In short: in Search you target by what people search for; outside Search, by who they are and what they've done. The decision that changes results the most isn't which audience you pick but the mode: observation measures without restricting, targeting restricts. Your own data is fact; the rest is Google's approximation, so start with yours.
On Search you target by what people search for; outside Search you target by who they are and what they have done. Google offers a dozen audience types with names that change every couple of years. This lesson is the stable map: which signal each one comes from, where it is useful and how to use it without restricting more than you should.
Which audience types are there in Google Ads?
| Type | Signal it comes from | Example | Typical quality |
|---|---|---|---|
| Your data (data segments, formerly "remarketing") | Your site/app (the tag), your customer lists (hashed emails), your YouTube videos | 30-day visitors; abandoned baskets; customers | The highest: they already know you |
| In-market | Google detects active buying behaviour in a category | "Shopping for a new car", "Athletic footwear" | High: they are buying, maybe not from you |
| Affinity | Stable interests from browsing history | "Running enthusiasts", "Cooking lovers" | Medium-low: interest, not intent |
| Life events | Signals of life changes | Moving house, wedding, graduation, retirement | Medium; very useful in certain sectors |
| Detailed demographics | Parental status, marital status, education, homeownership | Parents of young children; homeowners | A complement, not a base |
| Custom segments | You define it: terms searched on Google, URLs visited, apps used | People who searched for «pull-out kitchen tap» or visit home-renovation sites | High if defined well: approximate intent outside Search |
| Combined | Intersection/union of the above | In-market footwear and running affinity and not customers | Very high, small |
| Similar / expansion | Google looks for people like your data | "Similar to my buyers" | Variable; today it is called "optimised targeting" or expansion (lesson 3) |
| Basic demographics | Age, gender, income (where available) | Women 25-44 | Always as a filter, never as a base |
Where can each audience type be used?
| Network | Your data | In-market / affinity | Custom | Demographics |
|---|---|---|---|---|
| Search | Yes (Observation or Targeting; bid adjustments) | Yes (Observation) | Yes | Yes |
| Display | Yes (minimum 100 active users in 30 days) | Yes | Yes | Yes |
| Video / Demand Gen | Yes (minimum 1,000 in some cases) | Yes | Yes | Yes |
| PMax | As a signal, not targeting | As a signal | As a signal | As a signal |
| Shopping | Observation / adjustments | Observation | — | Observation |
What is the difference between observation and targeting?
When you add an audience to a campaign you choose:
- Observation (formerly "bid only"): the campaign keeps showing to everybody; the audience only measures (what CTR and conversion rate that group has) and lets you apply bid adjustments (with manual bidding). It restricts nothing.
- Targeting (formerly "target and bid"): the campaign shows only to that audience. It restricts your reach.
Rule: on Search, almost always Observation (you learn who converts without losing searches); on Display and video, Targeting (without it, Display means "everybody"). With Smart Bidding, audiences in Observation feed the algorithm as a signal even if you never touch an adjustment.
Why is your own data the best audience?
Everything else is an approximation; your data are facts: who visited which page, who bought, who abandoned. They need the Google tag (or GA4 linked) with the right consent, and for customer lists, hashed emails uploaded (Customer Match). The whole next lesson is about squeezing them.
Which audiences should you start with in a new account?
- Add in Observation to your Search campaigns: your data (all visitors, buyers), 3-5 in-market segments from your sector, 3-5 affinity segments. In a month you will know which group converts.
- Create a Display remarketing campaign (your data) before any other audience.
- Once remarketing works, try one custom segment (the search terms at the core of your business) on Display or Demand Gen, with a capped budget.
- In-market and affinity as Targeting only with creative and an awareness budget.
💡 Ninja trick: the Observation table on Search is the cheapest source of audience intelligence there is: after a month, the segment "in-market: X" with double the average conversion rate tells you who to speak to on Display and Demand Gen, and which signals to give PMax. Look at it before you create any audience campaign.
What you should remember
- Outside Search you target by who they are and what they have done: your data, in-market, affinity, life events, custom, combined.
- Observation measures without restricting (Search); Targeting restricts (Display, video).
- Your data are facts; everything else, approximations. Start there.
- In PMax, audiences are signals, not targeting.