Geo-Targeting
Geo-targeting is the practice of delivering ads, content, or experiences to users based on their detected or declared location, typically configured within ad platforms.
Key takeaways
- Signals rank by reliability: device GPS, then IP lookup, then browser language and time zone.
- IP accuracy is strong at country level and weak at city or street level.
- Presence and interest targeting are different settings, and defaults often include both.
- Treat a detected location as an overridable default, not a fixed fact.
- Wrong exclusions are invisible, since the excluded buyer appears in no report.
In depth
Delivery hinges on a location signal resolved at request time. The strongest is device GPS, available only in apps or with browser permission. The most common is the IP address, mapped through a geolocation database to a country, region and often a city. Weaker but useful signals include the language and time zone the browser reports, a postal code the user typed, and a shipping address on file. The platform compares whatever it resolved against your include and exclude lists and decides whether to enter the auction.
Accuracy falls as the radius shrinks. IP lookup is dependable at country level and unreliable at street level, because mobile carriers route traffic through regional gateways, corporate VPNs terminate somewhere other than the user, and residential ranges are reassigned. Most platforms also distinguish people physically present in a place from people showing interest in it, and the default often includes both. That setting quietly changes who sees the ad, and a campaign meant for local buyers can spend on distant researchers.
The same signal drives on-site behaviour once the click lands. A page can preselect a currency, show the nearest office, or swap a case study for a local one before the visitor does anything. The safe pattern is to treat the detected value as a default that the visitor can override, and a quiz makes that explicit: a location question presented with the detected answer preselected confirms or corrects the reading, and the confirmed value, not the guess, feeds routing and pricing.
Location signals describe a device at a moment, not a buying decision. A laptop connecting through a corporate VPN reports the office, a phone on a train reports the wrong city, and a traveller reports a market they will never buy in. Exclusion carries the greater risk, because a wrongly excluded buyer never appears in any report and the loss is invisible. Where the stakes are high, prefer a declared location over a detected one and keep exclusions coarse.
Example in practice
How to measure it
Compare the location the platform inferred with the location the visitor confirmed. Take completed forms where both values exist and calculate the share that agree; a low agreement rate in one channel points at that channel's signal quality rather than at the audience. Break the rate down by device, because mobile traffic usually shows the widest gap between reported and confirmed location.
Read performance by geography with volume attached. Conversion rate per region is meaningless below a few dozen sessions, so aggregate small areas before drawing conclusions. For radius campaigns, plot conversion against distance from the centre to find where relevance actually fades, then set the radius at that point instead of the round number someone picked at setup.
Common mistakes
The frequent error is leaving the presence-versus-interest setting at its default and then blaming the audience. A campaign for a single metro quietly serves anyone reading about that metro from anywhere, which inflates impressions and dilutes conversion. Open the location options, choose presence explicitly when you sell locally, and check the setting again after any campaign duplication, because copied campaigns often reset it.
The second error is hard-locking the page to the detected country with no way out. A visitor whose IP resolves to the wrong market sees prices in a currency they cannot pay in and no control to change it, so they leave. Always render a visible selector alongside the detected default. Teams also forget that a location captured at click time can differ from the buying region, so store both rather than overwriting one with the other.
Frequently asked questions
How accurate is IP-based geo-targeting?
Country level is generally dependable; city level is a reasonable guess; anything finer should not be trusted for a decision that matters. Mobile carriers, corporate VPNs and reassigned address ranges all push the reading away from the user. Confirm the location in the funnel whenever routing, pricing or eligibility depends on it, and keep the detected value only as a default.
What is the difference between presence and interest targeting?
Presence targets people physically in the location. Interest targets people showing interest in it, including those searching about it from elsewhere. Interest widens reach and suits travel or relocation offers; presence suits local service businesses. The default in several platforms includes both, so a local campaign can spend heavily on distant researchers unless you change it.
Can I exclude locations as well as include them?
Yes, and exclusions take precedence over inclusions where they overlap. Use them to remove regions you cannot serve or markets that consistently produce unqualified traffic. Keep exclusions coarse, because a wrongly excluded buyer never enters any report and the loss is invisible. Review the exclusion list on a schedule, since territories and coverage change faster than campaign settings do.
Should I set different bids by location?
Once you have enough conversions per region to trust the difference, yes. Bid adjustments let one campaign pay more where value per lead is higher rather than splitting into separate campaigns and fragmenting learning. Below roughly a few dozen conversions in a region, the observed difference is usually noise and adjusting on it will make performance worse.
Does geo-targeting also personalise the on-site experience?
It can, using the same signal the ad platform used. Common uses are preselecting currency, showing the nearest office, and swapping proof for a local example. Keep the change reversible with a visible selector, and avoid hiding content based on detected location, because a wrong reading then removes information the visitor needed rather than merely reordering it.
How does geo-targeting interact with a quiz funnel?
The platform setting decides who reaches the quiz; a location question inside it decides what happens next. Presenting the detected value as a preselected answer lets the respondent confirm or correct it in one tap, which repairs bad IP data at the point it matters and gives routing a declared value rather than an inference to work from.