Adjacent audiences advertising: a framework for testing topics next to the core
Part of Lookalike websites
The short answer
Adjacent audiences are readers of topics next to a campaign's core subject, reached through the sites they read, not data about them. Test a few adjacent topics, each with a stated reason, a capped budget in its own line and one metric set in advance, and move a topic into the core plan only after it passes twice.
Key takeaways
- An adjacent topic is chosen for a reason about the reader, not because it shares a word or a category with the core.
- Adjacent sites belong in their own campaign, line item or deal, because mixing them into the core list hides both results.
- Budget an adjacent test by what it takes to read the result and what the client can afford to learn from, decided before launch.
- Placement reports show where ads ran, not whether a topic worked, so the success metric has to be chosen first.
- A topic joins the core plan only after it holds up in a second flight, with the same quality checks as every core site.
What are adjacent audiences in media planning?
Definition
- Adjacent audience
- Readers of content next to a campaign's core subject, who plausibly share the core reader's needs at a different moment. In contextual planning they are reached through the topics and sites they read, not through data about the people themselves.
Every site list has a core: the topics the brief names, where the product obviously belongs. Around it sit topics the brief does not name but the same people read. A running shoe brief has a core of running publications; its adjacent topics might be physiotherapy advice, marathon travel or healthy cooking for training. Adjacent topics can bring scale and new readers, and they are also where a budget can leak into content that never had a reason to be there.
Adjacent audiences are not lookalike audiences. A lookalike audience is a set of people modeled on a customer list. An adjacent audience in this framework is a set of topics and the publishers that cover them, so it needs no personal data and can be checked page by page. It is also different from expanding a list that underdelivers: expansion fixes scale on the core, while adjacent testing asks whether a new topic belongs in the plan at all.
How do you pick adjacent topics to test?
Start from the reader, not the taxonomy. Ask what else the core reader is doing in the weeks before and after the purchase, and which publishers cover that. A category tree can then suggest neighbors: the IAB Tech Lab Content Taxonomy 3.1 (opens in a new tab) gives each category a parent ID, so siblings under one parent, such as Gardening and Landscaping under Home & Garden, are easy to list. Each route to an adjacent topic has strengths and weaknesses.
| Route | How it works | Strength | Weakness |
|---|---|---|---|
| Taxonomy sibling | Take categories that share the core category's parent | Fast, standard and easy to explain | Siblings can share a parent but not a reader, such as two unrelated sports |
| Next problem | Follow the reader to the task before or after the purchase | Rooted in a real need the product meets | Depends on the planner's judgment; needs the reason written down |
| Same reader, other interest | Use what the core publishers' readers also read about | Finds new publishers with a familiar audience | Easy to drift into topics with no link to the product |
| Same product, other use | Find content about a different use of the same product | Opens new buyers for the product | The message may not fit the new context without new creative |
Before an adjacent topic goes into a test
- The topic has a one sentence reason that links its reader to the product.
- The topic is not one of the brief's avoid topics, and is not next to one.
- At least a few publishers on the topic pass the same quality and made for advertising checks as the core.
- The creative makes sense beside the topic's articles, or new creative is planned.
- The client has seen the topic and the reason before money is spent.
How much budget should adjacent tests get?
There is no correct share that fits every plan. The budget for an adjacent test is set by two limits. The floor is what the test needs to produce a readable result on the metric chosen in advance: enough delivery on each topic to tell a real difference from noise. The ceiling is what the client can afford to spend on learning without putting the core campaign's goals at risk. If the floor is above the ceiling, test fewer topics, not each topic more thinly. Where the platform supports it, a controlled split helps: Google Ads custom experiments (opens in a new tab) let a trial take a chosen percentage of the original campaign's budget, for Search, Display, Video and Hotel campaigns.
Five steps to size an adjacent test budget
- Write the question. Name the one thing the test must answer, such as whether readers of a topic respond to the offer.
- Pick one metric. Choose the metric that answers the question and the comparison it will be judged against, such as the core line.
- Estimate the floor. Work out the delivery each topic needs to read that metric, using past campaigns where they exist.
- Set the ceiling with the client. Agree in writing how much can go to learning in this flight.
- Fix the rules before launch. Record the dates, budget per topic and the result that would count as a pass.
How do you keep adjacent sites apart from the core list?
