Demographic Targeting

Demographic Targeting applies at the ad group or campaign level across age brackets, gender, parental status, and, where the data is available, household income, using profile data Google has either explicitly collected from users (declared) or modeled statistically (inferred) for signed-in accounts. Every ad group carries an implicit Unknown bucket alongside the named brackets, representing the share of impressions Google can't confidently classify into any of them.

For each bracket, an advertiser has three options: leave it in at the default rate, exclude it entirely, or apply a bid adjustment - raising or lowering competitiveness for that segment without removing it. In practice, bid adjustments are the more common tool once an account has enough conversion history to show that some brackets convert measurably better or worse than others, rather than guessing which to cut before any data exists. These controls stack with every other targeting layer already applied to the ad group - keywords, audiences, placements - so each additional demographic restriction narrows delivery further on top of whatever narrowing already exists.

Demographic Targeting earns its place mainly as a subtractive control, used once conversion data shows a bracket is unprofitable, or where a business legitimately cannot serve a segment - an age-restricted product, or a B2B service with no relevance to certain income tiers. It matters less as a primary way to find intent, because demographic data alone says little about purchase readiness: two people in the same age and income bracket can be at completely different points in a buying decision, and demographic controls have no way to distinguish between them.

The most common mistake is excluding the Unknown bucket alongside a genuinely underperforming bracket, on the assumption that unclassified traffic is inherently low quality. Unknown is frequently a meaningful share of total impressions and sometimes converts perfectly well, so cutting it removes real, sometimes profitable traffic that simply wasn't identifiable rather than traffic that was actually bad. A second mistake is assuming these controls are available on every campaign - housing, employment, and credit-related categories are blocked from demographic targeting entirely under Google's restricted-category advertising policies, specifically because narrowing delivery by age, gender, or income in those verticals raises fair-lending and equal-opportunity concerns that the policy is designed to prevent. A third is applying a bid-down when the actual goal was to remove a bracket completely - a modest bid reduction still lets that segment see the ad, just at reduced competitiveness, which isn't the same outcome as an exclusion.

In reporting, the Demographics report broken out by age and gender inside the platform shows conversion rate and cost per conversion by bracket, and that's the number to check before making any exclusion decision - not impression share, which says nothing about whether a bracket is actually profitable. Because Unknown often carries a meaningful share of conversions despite offering no usable targeting signal on its own, exclusion decisions should be based on sustained underperformance across enough volume to be statistically reliable, not on a few days of noisy, low-sample data in a single bracket.