Most paid media teams can tell you their blended ROAS to two decimal places. Fewer can tell you what the number would look like with retargeting pulled out of it, and fewer still can tell you how much of their monthly spend is actually going to warm audiences versus cold. That gap is the whole problem. Retargeting is the one line item in a paid program that almost always clears its benchmarks, and the one most consistently underfunded relative to what it returns.
This piece is not another explainer on how pixels work or what a lookalike audience is. It is an argument about budget: why the retargeting layer is systematically starved, what the organizational and attribution reasons for that are, and how to rebalance without falling into the opposite trap of over-harvesting a small warm pool.
So what should a defensible retargeting budget actually look like?
The Returns Are Not Subtle
Start with the numbers that make the under-budgeting indefensible. On Meta, retargeting medians 3.61x against 2.11x for prospecting, a 71% gap that holds across a dataset of more than 5,000 accounts. Click-through rates are not close either: display retargeting averages roughly 0.7% CTR versus 0.07% for standard display, with retargeted search ads clearing 5.2%. Segmentation widens the gap further. Segmented retargeting lifts CTR 76% and conversions 147% over generic retargeting setups.
The behavioural math underneath is just as blunt. Baymard's meta-analysis puts average cart abandonment at 70.22%, and cart retargeting recovers roughly 26% of those carts on its own. Cross-channel retargeting converts 48% better than single-channel. On mobile, Criteo found retargeted apps see 30% higher retention and 224% more product views per user. These are not marginal effects.
Yet the share of the budget keeps lagging the share of the returns. About 35% of brands spend more than 10% of digital ad budget on retargeting, which means roughly two-thirds spend less than that on the single most efficient dollar they can deploy.
Why the Budget Keeps Getting Starved
If the ROAS is this good, why does the money not flow? Four reasons, in roughly the order you will find them inside a real account.
The first is attribution politics. Retargeting looks suspicious to anyone with an incrementality background, and often rightly so. A warm visitor was going to buy anyway, the argument runs, so the retargeting ad is just taking credit. The directional truth is real: holdout tests often show true incremental lift of 25-30% against reported ROAS. But teams tend to overcorrect, cutting retargeting hard instead of running the holdout and sizing the incremental layer honestly.
The second is channel ownership. Prospecting is run by whoever owns growth. Retargeting is run by whoever owns conversion, email, or CRM. The budget conversation almost never crosses that line, and the retargeting owner usually has less political weight. The result is a cap, not a strategy.
The third is audience ceiling. A retargeting pool is only as big as the prospecting funnel feeding it. Teams see diminishing returns inside the warm segment, correctly conclude they cannot double the budget tomorrow, and then wrongly conclude the current spend is right. The ceiling is real; the gap between current spend and the ceiling usually is not.
The fourth is creative starvation. Most retargeting still runs on static product ads refreshed quarterly. Meta's own frequency research shows CTR drops 40% and conversion drops 60% after four exposures of the same creative. The budget looks maxed because the creative is, not because the audience is. This is a solvable problem, and worth revisiting your creative refresh cadence around.
A Framework for Rebalancing Across Layers
The honest version of a retargeting strategy is three concentric layers, each with its own ROAS target, its own creative refresh clock, and its own ceiling. Treating all warm audiences as a single bucket is the thing that produces the "retargeting is only 8% of spend" problem in the first place.
- High-intent harvest. Cart abandoners, checkout dropouts, pricing-page visitors inside a 7-day window. Target ROAS the highest in the account; budget sized by audience throughput, not by percentage. Refresh creative every two to three weeks.
- Mid-funnel reconsideration. Product-page visitors, repeat sessions, non-converting video watchers at 50% or more. ROAS target closer to the blended account average; this is where sequencing and offer testing earn their keep.
- Lapsed and win-back. Past customers outside their normal repurchase window, dormant list members. Treated as retention rather than retargeting; cadence rules matter more than frequency caps.
The budget allocation that falls out of this looks less like a fixed percentage and more like a demand-driven floor. If the high-intent pool is converting at target ROAS and the pool is growing, the layer gets fed. If the mid-funnel layer is burning frequency without lift, creative gets refreshed before more budget. Deciding which conversion event sits at each layer is a different exercise, and worth doing deliberately against your conversion goal map before you rewire budgets.
What a Defensible Allocation Looks Like
Fixed percentage splits are useful mostly as starting constraints. Published ranges vary more than you would expect: Hawke Media recommends 60-90% prospecting, Criteo's guidance clusters around 10-20% retargeting, and BattleBridge's 2026 analysis settles on 70-85% prospecting and 15-30% retargeting for established accounts. The common thread: retargeting is a harvesting layer, not a growth engine, and the ceiling on it is set by the prospecting funnel upstream.
What this means in practice:
- Under $10k/month in paid spend: 85-90% prospecting, 10-15% retargeting, with retargeting restricted to cart abandoners until the warm pool justifies broader segments.
- $50k-$200k/month: 70-80% prospecting, 20-30% retargeting, split across the three layers above.
- $200k+/month: 65-70% prospecting, 30-35% retargeting, with mandatory holdout tests at least quarterly to size incrementality.
The number to watch is not the percentage but the frequency and the pool size. If retargeting is below 10% of spend and your average frequency is under 2, the layer is underfunded. If retargeting is above 35% and frequency is above 7, the layer is cannibalizing demand that would have converted anyway. Both signals are visible in any competent cross-channel reporting setup, and both get missed in dashboards that look only at blended ROAS.
Running the Incrementality Check
Any argument for more retargeting spend has to answer the "would have converted anyway" question. The answer is not vibes; it is a geo or audience holdout, run on a schedule.
The simple version: for one month, suppress retargeting ads to a randomly selected 10-20% of eligible users (or a matched geographic region), then compare converted-user rates between the holdout and the exposed group. The lift is your real retargeting ROAS. Repeat quarterly, because incrementality drifts with creative, offer, and funnel changes.
Two findings tend to show up. First, high-intent harvest retargeting is almost always genuinely incremental, often 40-60% of reported ROAS. Second, broad site-visitor retargeting is often only 15-25% incremental, which is the finding that justifies narrowing the audience rather than cutting the budget. The segmentation lifts in the broader data back this up: tighter audiences, better creative, better returns. Platform-native tooling can run these splits, and audience segmentation that honors recency and intent is where most of the lift lives.
Where This Fits in the Broader Program
Rebalancing retargeting is a small, high-leverage move inside a larger program. The adjacent questions, which this piece deliberately did not try to answer, are worth putting on the next quarter's roadmap: what the incremental cost of a retargeting layer looks like when added to a mature account, which AI remarketing benchmarks are worth tracking against your own baseline, and whether visitor-level identity resolution changes the audience ceiling enough to justify revisiting the whole budget split.
Retargeting is one of the most consistently underused digital marketing tactics, not because marketers do not know it works, but because the organizational and attribution reasons to under-fund it compound quietly. The budget question is simpler than the attribution debate makes it feel: fund the layer that converts, cap it at the frequency where lift decays, and keep the prospecting funnel wide enough that the warm pool never runs dry. That is a retargeting strategy. The ratio takes care of itself.