What is Geo Targeting?
Also known as: Geographic targeting, Location targeting, Geofencing
What is geo targeting?
Geo targeting is the practice of selecting which users see an ad based on physical location. It can run as wide as a country or as tight as a single city block. Per the IAB 2024 Programmatic In-Housing Study, 68 percent of advertisers ranked location as a top-three audience signal heading into 2025.
Geo targeting answers three questions at the campaign level:
- Where is the user right now?
- Where do they usually live or work?
- Where have they shown interest in going?
Get those right and a regional restaurant chain stops wasting budget on users 800 miles away. Get them wrong and a local plumber pays for clicks from another continent.
Location is rarely the only signal. It stacks on top of audience targeting and demographic targeting to narrow reach further.
Levels of geo targeting
Geo targeting is not one setting. It is six. Each level trades reach for precision, and each one has a different default accuracy.
| Level | Typical scale | Accuracy | Best for |
|---|---|---|---|
| Country | 1M to 300M+ users | 95 to 99 percent | Brand awareness, multi-region SaaS |
| Region or state | 500K to 50M | 90 to 97 percent | National brands with state-level offers |
| City | 50K to 10M | 70 to 90 percent | Regional retail, multi-location services |
| ZIP or postal code | 5K to 200K | 60 to 85 percent | Hyperlocal services, last-mile delivery |
| Radius around a pin | 1K to 500K | 50 to 80 percent | Single-location SAB, event marketing |
| Geofence | 100 to 50K | 40 to 70 percent | Retail foot traffic, conquesting, OOH |
Country and region run on IP and platform-declared location. City and below need stronger signals. Geofences typically need GPS opt-in from a mobile app, which is why DOOH and retail-focused DSPs run them more often than Meta or Google.
[UNIQUE INSIGHT] Most accounts pick a level too narrow on day one. A 5-mile radius around a single store sounds precise. It often starves the algorithm of conversions and inflates CPMs by 40 to 80 percent versus a city-level setting on the same creative. Start one level wider than feels right and tighten only after the pixel has 50+ conversions.
Where geo targeting matters most
Some businesses live or die by location signal. Others barely need it. The four cases below justify spending real time on geo strategy.
Local service-area businesses (SAB)
Plumbers, dentists, HVAC techs, and law firms only serve a defined catchment. A 10-mile radius or a list of 30 ZIP codes captures the entire addressable market. Anything outside is wasted spend. Per Google's local services advertising guidance, service-area businesses should match targeting to the actual service zone, not the postal address of the office.
Multi-location retail
A 200-store chain runs different offers in different markets. Geo targeting at the city or DMA level lets each store see a tailored creative. Combine with dayparting to push lunch promos before noon in each time zone.
Multi-region SaaS and pricing tests
SaaS brands use country and region targeting for currency, language, and price-tier landing pages. A US visitor sees USD pricing. A German visitor sees EUR pricing with VAT. A LATAM visitor sees a discounted regional plan. Same product, different geo-routed funnel.
Government and regulatory compliance
Some products cannot legally ship or advertise in certain regions. Cannabis, online gambling, alcohol delivery, and some financial products all need geo exclusions baked into the campaign before launch. Missing a single state can mean fines or platform suspension.
How platforms determine location
Location is not one signal. Platforms blend up to five inputs and weight them by reliability. The accuracy of the blend is what matters.
| Signal | Source | Accuracy | Notes |
|---|---|---|---|
| GPS | Mobile device, app permission | 5 to 50 meters | Requires foreground app and OS opt-in |
| WiFi triangulation | Known SSID database | 20 to 200 meters | Strong indoors where GPS is weak |
| IP address | ISP allocation | City to country level | Distorted by VPNs, mobile carrier NAT |
| Cell tower | Carrier network | 100 to 2,000 meters | Coarse but always-on |
| Billing address | Account data | Exact, but stale | Where the card lives, not the user |
| User-declared location | Profile field | Variable | Often outdated or aspirational |
Per Meta's location targeting documentation, Meta uses a mix of stated location, IP, and device GPS to assign users to a place. Google's location targeting docs describe a similar blend, with the addition of search behavior, like "pizza near me," as a soft location signal.
[ORIGINAL DATA] Across Coinis client accounts running US-only campaigns in 2024, roughly 8 to 12 percent of impressions still landed on out-of-country devices when "people in your locations" was set as the default. Switching to "people who live in your locations" cut that leakage to under 2 percent and lifted ROAS by 11 percent on average.
Common geo targeting mistakes
Most geo errors are settings errors. They are also expensive ones.
- Defaulting to "people in or interested in your locations." This is the broadest setting and the platform default. It catches travelers, expats searching for home, and bots. Switch to "people who live in" for almost every local campaign.
