Facebook Retargeting Ads: Turning Browsers Into Repeat Orders

Almost everyone who visits your site leaves without ordering. That’s normal, and it isn’t a failure of the site. Someone looked at a $70 catering tray on their phone at 11pm, got distracted, and never came back. They wanted it. They just weren’t going to finish the transaction standing in their kitchen.

Facebook retargeting ads exist for that person. They’re the only ads in your account aimed at people who have already told you what they want, which is why they’re consistently the cheapest orders you’ll buy and the ones small businesses most often never set up.

For a business that delivers, retargeting does two jobs rather than one. It recovers the cart somebody abandoned, and it restarts the customer who ordered once in March and hasn’t since. The second job is worth more, because a delivering business makes its money on the reorder. If you’re still setting the account up, the broader guide to Facebook Ads for businesses that deliver covers objectives, radius targeting and measurement; this piece is about the warm audiences specifically.

The Bottom Line

• Set the audience up before you need it. A website custom audience collects people from the day you create it and can’t backfill, so creating it a month early costs nothing and saves a month.
• Meta needs at least 100 people in a custom audience before it will run, and performs better at 1,000 or more. A 7-day cart-abandoner list at a small business often won’t reach that.
• Standard website custom audiences hold up to 180 days of activity. Since May 2026, audiences built on purchase events can be set as far back as 730 days, which is the setting that makes reorder campaigns possible.
• Segment by how close somebody got. A cart abandoner and a homepage visitor need different offers, and treating them as one audience wastes the cart abandoner.
• Always exclude recent purchasers, unless the whole point of the campaign is the reorder.

What Facebook retargeting ads are

Retargeting means showing ads to people based on something they already did, rather than on who Meta thinks they are.

Mechanically it’s built on two things. The Meta pixel on your website records visits and actions: page views, product views, add-to-carts, checkouts started, purchases completed. Then Meta assembles those people into a Custom Audience you can target like any other, except that this audience has already interacted with you.

You can also build custom audiences from sources other than your site: an uploaded customer list, people who watched a video, people who opened a lead form, people who engaged with your Instagram account, activity in your product catalogue. For most small businesses the website and the customer list do nearly all the work.

The reason the economics are so different from cold advertising is simply audience size. Cold campaigns pay to find interested people among thousands of uninterested ones. A retargeting audience is a few hundred people who are all already interested, so reaching every one of them repeatedly costs very little. It also means these campaigns can’t scale: retargeting harvests demand, it doesn’t create it. You still need cold campaigns filling the top, and the formats worth running for a delivering business apply to both.

The audiences worth building first

Don’t build twelve audiences. Build these, in this order.

• All website visitors, 30 days. Your base audience and the one most likely to reach a usable size. Useful on its own for a general offer, and essential as the pool you carve the others out of.
• Product viewers who didn’t add to cart, 14 days. Interested, but not convinced. Usually a price or a delivery-window objection.
• Add-to-cart, no purchase, 7 to 14 days. The highest-intent group in your account. At a small business this list is short, which is exactly why it converts.
• Checkout started, no purchase, 7 days. Where the delivery fee usually kills the order. Small, and worth its own ad.
• Past purchasers, 180 to 730 days. The reorder audience, and for a delivering business the most valuable list you own.
• Your customer list, uploaded. Email addresses and phone numbers from your ordering system, including the wholesale and catering accounts that never came through a web checkout.

Two constraints shape how you segment. Meta won’t serve an audience with fewer than 100 people in it, and it works noticeably better with 1,000 or more. So if you get 300 site visitors a month, a 7-day add-to-cart audience will never populate, and the honest move is to widen the window to 30 days rather than build a list that sits idle.

How long to keep people in a retargeting audience

Retention window is the setting people leave on default and shouldn’t.

Standard website custom audiences allow a maximum of 180 days. In May 2026 Meta extended that for audiences built on purchase events specifically, up to 730 days, and existing 180-day purchase audiences were migrated automatically. That change matters for anyone selling on a repeat cycle, because it means a customer from eighteen months ago is still reachable.

