The Circle on the Map Is Not Your Service Area
Radius targeting is the default in Meta Ads Manager because it is the easiest control to build. Drop a pin, drag a number, watch the estimated audience size move. It rests on one assumption: that distance from a pin is a reasonable proxy for how likely someone is to become a customer.
In a metro built on a grid, that assumption is close enough to true. Fifteen miles in any direction from a pin is roughly fifteen miles of driving in any direction, and the circle approximates a drive time well enough to spend money against.
Tampa Bay is not that metro. The bay sits in the middle of it, and the practical distance between two points here depends far more on which side of the water they are on than on how far apart they are. Radius targeting has one input — distance — for a market whose real constraint is direction. That mismatch is why so many local advertisers here see healthy reach, acceptable cost per click and leads from people who were never going to drive to them.
What a 15-Mile Radius From Downtown St. Pete Actually Buys
Draw it and look at what is inside.
East of downtown St. Petersburg you are over water within a mile, and you stay over water until the far shore around ten or eleven miles out. To the south there are a few miles of Pinellas Point before the lower bay and the approach to the Skyway. To the west there is roughly eight miles of land before the Gulf beaches, then the Gulf. Only the northern arc is continuous land, and it runs out around Largo and Belleair — short of Clearwater, well short of Dunedin, Palm Harbor and Tarpon Springs.
Something close to half of that circle is open water. Of the land it does reach, a meaningful share sits in Hillsborough County: South Tampa, the Westshore business district, the airport corridor. All of it on the other side of a bridge.
The instinct is to tighten the radius, and this is where the trap closes. Shrinking to ten miles removes some of Hillsborough — and also removes Seminole, most of Largo, and part of the beach communities, which are exactly the customers who can reach you in twenty minutes without crossing anything. Widening to twenty-five miles to pick up Clearwater and Dunedin also picks up Brandon and Riverview. No number solves it, because you are using a distance control against a direction problem.
Worse, the circle does not just include the wrong people — it often prefers them. Within a single ad set, delivery flows toward whichever impressions are cheapest to buy, and a larger, denser market across the bay frequently is.
Bridges, Not Miles: Howard Frankland, Gandy, Courtney Campbell
There are three crossings between Pinellas and Hillsborough. The Howard Frankland carries I-275 from the Gateway area into Westshore. The Gandy runs from south St. Petersburg into South Tampa. The Courtney Campbell Causeway connects Clearwater to the Rocky Point side of Tampa. To the south, the Skyway carries traffic into Manatee County.
Someone in Westshore is about fourteen miles from downtown St. Petersburg as the crow flies. As the car drives, that is a bridge, twenty-something minutes with no traffic and considerably more at peak — and when a crossing backs up there is no alternative for several miles. You sit in it, or you drive to a different bridge.
Everyone who lives here has priced that in, which is the part the radius tool cannot know. A homeowner in South Tampa looking for a dentist, a gym, a med spa, a lawn service or somewhere to eat on a weeknight is not crossing the bay for it. They will cross for something genuinely scarce — a specialist, a specific car, a venue, a supplier with no local equivalent.
That gives you a targeting rule you can actually apply: cross-bay targeting is justified by the scarcity of your offer, not by proximity on a map. Work through your service list and mark which offers survive a bridge. If none of them do, exclude Hillsborough outright and put the money into frequency on your own side of the water.
Pinellas, Hillsborough, Pasco: Three Counties, Three Ad Sets
Splitting by county is not housekeeping. It buys you three specific things.
Cost separates. Hillsborough is larger and more contested, and auction pressure there does not track what you see in Pinellas. Blended into one ad set, the average tells you nothing about either.
Creative separates. A line about free parking and no bridge is persuasive in Seminole and meaningless in Brandon. Landmarks and neighbourhood references land with residents on one side and read as noise on the other.
Reporting separates. This is the one that matters most, and the final section explains why: your ad set boundary is the only geographic reporting Meta will give you.
Pasco deserves a note of its own, because it is really two markets. West Pasco — New Port Richey, Trinity, Hudson — is reached from north Pinellas by road, no bridge involved, and behaves like an extension of the Pinellas market. Central and east Pasco — Wesley Chapel, Land O’Lakes, Zephyrhills — is reached from Tampa on I-75 and I-275 and behaves like an extension of Hillsborough. Targeting “Pasco County” as one geography puts two unrelated audiences in the same ad set and averages their results together.
One more mechanical point: give each ad set its own budget rather than letting campaign budget optimisation distribute across them. Automated allocation moves money toward the cheapest result, and cheap results in a market you cannot serve are the most expensive thing in the account.
‘Living In’ vs ‘Recently In’: The Setting That Fills Your Leads With Tourists
Underneath the location box sits a location-type control that most advertisers never open. Depending on the objective, Meta offers some combination of people living in the location, people recently in it, and people travelling in it — and the default on most objectives is the broader option that includes recent visitors.
