THE SHORT ANSWER
Key points
- The current request is the main targeting signal.
- A match can use needs and constraints, not only keywords.
- Eligibility rules narrow the available campaigns.
- A relevance threshold should allow no ad to win.
01
A simple example
A person asks, “I need a carry-on for a week in Europe. What should I look for?” The useful context includes the product category, trip length, destination, and cabin-size constraint.
A campaign for a lightweight carry-on may fit. A campaign for checked luggage, a local restaurant, or accounting software does not. Matching the whole need is more useful than matching the word “Europe.”
02
What happens before an ad appears?
- 01
Understand the request
Identify the need, constraints, and commercial intent in the current message.
- 02
Filter eligible campaigns
Apply location, category, budget, safety, and publisher rules.
- 03
Score relevance
Compare each eligible offer with the request and require a minimum match quality.
- 04
Return a placement or nothing
Show a labeled suggestion only when it improves the person's next step.
03
Why does this matter for advertisers?
A person asking an AI product what to buy or do is expressing an immediate need. That creates a chance to be useful at the moment a decision is being made.
The goal is not to place more ads. It is to earn attention by matching fewer, better moments.
COMMON QUESTIONS
What else should you know?
Is contextual targeting the same as keyword targeting?
No. Keywords can be one signal, but contextual matching considers the meaning, constraints, and goal of the current request.
Does contextual targeting require cookies?
No. It can work from the context supplied for the current request without third-party cookies.
What happens if no campaign is relevant?
No sponsored placement should appear. A no-ad result is an important part of protecting quality.
FOR ADVERTISERS