Separation is what makes an adjacent test readable. In Google Ads, when content targeting types are combined in an ad group, Google says ads show on content that matches any (opens in a new tab) of the topics, placements, or Display or Video keywords targeted. Adding adjacent sites to the core ad group therefore blends them into one result. The same logic holds in any buying platform: one list, one line, one result.
| Layer | Core | Adjacent test | Why it matters |
|---|---|---|---|
| Site list | The approved core list | One list per adjacent topic | A topic can be judged, kept or dropped on its own |
| Campaign, line item or deal | Core budget and goals | Its own line or deal with a capped budget | Results are reported separately and the core budget stays protected |
| Naming | Brief name plus core | Brief name plus topic and test dates | Anyone reading a report knows what each line was for |
| Overlap | Sites only on the core list | Sites removed if they are already on the core list | A site counted in both places muddies both results |
| Approval | Approved with the plan | Approved as a test, with an end date | Nothing stays live by default after the test ends |
Placement reports confirm the separation held, but they are not a verdict on the topic. Google's Where ads showed report (opens in a new tab) says placements are shown only above minimum thresholds and that its performance data is generally not statistically significant. Use it to check where ads ran, and use the metric set before launch to judge the topic.
When should an adjacent topic join the core plan?
A topic joins the core when it has passed the rule set before launch, and passed it again. One good flight can be luck, a seasonal spike or one unusually strong site carrying the whole topic. Read results by topic and by site: if a single publisher produced most of the result, the finding is about that site, not the topic, and that site can be considered for the core list on its own.
| Result | Decision | Next step |
|---|---|---|
| Passed in two flights | Join the core | Move its sites onto the core list after the standard checks |
| Passed once | Test again | Repeat in the next flight with the same rule |
| Passed on one site only | Keep the site, not the topic | Consider that site for the core list on its own |
| Did not pass | Retire | Record the result so the topic is not retested without a new reason |
| Could not be read | Fix the design | Check delivery and separation before spending again |
How does PlacementGraph suggest adjacent audiences?
What does an adjacent test look like on a real brief?
Fictional example
A refillable cleaning products brand tests two adjacent topics in Ohio
An unnamed fictional maker of refillable household cleaning products plans display across Ohio, focused on Columbus, Cleveland and Cincinnati, in English. The core list covers home organization and cleaning publishers. The planner proposes three adjacent topics, each with a reason: zero waste living, on the idea that its readers already look for refills; new parents, on the idea that baby gear needs frequent cleaning; and pet care, on the idea that pet owners clean often.
The client drops pet care to keep the test small. Zero waste and new parent publishers pass the same quality checks as the core, and two sites that already sit on the core list are removed from the test lists. Each topic gets its own line with a capped budget, dates and one agreed metric: visits to the refill subscription page compared with the core line.
After the flight, zero waste passes the rule and is booked for a second test. The new parent topic does not pass as a topic, but one family lifestyle site performs well on its own and is reviewed for the core list. The decision record shows each topic's reason, result and next step.
Frequently asked questions
Can an adjacent test run on the same deal as the core list?
It is better not to. A curated deal or private marketplace deal reports results for the whole domain list it carries, so adding adjacent sites blends them with the core. Ask for a separate deal, or a separate line on the same deal where the platform reports it apart, for each adjacent topic.
How many adjacent topics should one flight test?
As few as the question needs, such as two or three. Every added topic splits the test budget further and makes each result harder to read. If the client wants more ideas covered, run them in later flights rather than thinning every topic in one.
What if an adjacent topic beats the core?
Check the design before celebrating. Confirm the two lines ran at similar times, with the same creative and comparable delivery, and that one site did not carry the whole result. If the finding holds in a second flight, it may mean the brief's core was drawn too narrowly, which is worth raising with the client.
Is adjacent testing the same as letting a platform broaden targeting?
No. Automated broadening lets the platform choose new placements or people with its own model. An adjacent test is chosen by the planner, with a written reason for each topic, its own list and a result read on its own. Both can coexist when they run in separate lines.
Sources
- Content Taxonomy 3.1 (TSV file) (opens in a new tab)IAB Tech Lab
- About average daily budgets (opens in a new tab)Google Ads Help
- About combining content targeting types (opens in a new tab)Google Ads Help
- About the Where ads showed report (opens in a new tab)Google Ads Help
- About custom experiments (opens in a new tab)Google Ads Help
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.