Performance Marketing

Retargeting Without Waste: Audiences, Frequency and Exclusions

Build retargeting ads that pay: which audiences to target, what to exclude, how to control frequency and how to prove the sales are incremental.

TL;DR
  • Most retargeting waste comes from paying to reach people who would have bought anyway: recent buyers, people already in your email flows, and shoppers who were going to come back on their own.
  • Build a small set of audiences by intent and recency, exclude purchasers and overlapping lists, and control frequency through audience size and budget, since conversion campaigns on Meta and Demand Gen on Google offer no manual frequency cap.
  • Judge retargeting on incremental profit from a holdout test, not on platform ROAS, because retargeting is the campaign type that platform attribution flatters the most.

Why does retargeting waste so much budget?

Retargeting shows ads to people who already visited your store, viewed a product or started checkout. It usually reports the best return in the account. That is exactly the problem.

The people in a retargeting audience are already close to buying. Many of them will come back through a branded search, an email or a direct visit whether you pay for an ad or not. Platform attribution still credits the ad they saw, and with view-through conversions it can credit an ad they scrolled past. The result is a campaign that looks like your most efficient spend while adding little new profit.

Three patterns cause most of the waste:

  • Paying for buyers. Recent purchasers stay in visitor audiences unless you remove them, so you pay to show them the product they just bought.
  • Paying for people your owned channels already reach. A shopper who abandoned a cart is getting your abandoned cart emails and texts. A paid ad on top of that often claims the same sale.
  • Paying too often. Small audiences with large budgets show the same ad many times a week. Extra impressions to the same person rarely change the outcome, but you pay for each one.

The fix is to retarget fewer people, for a shorter time, with clear exclusions, and to measure what the spend adds.

Which retargeting audiences are worth paying for?

Sort audiences by intent (how close someone is to buying) and recency (whether they still care). A cart abandoner from yesterday and a homepage visitor from 90 days ago are very different people.

Audience Suggested window Intent Overlap with email and SMS Role
Checkout started, no purchase 1 to 7 days Very high High if they entered an email Short, low-budget reminder for people you cannot email
Added to cart, no purchase 1 to 14 days High Medium Product reminder with the exact item
Viewed product two or more times 1 to 30 days Medium to high Low Dynamic product ads, reviews, comparisons
Engaged site visitors (time on site, multiple pages) 7 to 30 days Medium Low Category and bestseller creative
All visitors 30 to 90 days Low Low Usually skip, or fold into prospecting as a signal
Past purchasers, outside replenishment window 60 to 365 days Proven buyers High Cross-sell and repeat purchase, often better handled by email

Past buyers are valuable, but your email program reaches them for free. Use paid ads for buyers who are not subscribed or not engaging, and leave the rest to owned channels.

Audience size and platform rules

Small stores used to struggle with minimum list sizes. Google has lowered the minimum audience size for remarketing and customer lists to 100 active users across Search, Display and YouTube, so most stores can now run segmented lists. On Meta, website custom audiences built on the purchase event can now be retained for up to 730 days, up from 180. That helps with long purchaser exclusions, but it is not a reason to retarget anyone for two years.

If you use GA4 audiences shared to Google Ads, you can build these segments once and reuse them across Search, Shopping and Performance Max.

What should you exclude from every retargeting campaign?

Exclusions save more money than any bid strategy. Start with this list:

  • Recent purchasers. Exclude buyers from the last 7 to 30 days, longer for products people buy once. For consumables, exclude them until they approach their replenishment cycle.
  • Active email and SMS engagers in the same stage. If someone is inside a cart flow, give the flow 24 to 72 hours to work before paid ads join in. You can do this by starting the paid window at day 2 or 3 rather than day 0.
  • Customer service and returns visitors. People visiting order tracking, returns or account pages are not shopping. Exclude those URLs.
  • Job seekers and wholesale visitors. Careers and wholesale pages bring traffic that will never buy at retail.
  • Very short visits. A minimum time or page depth rule removes accidental clicks and bots.

Then exclude retargeting audiences from prospecting campaigns, so the same person does not end up in three campaigns at once.

Meta and Google settings that matter

On Meta, the existing customer budget cap in Advantage+ campaigns is no longer available. To limit spend on existing customers now, exclude a customer custom audience in the ad set, or split the campaign into one ad set for existing customers and one that excludes them, then set a spending limit on the existing customer ad set. Our guide to Meta Advantage+ covers that setup.

On Google Ads, customer lifecycle goals let you bid only for new customers in Search, Performance Max, Shopping and Demand Gen. That is the cleanest way to stop prospecting campaigns from quietly becoming retargeting campaigns. It depends on a good existing customer list, so keep that list updated.

How much frequency is too much?

