How to build a contextual topic list a reviewer can test
Part of Contextual targeting
The short answer
A contextual topic list is the set of subjects a campaign wants its ads beside, each written as a description of an ideal article with a rule for what does not count. Build it by finding the reader situations a product belongs to, writing each as a testable description, splitting broad topics, and piloting the list on real sample articles.
Key takeaways
- Topics come from the situations where readers need the product, not from the product's own feature list.
- A testable topic has three parts: a short label, a one sentence description and a narrative that says what counts and what does not.
- Every topic on the list costs review time, so carry only topics that would change which sites are kept.
- Split a broad topic by reader task, depth or format until two reviewers can agree on any article.
- Pilot the list on a small set of real articles before the search, and rewrite any topic the reviewers disagree on.
What is a contextual topic list, and what makes it different from keywords?
Definition
- Contextual topic list
- The subjects a campaign wants its ads beside, each written as a description of the content that qualifies, so that planners, reviewers and matching tools can apply it the same way.
Google's help page on contextual targeting (opens in a new tab) draws the line in one sentence: "Keywords are individual words, whereas a topic is a concept for webpage or video." A keyword list asks whether certain words appear. A topic list asks whether the article is about something, which is a judgment a person can make and defend. Our guide to contextual keyword lists covers the word level; this guide covers the concept level that sits above it.
Information retrieval research met a version of this problem long ago. The Text REtrieval Conference, run by NIST, calls a natural language statement of information need a "topic" (opens in a new tab) to distinguish it from a "query", the data structure actually presented to the retrieval system. Its topics are written for human assessors who judge whether documents are relevant, which is the job a media reviewer does with candidate articles.
How do you turn a product into topics to include?
A client brief can describe the product, and products are not topics. A pool chemical is not something people read about for its own sake; they read about a cloudy pool the week before a party. The work is to find the reading situations where the product is genuinely useful, then describe the articles those situations produce.
| Lens | Question to ask the client | Kind of article it points to |
|---|---|---|
| Problem | What goes wrong that this product fixes? | Troubleshooting guides and fix it explainers |
| Task | What job is the reader doing when they need it? | Step by step how to articles and checklists |
| Decision | What choice comes right before buying? | Buying guides, explainers of options, cost breakdowns |
| Ownership | What does owning or using it involve over time? | Maintenance schedules, care guides, long term reviews |
| Occasion | When in the year or in life does the need appear? | Seasonal preparation stories and life event guides |
| Interest | What wider hobby or interest does it belong to? | Enthusiast features, community news, project stories |
From product to a first topic draft
- Run the six lenses with the client. Write down every answer, even ones that seem obvious, in the client's own words.
- Turn each answer into a reader situation. A situation names a person and a moment, such as a homeowner opening a pool for the first warm weekend.
- Cross out situations no publisher writes about. If a planner cannot picture a real article, the situation will not find inventory.
- Write one draft topic per surviving situation. Keep the client's wording in a note so the client recognizes their brief in the result.
What makes a topic description testable by a reviewer?
A topic is testable when two reviewers reading the same article reach the same verdict without discussing it. TREC topics achieve this with three fields: a short title, a one sentence description of the need, and a narrative that spells out what is relevant and what is not. The narrative is the part a media brief can leave out.
A relevant document will focus on the causes of the lack of integration in a significant way; that is, the mere mention of immigration difficulties is not relevant.
| Part | What it contains | The test a reviewer applies |
|---|---|---|
| Label | Two to six words used in reports and conversation | None; the label is a name, never the rule |
| Description | One sentence describing the ideal article and its reader | Is this article mainly about what the sentence describes? |
| Narrative: counts | The formats, depths and angles that qualify | Does the article take one of these forms? |
| Narrative: does not count | Passing mentions, wrong reader, wrong treatment | Is the subject only mentioned, or treated for another purpose? |
How many topics should one brief carry?
There is no industry standard number, and the right count depends on review capacity more than ambition. Every topic must be piloted, searched, reviewed and reported on, so each one should earn its place by changing which sites end up on the list. Two topics that would select the same publishers are one topic with two labels.
| Signal | Too few topics | Too many topics |
|---|---|---|
| Candidate sites | The same handful of large sites keep matching | Topics find only one or two sites each |
| Reviewer experience | Reviewers stretch one topic to cover unlike articles | Reviewers cannot remember which topic an article belongs to |
| Reporting | Results cannot be explained beyond one theme | Spend is too thin per topic to read anything |
| Client reaction | The client asks why obvious angles are missing | The client cannot say which topics matter most |
Rank the topics once the count settles. Mark one or two as core and the rest as supporting, so that if the list must be short, it fills from the core first.
