Meta Ads for Local Businesses: Small Budgets, Tight Radius, Real Attribution

The typical local ad budget is a few hundred a month against an audience of perhaps eighty thousand people within a sensible drive. That combination breaks most conventional structure advice, which assumes enough budget and audience for the system to learn its way to an answer.

Fix the measurement before spending anything

A substantial share of local businesses already running ads have no pixel, no conversions API, and no events configured. They are buying traffic they cannot retarget and cannot attribute. Spending more in that state simply loses money faster.

1. Install the Meta pixel on every page of the site. This is the minimum and takes minutes.
2. Configure the events that correspond to money — booking submitted, call button tapped, form completed, purchase. Page views are not a conversion.
3. Verify the domain in Business Manager, which is required for reliable event attribution.
4. Add the Conversions API where possible. Browser-only tracking loses a meaningful fraction of events to privacy controls, and for a small budget that loss materially distorts which ads look like they are working.
5. Confirm events are actually firing before launching. Test the booking flow yourself and watch it register.

Structure: fewer campaigns than you think

The instinct is to split by audience, placement and creative to see what works. On a small budget that fragmentation guarantees nothing ever exits the learning phase, and every ad set performs badly in a way that teaches you nothing.

Test creative, not structure. On local budgets creative accounts for most of the variance, and it is the cheapest variable to change.

Radius and audience

The most common local targeting error is a radius set by hope rather than by how far people will actually travel for the service. A restaurant draws from a much smaller area than a specialist clinic, and setting both to twenty-five miles wastes most of the budget.

• Set the radius to real catchment. Ask the client where existing customers come from — they usually know, and it is usually smaller than they first say.
• Use "people living in this location", not the default that includes recent visitors, unless the business genuinely serves tourists.
• Exclude existing customers on acquisition campaigns by uploading the customer list. Paying to reach people who already buy is pure waste on a small budget.
• Resist detailed interest stacking. On a small local audience, layering interests shrinks the pool below the point where delivery works properly.
• Let broad targeting do the work within a tight radius. Geography is the targeting.

Retargeting is where the margin is

For local business, retargeting is consistently the best-performing spend and consistently the most neglected. The audiences are small, which is exactly why they are cheap and why they convert.

• Site visitors in the last thirty days who did not convert — the highest-intent audience available.
• Video viewers past 50%, which for local often outnumbers site traffic and costs very little to build.
• Instagram and Facebook profile engagers, available without any site traffic at all — useful for a business with an active organic presence and a weak site.
• Past customers for reactivation offers, which in recurring categories is frequently the single highest-return campaign.

Allocate meaningfully to retargeting once there is any audience to retarget — often a quarter to a third of a small budget. The temptation is to put everything into acquisition because it feels like growth, but the people who already showed interest are cheaper to convert than new strangers.

Creative that works locally

Local ads succeed when they look local. Stock imagery and corporate polish read as a national brand and get ignored by people who are, in fact, looking for somewhere nearby.

• Show the actual premises, the actual staff, the actual work. Recognition matters — people respond to a storefront they have driven past.
• Name the place in the copy. "In [neighbourhood]" outperforms generic copy consistently.
• Lead with the offer or the outcome in the first line, since most of the audience will read nothing else.
• Use vertical video for Reels and Stories placements; a repurposed landscape ad wastes most of the screen.
• Refresh every four to six weeks. A small audience sees the same ad repeatedly, and fatigue arrives far sooner than on national campaigns.

The landing page decides the result

Plenty of local campaigns are killed for poor performance when the ads were fine and the destination was the problem. If the click lands on a slow homepage that requires three more taps to book, the spend is wasted regardless of targeting quality.

1. Send traffic to a page about the specific thing advertised, not the homepage.
2. Make the action available without scrolling on a phone — call, book, or message.
3. Keep forms to the minimum fields that let the business follow up. Every additional field costs conversions.
4. Test the full journey on a real phone on mobile data, not on desktop over office wifi.

Where the site itself is the bottleneck, that is worth naming to the client before launching — see the gaps local businesses actually have for how to frame it without sounding like an upsell.

Reporting honestly

Local attribution is genuinely imperfect. People see an ad on a phone and call from a landline, or visit in person a week later. Overstating precision destroys trust when the client compares your figures to their own sense of what happened.

• Report cost per booking or per qualified enquiry, not cost per click.
• Ask the client to log how new customers say they found them. Imperfect, but it is real-world signal the platform cannot see.
• Compare period over period against the same months last year where seasonality is strong.
• State the limits of tracking explicitly. A client who understands why the numbers are approximate stays; one who discovers it themselves does not.

For the reporting format that keeps clients engaged rather than confused, see our guide to reports clients actually read.