Frequency is the average number of times one person sees your ads in a period. The first few impressions do most of the work, and each extra one adds less while costing the same.

Platform controls are limited. Google’s frequency capping applies to Display and Video campaigns, and Google states that Demand Gen campaigns do not support it. On Meta, frequency caps are available when you optimize for reach or buy through reservation, but not for ad sets optimized for sales. So for most conversion campaigns, you control frequency indirectly.

The lever is simple: frequency is roughly budget divided by audience size and cost per impression. If the audience is small and the budget is large, frequency goes up. Use this to set budgets on purpose.

Worked example (illustrative numbers)

A store has a cart abandoner audience of 4,000 people over 7 days. Its CPM on that audience is $20, so 1,000 impressions cost $20.

  • At $100 per day, the store buys 5,000 impressions a day, or 35,000 a week. Spread over 4,000 people, that is about 8.75 impressions per person per week.
  • At $30 per day, it buys 1,500 impressions a day, or 10,500 a week. That is about 2.6 per person per week.

Now suppose a holdout test shows the $30 version recovers almost the same number of incremental orders as the $100 version. The extra $70 a day, about $490 a week, was buying repetition, not sales. Cutting it adds that amount straight back to contribution margin.

Watch reported frequency and creative fatigue together. When frequency climbs and click-through rate falls, the audience has seen enough. Lower the budget or rotate creative before you raise anything.

How do retargeting, email and SMS work together?

Treat owned channels as the first line and paid retargeting as the backup. Email and SMS cost little per message. Paid ads cost money for every impression.

A practical order for a cart abandoner:

  1. Hours 0 to 24: cart email and, if they opted in, a text. No paid ads yet.
  2. Days 2 to 7: paid retargeting for people who have not purchased, with the exact product and a reason to buy (reviews, shipping speed, returns policy).
  3. Days 8 to 14: lighter frequency, broader creative, or stop.

For shoppers who never gave you an email, paid retargeting is the only way back, so they deserve more of the budget. That is why growing the list with abandoned cart emails without discounts and a solid pop-up also lowers paid retargeting costs over time.

How do you prove retargeting is incremental?

Platform ROAS cannot answer this. It counts every sale where an ad was seen or clicked, including the ones that would have happened anyway. You need a comparison group that saw no ads.

  • Holdout test. Split the retargeting audience randomly. Show ads to one part and nothing to the other, a holdout group. Compare purchase rates. The difference is the incremental effect.
  • Budget on and off tests. Pause retargeting for two to four weeks in a stable period and watch total store revenue, not channel revenue. This is weaker than a randomized holdout but better than nothing.

Then convert the result into profit. Incremental orders times contribution margin per order, minus spend, is the number that matters. Our guide to incrementality testing explains how to size and run these tests.

A low incremental return does not always mean cutting the campaign. It means cutting the parts that add little: broad windows, high frequency and audiences your email program already covers.

Retargeting audit checklist

  • [ ] Every retargeting campaign excludes purchasers from the last 7 to 30 days
  • [ ] Customer service, returns, careers and wholesale pages are excluded from visitor audiences
  • [ ] Audiences are split by intent and recency, not one “all visitors 180 days” list
  • [ ] Paid retargeting starts after email and SMS flows have had 24 to 72 hours
  • [ ] Weekly frequency per audience is reviewed alongside click-through rate
  • [ ] Existing customer spend on Meta is controlled with exclusions or a separate ad set with a spending limit
  • [ ] Google Ads campaigns meant for acquisition use the new customer goal with an updated customer list
  • [ ] At least one holdout test has measured incremental orders in the last 6 to 12 months
  • [ ] Results are reported as incremental contribution margin, not platform ROAS

FAQ

What share of the ad budget should go to retargeting?

There is no correct percentage. Retargeting spend is capped by audience size: once frequency gets high, more budget mostly buys repetition. Size the budget from audience size, CPM and a sensible weekly frequency, then confirm it with a holdout test.

Should retargeting ads include a discount?

Not by default. A discount in retargeting teaches shoppers to abandon carts and wait for an offer, and it reduces margin on sales that might have happened at full price. Lead with reasons to buy, such as reviews, guarantees and delivery times. If you test an offer, test it against a no-offer version and measure the margin, not just the conversion rate.

Is retargeting still possible with privacy changes and consent rules?

Yes, but audiences are smaller and less complete. Browser limits, ad blockers and consent choices remove some visitors from pixel-based lists. Server-side signals such as the Conversions API help with measurement, and customer lists help reach known buyers. Respect consent settings in every region where you require them.

Turn this into a plan for your store

The Free Profit Audit scans your store, sizes your biggest profit leaks and builds a 90-day plan.

Free Profit Audit