How do you split a broad topic into narrower ones?
Broad topics feel safe because they match a lot. In Google Ads topic targeting (opens in a new tab), selecting a top-level topic also targets all of its subtopics, and the IAB Tech Lab Content Taxonomy (opens in a new tab) arranges categories in tiers for the same reason: a parent category covers a great deal of unlike content. A topic on a brief should be split until reviewers can agree on any article in it.
Four rules for a clean split
- Cut on one facet at a time. Splitting by task and by format in the same step produces overlapping topics.
- Keep the pieces apart. If one article fits two child topics, the cut was on the wrong facet.
- Drop pieces with no publishers. A child topic that no site writes about is a gap in the market, not a target.
- Stop early when the reviewers already agree. Narrower is not better once the topic is testable.
How do you check a topic list against sample articles?
Run a pilot before the search. Collect a small set of real articles per topic: a few the planner expects to match, a few that sit on the edge, and a few that only mention the subject. Ask two reviewers to label each article against the topic descriptions without talking to each other, then compare.
Topic list pilot check
- Each topic was tested on expected matches, edge cases and passing mentions, all from real publishers.
- Two reviewers labeled the articles separately before comparing verdicts.
- Every disagreement was traced to a missing or vague line in the narrative, and that line was rewritten.
- No article was claimed by two topics; overlapping topics were merged or recut.
- Every topic found at least one plausible publisher in the pilot.
- The client saw the final descriptions, with the narratives, before the search began.
Once the list is live, the same descriptions become the reference for article level checks; our guide on how to verify contextual relevance sets out that review, with a scoring sheet and tie breaks.
What does a finished topic list look like?
Fictional example
A pool supply retailer in Georgia
A fictional pool and hot tub supply retailer with stores around Atlanta and Savannah sends a brief that lists its products: chemicals, pumps, filters and covers. The planner runs the six lenses with the client and finds four reading situations: opening a pool for the season, fixing cloudy or green water, choosing a replacement pump, and caring for a hot tub through the year.
The cloudy water topic is written as a description, a guide that helps a homeowner diagnose and fix cloudy or green pool water, and a narrative. It counts troubleshooting guides and water chemistry explainers for homeowners. It does not count news about public pool closures, articles for commercial pool operators, or listicles that mention cloudy water once among summer tips.
In the pilot, the reviewers split on a long summer entertaining feature with one paragraph on water clarity. The planner adds a line to the narrative, the article's main subject must be pool care, and the second pilot agrees.
How does PlacementGraph use a topic list?
Frequently asked questions
Should a topic list name the client's products or brands?
Keep product and brand names out of the topic descriptions. A topic built on a product name matches only pages that name it, which narrows inventory and leans toward review sites. Product names belong in the creative and, if needed, in a separate keyword list.
Can a topic list reuse IAB Content Taxonomy category names?
Use taxonomy categories as a cross reference, not as the topics themselves. A tier 1 category such as Home and Garden is a useful tag for reporting, but it is far too broad to test an article against. Write the description and narrative first, then note the nearest category beside it.
Should topics include seasonal or event based angles?
Yes, when the client's sales follow the calendar. Write the occasion into the description, such as preparing a pool for the first warm weekend, and give the topic a start and end date in the brief. Seasonal topics should be reviewed against recent articles, because older seasonal content can sit on a site for years.
What if the client insists on a topic the pilot shows has no publishers?
Show the client the pilot result and keep the topic as a documented gap rather than stretching it to fit unrelated sites. A topic with no publishers may point to a content partnership or a different channel, such as video, instead of a site list.
Sources
- English test questions (topics) (opens in a new tab)Text REtrieval Conference (TREC), NIST
- TREC ad hoc topics 401 to 450 (opens in a new tab)Text REtrieval Conference (TREC), NIST
- About contextual targeting (opens in a new tab)Google Ads Help
- About topic targeting (opens in a new tab)Google Ads Help
- Content Taxonomy (opens in a new tab)IAB Tech Lab
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.