You have a good base of happy customers. You know roughly who they are: homeowners in the suburbs, a lot of them working couples, mostly found you through word of mouth. If only you could find more people exactly like them. That is the promise of lookalike audiences on Facebook: give Meta a list of your best people, and it looks for others who resemble them.
The promise is real, but smaller than it sounds for many small businesses. Lookalikes depend heavily on the quality of what you feed in, and Meta’s broad targeting now does a similar job by itself in many cases. This post explains how similar audiences work, how to choose a good seed audience, when they help a small business, and what to try instead when they do not.
How lookalike audiences on Facebook work
A lookalike audience starts with a source, sometimes called a seed. This is a custom audience you already have: customers from a list, website visitors who bought, people who submitted a lead form, or people who engaged heavily with your Instagram. Meta studies the shared traits of the people in that source, drawing on everything it knows about how they use its apps, and then finds other people in your chosen country who resemble them.
You choose two things when creating one:
- The location where Meta should look, usually a country, sometimes a region.
- The size, on a scale from a small group that matches the source very closely to a much larger group with a looser match.
Meta then keeps the lookalike updated as the source changes. You use the lookalike like any other audience in an ad set, alongside location and age settings.
The seed matters more than anything else
A lookalike can only be as good as its source. Feed Meta a list of people who are not really your customers, and it will faithfully find more people who are not really your customers.
Good seed audiences
- Paying customers, uploaded from your records with consent for this use. The closer to real money, the better.
- Purchasers tracked by the Pixel, for an online shop with steady orders.
- Leads that became customers, if you mark which enquiries turned into sales and can build a list of just those.
- High-value customers, if your customer list includes a value column. Meta can weight the lookalike towards people resembling those who spend more.
Weak seed audiences
- All Page likes, especially on an older Page that has collected likes from contests or boosted posts over the years.
- All website visitors, which mixes serious buyers with people who bounced in seconds.
- Every form lead, including fake numbers and people who were never interested.
- A purchased or scraped contact list, which is against Meta’s terms and data protection rules as well as being low quality.
Quality over size is the rule. A smaller list of real customers usually produces a better lookalike than a large list of mixed contacts, as long as the source is big enough for Meta to work with. If it is too small, Meta will tell you when you try to create it.
Tracking which leads become customers is a habit worth building anyway. Our post on keeping track of which leads turn into business explains a simple way to do it.
Choosing the size
The size setting is a trade-off between closeness and reach. The smallest setting finds the people who most resemble your source; larger settings include more people who resemble them less strongly.
For a small business, a few principles help:
- Start with the closest match, because the audience will still be large in a country the size of India, the US or the UK.
- Combine it with your real service area. A lookalike across the whole country, narrowed by location to your city, gives you similar people who can actually reach you.
- Only widen the size if the close match stops delivering or becomes expensive.
Note that narrowing a national lookalike to a small local area can leave very few people in the final audience. If the ad set struggles to spend, that combination may be the reason.
When lookalikes help a small business
Lookalikes tend to earn their place when:
- You have a clean, decent-sized customer list, such as an online shop with a history of orders or a clinic with a patient database collected with consent.
- You sell nationally or across a large region, where Meta has a big pool to search and the lookalike narrows it sensibly.
- Your customers are distinct from the general population in ways that are hard to describe with interests, for example people who buy a particular niche product.
- Broad targeting has been patchy and you want to give Meta a stronger starting signal.
When they do not help much
For many local small businesses, lookalikes add less than expected:
- Tight local service areas. If you serve one part of a city, location already limits the audience heavily, and a lookalike on top may leave too few people.
- Small or messy sources. If your best list is a mix of old WhatsApp contacts and form leads you never checked, the lookalike inherits that mess.
- Accounts with good conversion tracking. If your campaign optimises for real leads or purchases and the Pixel reports them reliably, Meta is already learning who your customers are, and broad targeting often performs just as well.
Meta’s automated audience options have also changed how lookalikes behave. In some campaign setups, Meta may treat your lookalike as a suggestion and reach beyond it if it expects better results. Check your ad set settings so you know whether your audience is a firm limit or a starting point.
Alternatives with broad targeting
If lookalikes are not available or not working, these options often do a similar job:
- Broad targeting with a conversion objective. Set your location and leave interests open. Let the campaign learn from real leads or sales. Our guide to setting up Facebook ad audiences sensibly covers this approach.
- Creative that speaks to your best customers. Show the kind of customer you want, in their setting, with an offer that suits them. The ad itself steers delivery.
- Better conversion data. Improve the Pixel and, where suitable, the Conversions API, so Meta learns from real outcomes rather than clicks.
- Retargeting people who already know you. Not a replacement for finding new people, but often the best-value audience for a small business. See our post on reaching warm audiences with custom audiences.
Testing a lookalike properly
If you want to know whether a lookalike beats broad targeting for your business, test them fairly:
- Use the same objective, budget, ads and location in two ad sets: one with the lookalike, one broad.
- Run them at the same time, for long enough to get a reasonable number of results.
- Compare cost per real customer, not just cost per lead or click.
Meta also offers an experiments tool for split tests, which keeps the two audiences from overlapping. For the full campaign structure around a test like this, see the owner’s guide to Facebook and Instagram campaigns. Special ad categories such as housing, employment and credit limit the use of lookalikes, so check the rules if that applies to you. For how this sits within wider marketing for small businesses, our overview of digital marketing for small businesses covers the other channels.
Frequently asked questions
Can I create a lookalike from my competitor’s customers?
No. You can only build lookalikes from sources you own: your customer lists, your Pixel data, your Page and Instagram engagement, or your lead forms. There is no way to access another business’s customers, and any service offering their contact lists should be avoided.
Does a lookalike include my existing customers?
Meta generally builds lookalikes to find new people rather than the people already in the source. If you want to be sure existing customers do not see a first-time offer, add the source audience as an exclusion in the ad set as well.
How often should I refresh the customer list behind a lookalike?
Whenever your customer base has changed meaningfully, for example after a busy season. A lookalike built from a list uploaded long ago reflects who your customers were then. Updating the list with recent buyers, collected with consent, keeps the lookalike current.
If you are not sure whether your customer data is clean enough to build a useful lookalike, our Facebook and Instagram ads service can review it with you and test it against broad targeting.
Next in this series: Facebook Ads Metrics Explained: CPM, CTR, Cost per Result
