Remarketing that does not annoy people

Everyone has been chased around the internet by a pair of shoes they already bought. That experience is why most business owners feel uneasy about remarketing, and they are right to. Badly configured remarketing is cheap, easy to launch and actively damages the brand it is supposed to help. Done properly it is one of the highest return things in a paid account.

What makes it annoying is almost always the setup

Three settings cause the vast majority of the irritation:

• No frequency cap. The default behavior will happily show one person your banner dozens of times a week. There is no version of your business where that is persuasive.
• No membership duration limit. Someone who visited once four months ago is not a warm prospect. They are a stranger who has now seen your logo 300 times.
• No exclusion of converters. Advertising a product to someone who bought it yesterday is the single most visible way to look like you are not paying attention.

Fixing those three takes fifteen minutes and removes most of the harm. Everything else is refinement.

Segment by intent, not by the fact that they visited

One list called all visitors, one banner, one message. That is what most accounts run and it is why the results are mediocre. The person who read a blog post for 20 seconds and the person who abandoned a checkout are not the same audience and should never see the same ad.

A workable segmentation for most businesses:

1. Deep intent, recent. Reached a pricing page, started a form, added to cart, in the last 3 to 7 days. Highest bid, most specific message, and it should reference exactly what they were looking at.
2. Deep intent, stale. Same actions, 8 to 30 days ago. Different message. They did not buy for a reason. Address the reason: delivery time, warranty, financing, a comparison, a real customer story.
3. Shallow visitors. Homepage and content readers. Low bid, low frequency, brand level message. Many of these are not prospects at all.
4. Existing customers. Excluded from acquisition ads entirely, and used for a completely different campaign about the next thing they might buy.

The gain here is not just politeness. Separating these lists lets you spend properly on the small group that is actually close to buying instead of averaging your budget across everyone who ever loaded a page.

Set an end date, because interest expires

Pick a membership duration that matches your real buying cycle. If people decide within a week, a 30 day window is already generous and 540 days is absurd. If you sell something with a genuine six month research cycle, longer windows make sense, but the message must change over that time rather than repeating.

The same applies to frequency. My default starting point is a handful of impressions per user per week, not per day, and I would rather be too quiet than too loud. You can always increase it if the campaign is under delivering.

The consent situation changed, and it changes list sizes

Since mid 2026 consent is the single control for what data flows into advertising. In practice, only users who have granted advertising storage consent are eligible for your remarketing lists. If your consent banner is not set up correctly, or if a large share of your visitors decline, your lists will be smaller than your analytics traffic suggests and some of them will not reach the minimum size to serve at all.

Two consequences worth planning for:

• Do not build a strategy that requires very granular lists if your traffic is modest. You will end up with a dozen audiences that are all too small to run.
• Check your consent setup before you conclude that remarketing does not work for you. A large fraction of the audiences I have seen described as broken were configuration problems, not demand problems.

It is also worth remembering that Google now labels this area as your data rather than remarketing. Same idea, different menu.

How to build a campaign people do not resent

1. Exclude converters and existing customers first. Before anything else.
2. Build the four segments above. If your traffic is small, start with two: deep intent recent, and everyone else.
3. Cap frequency deliberately rather than accepting the default.
4. Write ads that acknowledge the situation. The best performing remarketing creative I have used is not a discount, it is the answer to the objection that stopped them. Sizing, delivery, contract length, what happens on the call.
5. Change the creative every few weeks. Fatigue is real and it is the main reason a good campaign turns into an annoying one over time.
6. Exclude your own placements where they do not belong. Mobile apps and games can produce enormous cheap impression volume with near zero value.
7. Measure incrementally where you can. Remarketing takes credit easily because it reaches people who were already going to come back. Pausing it for a period and watching total conversions is crude, but it is more honest than the reported numbers.

Where the line actually is

My rule is that a remarketing ad should be able to appear in front of the customer while they are sitting next to you without embarrassment. If the message assumes they have forgotten you and needs to shout, it is too aggressive. If it reads like a useful reminder from a company they were considering, it is fine.

Sensitive categories deserve extra care. Health, finance, legal, anything personal. Being followed around by an ad for a medical service is a different experience from being followed by an ad for a chair, and the rules around what you may build audiences from are stricter for good reason.

What to do next

Open your audience settings and check three things today: is there a frequency cap, how long is the membership duration, and are converters excluded. Fixing those three usually improves both results and reputation in the same afternoon. If you want someone to look at whether your lists are even being populated correctly under the current consent rules, book a free consultation.