Ask a jeweler what their retargeting is and you will almost always get the same answer: we chase the people who put something in the cart and left.
That is not retargeting. That is one narrow, pessimistic corner of retargeting, and it is the corner with the fewest people in it and the worst attitude. Somebody who filled a cart and walked away did not forget. They decided. Following them around the internet with the same ring at a slightly more desperate angle is not persuasion. It is asking somebody out who has already said no, twice, in writing.
Meanwhile the people who watched your custom build video to the end, saved three posts, spent four minutes on a product page and never got near a cart are sitting there unspoken to. They are the pool. The cart is a puddle.
Build one broad warm audience of everybody who has shown brand interest, not a collection of clever little segments. Site visitors, video watchers, post and ad engagers, page and profile followers. Start at a 30-day lookback and extend to 60 or 90 only as far as you need to clear roughly 10,000 people, because below that size delivery gets expensive and unstable. Run zero exclusions, including past purchasers, for reasons that will annoy you. Cart abandoners stay in the pool as a slice of it, never as the whole thing. And retargeting cannot save a product nobody wanted at a price nobody accepted, so if the cold side is not working, this will not rescue it.
IWhy the cart is the worst place to build your audience.
Three problems, and they compound.
It is tiny. For a jeweler with normal traffic, the number of people who reach a cart in a month is small. Then you slice it by recency and it gets smaller. You have built an audience so precise that the platform cannot deliver against it economically, which we will come back to, because the size floor is the thing that quietly ruins most of these setups.
It is the most negative signal in the funnel. Everybody else in your warm pool stopped somewhere neutral. They got interested, got distracted, got on with their day. The cart abandoner got all the way to the decision point, looked at the total, and said no. That is a person who has already weighed it. They are not more likely to buy than the browser. For a considered purchase they are frequently less.
It teaches you nothing. A cart-abandoner campaign has one message: come back. That is not a creative strategy, it is a nag with a budget. You cannot learn which angle, which piece, which proof point moves people, because you only ever say the one thing.
Call it the abandoned-cart dragnet. It feels like precision. It is mostly a very expensive way to annoy eleven people.
IIWhat the warm pool actually should be made of.
Everybody who has shown brand interest, in one audience, not fourteen.
- Website visitors. All of them, not just product pages, not just carts.
- Video viewers. Anybody who watched a meaningful chunk of a bench video, a custom build, a stone-setting clip. In jewelry this is often the richest group, because watching somebody make a thing for ninety seconds is a real signal.
- Post and ad engagers. Likes, comments, saves, shares, taps.
- Page and profile followers. They opted in. Talk to them.
- Anybody who messaged, enquired or booked. Obviously.
One pool. The instinct to split this into a dozen tidy segments, each with its own message, is understandable and almost always wrong at jeweler scale. You end up with a dozen audiences that are each too small to deliver, each fighting the others in the same auction, each starved of the events it needs to stabilize. You have not built precision. You have built twelve tiny failures and a spreadsheet to admire them in.
IIIThe size floor, which is where most setups quietly break.
This is the mechanical bit and it is the one nobody mentions.
An audience that is too small does not fail loudly. It just gets expensive. Delivery narrows, CPMs climb, frequency rises because the system keeps showing the same ads to the same handful of people, and the optimization has too few events to work with. Nothing errors. The dashboard looks normal. The money just goes further and further for less and less.
Our published rule from the first 30 days of a fine-jewelry Meta ad account: start at a 30-day lookback, and if 30 days does not get you to roughly 10,000 people, go back further, 60, then 90, until it does. The 10,000 floor is a delivery requirement, not a recency requirement. A 90-day-old visitor in a working audience is worth more to you than a 7-day-old visitor in an audience too small to deliver.
That one adjustment fixes more broken retargeting than any creative change, and it takes four minutes.
IVWhy we run zero exclusions.
Here is the part that will bother you, because it contradicts what nearly everyone does.
The standard practice is to exclude. Exclude purchasers so you do not pay to advertise to someone who already bought. Exclude recent converters. Exclude, exclude, exclude, until the audience is a carefully pruned bonsai of people who have done exactly the right things in exactly the right order.
We do not exclude. Not purchasers, not recent buyers, nobody.
Two reasons. The first is mechanical: every exclusion shrinks the pool, and we just spent a section on what happens when the pool shrinks. Exclusions are the most common cause of an audience sliding under the floor, and they do it invisibly, because the exclusion is a checkbox somebody ticked eight months ago and nobody has looked at since.
The second reason is the business one, and it is the more important of the two. Your best future customer is your last customer. In fine jewelry the second purchase is where the margin lives, the referral comes from somebody who already owns a piece, and the anniversary is a date you already have. Excluding purchasers from your warm audience is paying to acquire somebody, succeeding, and then deliberately never speaking to them again. Ads acquire. Retention compounds. You do not get the compounding if you filter out everyone who compounds.
If the objection is that you do not want to show an engagement ring ad to somebody who bought one last month, the answer is better creative for the warm pool, not a smaller pool.
VWhat to actually say to the warm audience.
Different job from cold, so different content. Cold has to earn attention. Warm has to close the gap between interest and trust, which in a four-figure purchase is the entire distance.
