How AI picks an audience for your campaign
What AI takes into account when it builds an audience, and where to adjust it by hand so you don't pay for impressions shown to the wrong people.
AiryAds Team
Guides2 min read
When you create a campaign, AI suggests an audience for you, so there is no need to set up targeting from scratch. But the suggestion rests on what you told it about your business, so a few minutes of checking before launch pay off within the first week.
What an audience is made of
In the dashboard, a campaign audience has four parameters: age, gender, interests and cities. All four are visible on the “Audience” step, and each one can be changed before you send the campaign for moderation.
What AI takes into account
The first thing AI looks at is what you are promoting: a product, a service, a venue or a track. Next come your business description and the channels you picked for the campaign.
For a coffee shop in the city centre, it narrows the geography to one city and picks interests around coffee and breakfast. For an online store that delivers nationwide, there will be more cities and the interests will be broader.
Where to adjust it by hand
AI doesn't know what only you know. Here are three places worth a look before launch.
| Cities | If delivery works in one city only, remove the rest. Impressions outside your delivery area won't bring orders. |
| Age | An empty field means “no limit”. Set a range only when you know exactly who buys from you. |
| Interests | Keep the ones that relate directly to your product. Every adjacent interest widens the audience with people who may not need it. |
The reach forecast is calculated for the saved audience. If you changed the parameters, save them and the figure will update.
What happens after launch
While the campaign runs, AI moves budget to wherever results are cheaper. You can change the audience at any time, and the reach forecast will be recalculated.
The report shows conversions by age. If most results come from one age group, narrow the audience to it and compare the cost per result over the following week.
