We've all been chased around the internet by the same shoe for two weeks. That's harassment with a budget. Done right, retargeting can be the most efficient spend in your account.

To retarget without being creepy, segment people by how close they are to buying and show each group a message that moves the story forward. Keep it to a handful of impressions a week. Exclude people who've already bought. Then run a holdout test to prove the spend is creating sales.

The lazy setup does none of it. One "all website visitors, last 180 days" audience, one ad on repeat, and an owner wondering why results slide and comments turn nasty.

Why retargeting goes wrong

Retargeting usually fails because it's set up once and left alone. Someone builds a single warm audience and drops in a product ad. The audience is small and the budget keeps running, so the same few people see that ad over and over.

The numbers look fine for a while, because warm audiences were already interested. Underneath, three things are going wrong.

  • The same people are being hit far too often. What felt like a friendly reminder on day two feels like being followed by day ten.
  • Everyone gets the same message. Someone who glanced at your homepage and someone who abandoned a full cart are treated identically, so the ad fits neither.
  • The credit is inflated. Some of the people who "converted from retargeting" would have come back anyway.

Segment by intent, not just visits

A homepage bouncer and a cart abandoner need different messages and different urgency. Build your audiences around where each person is in the decision.

  • Engagers watched a good chunk of a video or interacted with your page, but never visited the site. Top of the funnel. Reintroduce the brand and skip the hard sell.
  • Browsers viewed a page or two and left. They get the same soft treatment.
  • Considerers looked at a product or pricing page, or spent real time on the site. Middle of the funnel. Handle objections here with reviews and guarantees.
  • Hand-raisers added to cart or began a form. Bottom of the funnel. Remove friction with a reminder and a one-tap path back.
  • Buyers come out of the sales sequence and go into a separate audience for repeat purchase or referral offers, where that makes sense.

Each audience should exclude the ones below it, so nobody sees three sequences at once.

Holding back on frequency is a form of respect, and respect converts.

Sequence the story

Good retargeting feels like a conversation, where each impression adds something the last one didn't. We build sequenced funnels where the creative evolves as the relationship warms, so by the time someone buys, it feels like their idea.

A sequence for a considered purchase might run like this:

  1. Days 1 to 3: the "why us." A short video or carousel that explains what makes you different. No discount, no pressure.
  2. Days 4 to 7: the proof. Customer stories, reviews, before-and-after results, behind-the-scenes of how you work.
  3. Days 8 to 14: the objection handler. Answer the question that stops people buying. Price, timing, risk, "will this work for me." A guarantee or a comparison goes here.
  4. Final days: the nudge. A reminder and a simple path back. If you use an incentive, this is the only place it belongs.

The timing depends on how long your buyers take to decide. A $40 impulse product might compress the whole thing into three days. A high-ticket service might stretch it across a month. Either way, never show someone the same thing twice when you could show them the next thing.

Creative wears out fastest in retargeting because the audience is small. If you want to understand how quickly winning ads burn out now, read why your winning ad dies faster than it used to.

Control the frequency before it controls you

Left alone, a retargeting campaign will show one person the same creative again and again in a week. You paid to earn that person's goodwill, and now you're paying to burn it.

  • Pick a ceiling. Our rule of thumb is three to five impressions per person per week for a considered purchase, more for impulse buys and far fewer for anything expensive.
  • Know where a hard cap exists. Google lets you limit how many times ads appear to the same person on Display and Video campaigns. On Meta, a set cap is only available on ad sets with Reach or ThruPlay as the performance goal. On a Meta sales campaign you steer frequency with shorter audience windows, a budget sized to the audience and more creative in rotation.
  • Set an expiry. Every audience needs one. Someone who visited 90 days ago and never came back is cold traffic you are overpaying to reach.
  • Suppress your buyers. Advertising a product to the person who bought it yesterday makes you look careless. Setting up exclusions takes five minutes and nothing else in the account pays back faster.

Suppression depends on the platform knowing who bought. With patchy purchase tracking, buyers slip through and keep seeing ads. That's one more reason to fix your conversion data, which we cover in the boring number quietly deciding your ad results. As a backstop, Meta lets you build a custom audience from your own customer information, which you can then exclude.

The goal is to be seen at the moment the decision is being made. Being seen more often is something else.
Build me a retargeting funnel that converts →

A worked example: the online homewares store

Here's a hypothetical. Say you run an Australian online homewares store spending $3,000 a month on Meta ads, with about a third of that going to one retargeting campaign aimed at everyone who visited in the last 180 days. It's one carousel of best-sellers, running for months. The comments now include "stop following me."