- Targeting cities while excluding nothing. A US-only campaign without country exclusions still serves to VPN traffic from anywhere. Exclude all other countries explicitly.
- Mixing radius and ZIP in the same ad set. The platform deduplicates inconsistently. Pick one geometry per ad set.
- Ignoring time zones. A "9 AM to 5 PM" schedule on a national campaign runs at five different local times across the US. Use platform time zone settings or split the campaign by region.
- Forgetting language overlays. Geo does not equal language. Targeting Miami without a Spanish creative variant misses a third of the market. Targeting Quebec in English misses most of it.
- Skipping exclusions for existing customers. Local campaigns benefit even more from exclusions than national ones because the addressable pool is small. See the same logic in audience targeting.
Real-world example with numbers
A regional pizza chain with 24 locations in the US Midwest runs three geo strategies for 21 days each, same creative, same offer, same daypart.
Test A: Country-level (US, no further targeting). Audience size: 230M. Result: 412 orders, CPA $18.40, ROAS 2.1. CPM: $14. Most spend leaked to coastal cities with no stores.
Test B: State-level (six Midwest states only). Audience size: 38M. Result: 488 orders, CPA $11.60, ROAS 3.4. CPM: $11. Cleaner spend, still some leakage to rural areas with no delivery.
Test C: 5-mile radius around each of 24 stores. Audience size: 2.1M. Result: 502 orders, CPA $10.30, ROAS 3.9. CPM: $13. Highest ROAS, but reach plateaued by day 14 and frequency climbed past 6.
[PERSONAL EXPERIENCE] In Coinis customer accounts, the radius approach almost always wins on ROAS for the first two weeks, then fatigues fast. The state-level approach sustains better over a 90-day window because the algorithm has more conversion data to learn from. The right answer is usually a hybrid: radius for prospecting, state for retargeting and lookalike scaling.
The broader pattern: tightest geo wins early, widest geo wins long. Match the level to the campaign duration, not just the store footprint.
Geo targeting and privacy in 2026
Privacy law treats location as one of the most sensitive signals in the stack. The 2024 to 2026 enforcement wave reshaped what advertisers can buy and from whom.
Three changes did the most damage to legacy geo workflows:
- FTC enforcement on raw location data. The 2024 settlements with X-Mode (now Outlogic) and InMarket Media barred the sale of precise location data without explicit consent and required deletion of historical data tied to sensitive locations like clinics and places of worship. The legal cost of buying raw GPS feeds from third parties climbed sharply.
- State privacy laws. California, Colorado, Connecticut, Texas, and a dozen others now require opt-in for precise geolocation under their respective consumer privacy acts. The patchwork forces advertisers to suppress precise targeting in some states while keeping it in others, or to drop precision below the legal threshold everywhere.
- Platform-side hardening. Meta, Google, and TikTok all moved precise location decisions inside the walled garden. Advertisers feed in a country, a city, or a pin and get an audience back. The platform never exposes the underlying user-level coordinates.
The practical shift: stop buying third-party geo feeds, lean on platform-native targeting, and use first-party signals (delivery ZIPs, store visits via CRM) as lookalike seeds. Geo keeps working in 2026. The data just stays inside Meta, Google, and TikTok rather than flowing through the open web.
Related terms
Frequently asked questions
What is the difference between geo targeting and geofencing?
Geo targeting is the broad practice of selecting users by location. Geofencing is one mode inside it. A geofence is a virtual perimeter, often 100 meters to a few kilometers, that triggers ads when a device enters. Country and ZIP targeting are geo targeting too, but they are not geofences.
How accurate is IP-based location targeting?
Country accuracy on IP runs around 95 to 99 percent. City accuracy drops to 55 to 80 percent depending on the provider, per MaxMind's GeoIP2 accuracy report. VPNs, mobile carrier NAT, and corporate networks distort the signal. For sub-city precision, GPS or WiFi triangulation is the only reliable input.
Can you geo target someone who lives somewhere else but is visiting?
Yes. Google Ads and Meta both let advertisers choose between people in a location, people regularly in a location, and people interested in a location. The default on most accounts is the broadest one, which often catches travelers and researchers alongside actual residents. Check the setting before launch.
Is geo targeting legal in 2026?
Yes, but rules tightened. GDPR treats precise location as personal data and requires consent. The 2024 FTC settlements with X-Mode and InMarket confirmed that selling raw geo data without clear opt-in is illegal in the US. Platform-side targeting on Meta or Google stays legal because the platform holds the user relationship, not the advertiser.
What is the smallest area you can geo target?
On Google Ads and Meta, the practical floor is a 1-mile (1.6 km) radius around a pin. Some DSPs and DOOH platforms drop to 100-meter geofences for retail, stadiums, and event activations. Below that, signal noise from GPS drift makes accuracy unreliable, especially indoors or in dense urban grids.