers over a bad one just long enough to hide the problem. Budget it that way and the line item earns its place; budget it on last-click ROAS alone and it will quietly eat the margin it appears to be creating.