The right window follows your buying cycle, not a best practice.

• Impulse and occasion products, like a bouquet or a dessert tray, decay fast. A 7 to 14 day window catches the intent while it exists; at 60 days you’re advertising to people whose event has passed.
• Considered purchases, like wedding catering or a corporate account, run on a much longer clock. Thirty to ninety days is reasonable, because the decision takes that long.
• Reorder audiences should be set from your average gap between orders. If regulars order every three weeks, target purchasers at 21 to 45 days, not at 7, because at 7 days they already have a fridge full of your product.

Shorter windows mean smaller, hotter, cheaper audiences. Longer windows mean larger, colder ones. If an audience is too small to run, lengthen the window before you loosen the criteria.

Matching the offer to how close somebody got

The most common retargeting mistake is running one ad against one big audience. A person who glanced at your homepage and a person who abandoned a full cart need different things said to them.

Someone who viewed a product and left usually has an unanswered question. Show that product again with the thing that was missing: the price of delivery, the areas you cover, the cutoff time for tomorrow, a photograph of it packed in the box. No discount. They haven’t earned one and you haven’t established that price was the problem.

Someone who abandoned a cart has a specific objection, and at a delivering business it’s most often the delivery fee appearing at the last step. Address it directly. Free delivery over a threshold, or a clearly stated flat fee, converts this group better than a percentage off the product. This is also the one group where an incentive is worth its margin, because you’re spending it on an almost-complete sale.

Someone who started checkout needs the smallest possible nudge and a reminder that stock or slots are finite, if that’s true. Don’t invent scarcity; the order cutoff you already have is scarcity enough.

Someone who bought once needs a reason to come back on a schedule. The strongest framing for a delivering business is a routine: the same order, standing, every Friday. Reorder ads that show the exact items somebody bought last time do better than generic “we miss you” creative.

Your uploaded customer list is where the wholesale and catering side lives. These people may never have touched your web checkout at all, and the offer is usually an account conversation rather than a product.

Catalogue-based formats carry a lot of this automatically. Meta’s guidance on retargeting with Advantage+ catalogue ads covers the setup, and dynamic creative that shows each person the product they looked at is the single biggest creative upgrade available to a retargeting campaign.

Exclusions: the setting that stops you wasting the budget

Every retargeting audience needs an exclusion list, and this is where most accounts leak.

• Exclude recent purchasers from acquisition and cart-recovery campaigns. Advertising a first-order discount to somebody who ordered yesterday is how you annoy a good customer and teach them to wait for a deal.
• Exclude the hotter audience from the colder one. If you’re running both a 30-day all-visitors audience and a 14-day cart abandoner audience, exclude the abandoners from the broad one. Otherwise they see both ads, the broad campaign takes credit for the conversion, and you can’t tell which one worked.
• Exclude people outside your delivery zone. Retargeting doesn’t override geography. Someone can visit your site from three states away, join your custom audience, and receive ads for a delivery you can’t make. Keep the location targeting on the retargeting ad set exactly as tight as it is on the cold one.
• Exclude current active customers from reorder campaigns aimed at lapsed ones, which means setting the audience as purchasers in the last 365 days excluding purchasers in the last 30.

How much to spend on retargeting

Retargeting should be a small, separate, protected line in the budget, not a share of the main campaign.

Small, because the audiences are small. A few hundred people don’t need $500 a month to reach; they need enough to be seen a handful of times, which is often $50 to $150.

Separate, because if it shares an ad set with a cold audience, Meta will spend most of the money on the bigger pool and the warm audience gets whatever’s left.

Protected, which means the budget sits at the ad set level rather than being pooled at the campaign level, so the system can’t drain it into an ad set with more headroom.

For context on what you’re paying into: WordStream’s 2026 benchmarks put the median Facebook cost per click at $0.60 on traffic campaigns and $0.45 in restaurants and food, and Triple Whale’s August 2026 analysis of more than 40,000 ecommerce brands puts the median cost per thousand impressions at $15.06 and median cost per acquisition at $38.99. A well-built retargeting audience should come in meaningfully under that acquisition figure. If it doesn’t, the problem is usually the offer rather than the targeting. Full benchmark detail sits in the guide to what Facebook ads cost in 2026.

