Retarget visitors with ads that match what they viewed

Most people who visit your site will not sign up on the first visit. Retargeting lets you reach them again while they are still comparing options. A single generic ad for everyone wastes money. Ads that follow up on the exact topic a person looked at feel relevant and give them a clear reason to come back.

Why it works

Few people buy on a first visit because they are still comparing options. Retargeting puts you back in front of them during that comparison, when a timely reminder can tip the decision. Using the pages they read as a guide lets each ad pick up the exact question they had.

These audiences already know your name, so each ad builds on a visit they chose to make instead of starting cold.

How to do it

1. Install the ad platform tags properly. Add the tags for each platform you use, ideally through a tag manager, and tie them to your consent banner.
2. Build audiences by page intent. Create separate groups for pricing page visitors, visitors of each key feature page and blog readers. Use shorter time windows for high intent pages.
3. Exclude people you do not need to reach. Remove existing customers, people who started a trial and employees. Feed your signup or purchase event into the platform to keep exclusions up to date.
4. Write ads for each audience. Pricing visitors might see an answer to a pricing objection. Feature page visitors see a short demo of that feature. Blog readers see a related guide or tool.
5. Set frequency caps. Limit how often one person sees your ads per day and per week. Rotate creative so the same ad does not appear for weeks.
6. Tag every ad link. Use consistent UTM parameters so you can see each audience and ad in your analytics and compare them fairly.
7. Judge on downstream results. Look at trials, demos and paid accounts per audience. Cut audiences that only produce clicks.

Example

A B2B expense management tool builds three audiences: pricing page visitors from the last two weeks, visitors of its receipt scanning page and readers of its guide on expense policies. Customers and current trial users are excluded through a signup event. Pricing visitors see an ad explaining what is included in each plan, feature visitors see a short clip of receipt scanning and guide readers get a free policy template. Each ad link carries UTM tags naming the audience. After a month, the team pauses the blog audience, which drives clicks but few trials, and moves the budget to the pricing audience.

Metrics to track

• Trials, demos or purchases per audience, with the cost of each.
• Lift against a holdout group that sees no ads, to measure what retargeting truly adds.
• Average frequency per person, which should stay inside your caps.
• Spend reaching excluded groups, which should be close to zero once exclusions work.

When it backfires

• Ads follow people everywhere. Without caps, repeated ads feel intrusive and can make people less likely to buy.
• You pay to reach existing customers. Missing exclusions waste budget and confuse users who already pay you.
• Retargeting takes credit it did not earn. Many people would have come back anyway. Compare with a holdout group before scaling spend.
• Tracking ignores consent. Firing ad tags without the right consent can break privacy laws and damage trust.

Questions people ask

How big does an audience need to be for retargeting?

Each ad platform sets its own minimum audience size, and small audiences may not deliver at all. If your segments are too small, combine related pages into one audience or extend the time window.

Do I need consent to retarget visitors?

In many regions, yes. Retargeting relies on tracking tags or cookies, which often need consent under privacy laws. Use a consent banner that controls your ad tags and check the rules where your visitors live.

How long should a retargeting window be?

Match it to how long people take to decide. Use short windows for pricing visitors who are close to buying and longer ones for blog readers, then adjust based on when conversions actually happen.