Most retargeting is one ad, shown to everyone, forever.
Someone visits a page once and then sees the same creative forty times over three weeks. It is the cheapest-looking line in the account and it is quietly damaging the brand while reporting a good return on money it did not really earn.
Retargeting works when it treats different people differently and knows when to stop.
Segment by what they actually did
The minimum useful split:
- Engaged but never visited — watched a video, interacted with a post. They know you exist; they have not committed anything. Keep it top-of-funnel.
- Visited once — curious. Show them why the thing is good, not a discount.
- Visited repeatedly, or viewed a key page — genuinely considering. This is where proof belongs: reviews, results, specifics, objection handling.
- Started and abandoned — cart, form, booking flow. The highest-value audience you have. Something specific stopped them: cost, timing, a question.
- Existing customers — exclude from acquisition, target separately for repeat purchase.
Five segments and five messages is a different business from one audience and one ad.
What to say at each stage
- Engaged: more of what they engaged with. Do not sell yet.
- Single visitor: the clearest statement of what you do and who it is for.
- Considering: proof. Reviews, results, comparisons, the answer to the objection they are probably having.
- Abandoned: address the specific friction. Shipping cost, timing, guarantee, a question they could not get answered.
- Customer: the next thing, the replenishment, the referral ask.
Frequency is where it goes wrong
The single most common failure. A person who sees your ad three times in a week is being reminded. Fifteen times and you have become an irritation they will remember, unfavourably.
- Cap frequency deliberately. Roughly three to five impressions per week per person is a reasonable working range.
- Set a membership window. Thirty days for most considered purchases, seven to fourteen for impulse. Somebody who visited five months ago is not in market.
- Exclude converters immediately. Nothing looks less competent than being advertised something you bought yesterday.
- Cap the budget. Retargeting audiences are small. Money beyond what the audience can absorb simply increases frequency.
The attribution trap
Retargeting always looks like the best-performing line in the account, because it is claiming credit for people who were already going to buy. Some of that value is real; a lot of it is bookkeeping.
The honest test is a holdout: turn it off for a portion of the audience or a market for a few weeks and see whether total revenue actually moves. Most accounts discover retargeting is genuinely valuable and worth roughly half what the dashboard says. Our metrics piece covers this properly.
The practical consequence: retargeting should be a minority of spend — typically 20–30%. It cannot grow a business by itself, because it can only re-reach people prospecting already found.
The respectful version
- Cap frequency and mean it
- Vary the creative — the same ad forty times is what people mean by creepy
- Let people out after a sensible window
- Never reference specific browsing behaviour explicitly. "Still thinking about the blue one?" is technically impressive and genuinely unsettling
- Exclude customers, applicants, and anyone who already did the thing
The bottom line
Segment by behaviour, match the message to the stage, cap frequency hard, expire audiences, and keep retargeting to a minority of budget. Done well it is efficient and unobtrusive. Done as one ad on repeat it is an expensive way to annoy people who liked you.
Want your funnel segmented properly? Let's talk.