Frequency: where retargeting turns into an irritation

The risk with a small warm audience is that you reach the same two hundred people forty times.

Watch frequency in Ads Manager. Somewhere between two and four exposures a week is productive; above that, click-through rate falls, cost per result rises, and negative feedback starts appearing. On a tight local audience you’ll hit that ceiling faster than any general guidance suggests.

Three fixes, in order of how well they work. Cap the budget, because a small audience with a large budget is the direct cause. Rotate the creative every couple of weeks, which is cheap if you keep a few product photographs in reserve. And shorten the retention window, so people age out instead of being pursued indefinitely.

The tell that you’ve got it wrong is a customer mentioning your ads. Not fondly.

Getting retargeting working in the first month

On day one, create the audiences. All site visitors at 30 days, product viewers at 14, add-to-cart at 14, purchasers at the longest window available. They start collecting immediately and they cannot look backwards, which is the whole reason to do this before you think you need it.

In week one, upload the customer list. Export from your ordering system, upload to Meta, and you have a reorder and lookalike source on day one instead of in six months.

In weeks two and three, wait. Check the audience sizes. Anything under 100 people can’t run yet. Resist the urge to merge everything into one audience to get over the line; widening the window is the better fix.

In week four, launch, smallest audience first. Start with cart abandoners and a delivery-fee offer, because that’s the group most likely to convert and the fastest way to see whether the plumbing works. Then add the product-viewer audience, then the reorder campaign.

From there on, compare cost per order between the retargeting ad sets and the cold ones. If warm isn’t beating cold by a wide margin, something is misconfigured, and the usual culprits are an audience that’s too broad, an exclusion you didn’t set, or a discount you didn’t need to give.

Frequently asked questions

How do Facebook retargeting ads work?

The Meta pixel on your website records what visitors do: viewing a product, adding to cart, starting a checkout, completing a purchase. Meta groups those people into a Custom Audience, and you target that audience with ads the same way you’d target any other. Because everyone in the audience has already interacted with you, the ads reach a small, warm group instead of paying to find interest among strangers.

How many people do you need for a Facebook retargeting audience?

Meta requires a minimum of 100 people before an audience can be used for targeting, and performance is generally better at 1,000 or more. If a narrow audience like 7-day cart abandoners won’t reach 100 at your traffic level, widen the retention window rather than loosening the behaviour you’re targeting.

How long do people stay in a Facebook retargeting audience?

Standard website custom audiences hold up to 180 days of activity. As of May 2026, audiences built on purchase events can be set to a maximum of 730 days. Within those limits you choose the window, and it should match your buying cycle: a week or two for impulse products, a month or more for considered purchases, and roughly your average reorder gap for purchaser audiences.

Is retargeting more effective than targeting new customers?

It produces a better cost per order, but it can’t replace cold campaigns. Retargeting only reaches people who already found you, so its ceiling is set by how many people your other marketing brings in. The two work as a pair: cold campaigns fill the audience, retargeting converts it.

Should retargeting ads include a discount?

Not by default. For someone who only viewed a product, information usually converts better than money off, and a discount trains people to abandon carts. Save the incentive for the checkout-abandoner audience, where the objection is often the delivery fee specifically, and where a free-delivery threshold addresses the reason they stopped.

Can I retarget people who ordered by phone?

Yes, through a customer list upload. Export the phone numbers and email addresses from your ordering system and upload them as a custom audience. This is how catering and wholesale accounts, which rarely pass through a web checkout, get included in your warm audiences at all.

What to do this week

Create the audiences today, even if you have no intention of running a retargeting campaign for another two months. They only collect forward, so every week you wait is a week of warm prospects you can’t reach.

Then start with the smallest and hottest list you have, address the objection that group has, exclude recent buyers, and keep the location targeting as tight as your delivery zone. Retargeting won’t grow your business on its own. It will stop you paying twice for customers you’d already found.