Rebuilt properly, the setup could look like this:

  • Browsers from the last 14 days: a short brand video on how the pieces are sourced and made, plus a carousel of styled rooms. Enough creative that nobody sees the same piece more than a few times a week.
  • Product viewers from the last 14 days: reviews and customer photos of the categories they viewed, then a piece answering the big objections: delivery times, returns, how colours look in real homes.
  • Cart abandoners from the last 7 days: a clean reminder of the item with a one-tap path back to the cart and a clear note on the returns policy.
  • Buyers: excluded from all of the above. After a sensible gap, a separate audience sees complementary products.
  • Holdout: a slice of each audience never sees ads, so you can measure what the campaign is adding.

Same $1,000 or so a month. Fewer wasted impressions, and an honest number at the end telling you whether it's working.

Measure incrementality, not credit

Retargeting looks brilliant in a last-click report because it's standing closest to the finish line, taking credit for a race someone else ran. Run a holdout instead. Keep a slice of the audience unexposed, even 10% of it, and compare conversion rates. If the exposed group doesn't clearly outperform, you are paying to reach people who were coming back anyway.

You don't need a data science team.

  1. Use the platform's own lift test if you qualify. Meta's Conversion Lift compares a test group that sees your ads with a control group that doesn't, and assigns people at random. Google has a version too, but it isn't available to every account and you request it through a Google rep.
  2. If you don't qualify, split manually. Randomly divide a customer or visitor list and only target one half. Compare purchase rates across both halves over the same period.
  3. Run it for a full buying cycle. A few days of data will mislead you.
  4. Look at the difference. What retargeting is worth is the gap between the exposed and unexposed groups. The number of sales the exposed group made tells you nothing by itself.

The same discipline uncovers waste across the whole account. If you suspect your spend is leaking elsewhere too, our ad spend audit walks through where to look.

Common retargeting mistakes

  • One audience for everyone, so the message fits nobody.
  • No expiry on audiences. You end up paying warm prices for cold people.
  • Discounting too early. Offer a discount on the first retargeting ad and you train visitors to leave and wait for one.
  • Advertising to buyers the thing they just bought.
  • Running one creative for months to an audience small enough to see it daily.
  • Trusting last-click, which flatters retargeting every time.

What to fix first this week

  1. Check your frequency. Open your retargeting campaigns and look at average frequency over the last seven days. Anything well above five for a considered purchase is a warning sign.
  2. Add buyer exclusions. Exclude purchasers from every sales campaign, using both your tracking and an uploaded customer list.
  3. Shorten your windows. Replace one long catch-all audience with at least two: recent visitors and high-intent visitors.
  4. Write the objection ad. Ask your sales team or check your inbox for what stops people buying, and make one ad that answers it directly.
  5. Set up a holdout. Even a rough one tells you more than a month of last-click reports.

Be the brand people are glad to see again

Bad retargeting teaches people to resent you. The good kind feels like a follow-up from a business that remembered what you were looking at and has something useful to add. Budget has little to do with which one you get.

That's how our advertising team builds retargeting: audiences mapped to intent, creative that moves the story forward, frequency caps and exclusions set on day one, and holdout tests that prove the spend is earning its keep. If your current setup is one ad chasing everyone for six months, there's money being wasted and goodwill being burned. Let's fix both.

Frequently asked questions

How often should a retargeting ad be shown to the same person?

For a considered purchase, around three to five impressions per person per week is a sensible ceiling. Impulse products can tolerate more, while expensive or sensitive purchases need far fewer. Rotate creative so people see a progression of messages and not one ad on repeat. If comments or your frequency numbers show irritation, tighten the audience window and refresh the creative.

How long should retargeting audiences last?

Match the audience window to how long your customers take to decide. Cart abandoners are warmest within a week, general site visitors within a few weeks, and considered services may justify a longer window. Every audience needs an expiry. Someone who visited months ago and never returned behaves like cold traffic, so paying retargeting prices to reach them is usually wasteful.

How do I know if retargeting is working?

Run a holdout test. Keep a random portion of your retargeting audience from seeing ads and compare their conversion rate with the exposed group over a full buying cycle. If the exposed group converts clearly better, retargeting is adding sales. If the rates are similar, you are mostly paying to reach people who would have returned anyway, whatever the last-click report says.

Should I offer a discount in retargeting ads?

Only at the end of a sequence, and only if your margins and brand allow it. Leading with a discount teaches visitors to leave your site and wait for one, which erodes full-price sales. Start with proof and answers to objections. Reserve any incentive for people who have shown strong intent, such as cart abandoners, and test whether it lifts conversions.