What works: the process. The bench, the setting, the stone selection, the thing being made. Proof that you are real, which means faces, the workshop, the people. Answers to the questions that stop a purchase, which are usually sizing, returns, warranty, lead time and whether it will look like the photo. Reasons to come in rather than reasons to buy now, because for a lot of jewelry the next step is an appointment, not a checkout.
What does not work: the same product shot they already scrolled past, now with a discount on it. Discounting into your warm audience teaches your most interested people to wait for the next one. You trained them. Congratulations, you now have a list of people who will never pay full price again, and you paid to build it.
VIWhat retargeting cannot fix.
Be honest with yourself before you spend here.
Retargeting amplifies what is already happening. If the cold side is genuinely working, retargeting makes it meaningfully more profitable. If the cold side is not working, retargeting changes nothing, because the pool is full of people who already looked at the offer once and passed.
It also cannot fix the website. More interested people arriving at a page that loses interested people is just a bigger sample of the same result. And it cannot fix a product nobody wanted at a price nobody accepted. No audience configuration has ever talked somebody into a piece they did not like.
Retargeting is the easiest place in an ad account to see good numbers, because you are advertising to people who already know you. Treat impressive warm-audience returns with suspicion rather than pride. Some of those people were going to buy anyway, and you are now paying to serve them an ad on their way to the till.
VIISelf-diagnostic: is your warm pool actually working?
The retargeting diagnostic.
- How many people are in your warm audience right now? Not roughly. The number.
- Is that number above 10,000, and if not, how far back does your lookback currently go?
- How many separate retargeting audiences are you running, and could any of them deliver on its own?
- Does your warm pool include video viewers and engagers, or only website visitors?
- What exclusions are currently applied, who applied them, and when did anybody last check?
- Are past purchasers excluded from your warm audience? Why?
- What does your warm audience actually see, and is it different from what cold sees?
- Is there a discount in your retargeting creative? Has there been one for a while?
- Is your cold side producing, or are you hoping retargeting will compensate for it?
- Of your last ten warm-audience conversions, how many would have happened anyway?
If 1 and 2 are a shrug, fix the pool size first. It is the cheapest fix in the account and it is usually the one that was broken.
Straight answers
What is retargeting for a jewelry website?
Showing ads to people who already engaged with your store, rather than to strangers. For jewelry that pool should include everybody who visited the site, watched a video, engaged with a post or ad, or follows the page, not just people who abandoned a cart. The job is different from cold advertising: cold has to earn attention, warm has to close the gap between interest and trust, which in a four-figure purchase is most of the distance.
How big should a jewelry store’s retargeting audience be?
Roughly 10,000 people or more. Below that, delivery narrows, CPMs climb, frequency rises and the optimization has too few events to work with. Nothing errors, it just quietly gets expensive. Start with a 30-day lookback and extend to 60 or 90 days as far as it takes to clear the floor. A 90-day-old visitor in a functioning audience is worth more than a 7-day-old visitor in an audience too small to deliver.
Should I only retarget cart abandoners?
No, and it is the most common mistake in jeweler accounts. Cart abandoners are the smallest group in your warm pool and the one that has already weighed the purchase and declined. They belong in the audience as a slice of it, never as the whole thing. The people who watched a build video to the end or spent real time on a product page without reaching a cart are usually a larger and more persuadable group, and almost nobody is talking to them.
Should I exclude past purchasers from retargeting?
We do not, for two reasons. Mechanically, every exclusion shrinks the pool, and shrinking the pool below the delivery floor is the single most common way these setups break, usually because of a checkbox nobody has reviewed in months. Commercially, in fine jewelry the second purchase is where the margin is, the referrals come from existing owners, and you already know their anniversary. If the worry is showing the wrong ad to a recent buyer, the fix is better creative for the warm audience, not a smaller audience.
What should retargeting ads for a jewelry store actually show?
Process and proof rather than product and price. The bench, the setting, the stone being chosen, the piece being made. Faces and the workshop, because trust is the barrier. Direct answers to the questions that stall a purchase: sizing, returns, warranty, lead time, whether it looks like the photograph. And for most local stores the call to action is an appointment, not a checkout. Avoid putting discounts in front of your warmest audience: you are training the people most likely to buy to wait for the next sale.
Why is my retargeting not working?
Usually one of four things. The audience is too small to deliver, which is the most common and the easiest to fix. Exclusions have quietly starved it. The creative is identical to what cold traffic sees, so it gives an already-interested person no new reason. Or the cold side is not working, in which case retargeting has nothing to amplify, because the pool is made of people who looked once and passed. Check the pool size first; it takes four minutes.
This is part of our fine-jewelry operating series. For how the warm pool gets built in the first place, see the first 30 days of a fine-jewelry Meta ad account. For the diagnostic order when an account is running and not producing, see why your jewelry store’s Meta ads are not producing sales.
Count your warm pool. Get it over the floor. Delete the exclusions nobody remembers adding. Then give the people who already know you something they have not already scrolled past. If you would rather have an operator build it properly, see how we work or email info@endicodatastrategic.com directly. The founder reads every inquiry within one business day.
See how your store scores on every signal in this report.
Written by William Endico and delivered within 5 business days. You get the findings, the fixes, and the order to do them in, whether or not you ever work with us.
Request the audit →