On a peninsula with St. Pete Beach, Treasure Island, Madeira Beach and Clearwater Beach on it, that default is expensive. A device that registered in a beach ZIP code for three days belongs to someone who has since flown home. They will see the ad, some of them will engage with it, a few will fill in the form, and none of them will book a recurring service.
The division is straightforward once you make it consciously. Use living in this location for anything involving a second visit — dentists, gyms, home services, med spas, professional services, B2B. Use the visitor-inclusive settings only when the visitor genuinely is the customer: hotels, waterfront restaurants, rentals, tours, attractions.
What hides this setting is that it flatters your metrics. Visitors are broader, cheaper attention, so cost per impression falls and click-through rate rises while qualified leads sit flat. If you are reading engagement rather than booked work, the wrong setting looks like the campaign improving. And because Meta derives location from device signals, treat any of these buckets as probabilistic rather than a verified list of who was where.
Snowbird Season Changes Your Audience Size, Not Your Settings
Pinellas is the most densely populated county in Florida before you count anyone seasonal. In the cooler months the beach communities and condo corridors take on a population that stays for weeks or months, and those people are residents in every way that matters to a local business. They book dentists, join gyms, hire contractors and eat out on a Tuesday.
Three things move as a result. Your addressable audience in coastal ZIP codes grows, so the same budget delivers at a lower frequency. Auction pressure rises as seasonal advertisers come in alongside the seasonal population. And a long stay generally registers as residency to Meta’s location signal, which means the living in setting already includes most of them.
What should not move is the targeting. The common error is widening the radius in the quieter months to keep reach numbers where they were — which is precisely how a St. Petersburg campaign turns into a Hillsborough campaign without anyone deciding it should. Hold the geography, move the budget and the creative, and watch frequency rather than reach: in the thinner months the same spend against a smaller pool climbs frequency until people are seeing the same ad often enough to resent it.
The reporting consequence matters too. Cost per lead in a high-season month and a low-season month are not measurements of the same market, and comparing them as though they were will send you re-optimising a campaign that was working.
ZIP and City Stacking: The Setup That Replaces Radius
The replacement build is not complicated, and it takes about an hour.
- Write down where customers actually come from. Not where you would accept work — where the last fifty customers lived. Most owners can name the towns from memory, and the list is almost always narrower than the radius they have been buying.
- Build from named places, not a pin. Meta accepts US cities and ZIP codes and lets you paste them in bulk, so a Pinellas audience becomes an explicit list: St. Petersburg, Gulfport, Kenneth City, Pinellas Park, Seminole, Largo, the beach communities, and north into Clearwater, Dunedin and Palm Harbor if you serve them.
- Use exclusions for the ragged edges. Where a city boundary pulls in an area you cannot serve, exclude the specific ZIP codes rather than redrawing the whole audience.
- One ad set per coherent drive-time market. Usually that maps to county, with west Pasco grouped toward Pinellas.
- Set the location type in every ad set. Never inherit the default.
- Fund each ad set separately so the account cannot quietly reallocate across the bay.
- Leave it alone long enough to read. Geography changes reset your ability to compare anything.
One caveat on granularity: a city in Meta covers the incorporated area plus a surrounding zone, which is coarser than it looks on screen. Where boundaries genuinely matter, ZIP codes are tighter. Where they do not, cities are far easier to maintain — and a list you will actually keep current beats a precise one you abandon.
Reading Delivery by Geography to Find the Waste
Here is the constraint that shapes everything above. Ads Manager will break delivery down by region and by DMA. Region is state, so every impression you buy here is Florida. DMA is Tampa–St. Petersburg (Sarasota), which spans Pinellas, Hillsborough, Pasco, Polk, Manatee and Sarasota among others — it collapses the exact distinction you are trying to see into a single row.
Since the platform will not report below that, you have to build the geography into the account. One ad set per market means spend, CPM, cost per result and lead volume per market stop being a question and become columns you can sort.
Instrument the other end as well. Ask for a ZIP code on the form — one field, and it is the single most useful thing you can collect. Pass a location parameter in the URL so your CRM records which ad set produced each lead. Ask callers where they are calling from and write it down.
Then look for three patterns. Cheap impressions and no leads concentrated in one geography means you are buying reach across the bay. Leads at a reasonable cost that never convert to booked work usually means the wrong side of the water, or the wrong location-type setting filling the pipeline with visitors. And an acceptable blended cost per lead where eighty percent of the leads come from one market means you do not have a campaign that works — you have one market working and another quietly funding it.
If you would rather have this diagnosed than rebuild it yourself, our social media ads work opens with exactly this: a geographic read of where spend has actually gone, a rebuilt audience from named places, and separate ad sets for St. Petersburg and Tampa so the next report answers the question instead of averaging it away. Where paid search is running alongside it, the same geographic discipline applies on that side too, and we handle both under PPC management.
