Competitive Advertising Intelligence: Build a Cited Brief

Competitive advertising intelligence is the process of collecting ad evidence, identifying the advertiser and translating observed messages into testable marketing hypotheses. A useful brief shows what was observed, where, and when. It separates those facts from guesses about spend, targeting, performance or strategy. Use this guide for a repeatable research brief. For a maintained execution path, inspect the ad intelligence Play documentation before running it; the current contract determines its supported sources and outputs.

​

Start with a research question

Choose a question that ad observations can help answer: which offers a competitor presents, what proof appears on landing pages, or how messaging differs between two markets. “Tell me everything about their strategy” does not produce a bounded research job. Write the advertiser list, region, observation window and source coverage into the brief. Include known company domains and brand aliases so the researcher can distinguish the intended business from a similarly named advertiser.

​

Use source evidence at the right level

Google’s Ads Transparency Center announcement describes a searchable hub with advertiser ads, region, format and last-run information. Use the current center to inspect the evidence available for your chosen advertiser and region. Availability and fields can differ; record what the source actually exposes. A landing page adds context about the offer and conversion path. It does not prove that every visitor saw the same page or that the campaign performed well. Keep the observed page URL and access time with your notes.

​

Build an evidence sheet

Use one row per observed creative or source record. Preserve a stable source ID where available. Store the advertiser identity, source link, region, format, observed date, ad text summary, offer, landing URL and notes about missing fields. Deduplicate exact repeats before counting themes. Keep creative variants distinct when the message or offer materially differs. Document the rule so another researcher can reproduce the same counts.

​

Worked brief: repeated setup messaging

The following is a fictional teaching sample, not an observation of a real advertiser. In a live brief, replace each evidence ID with a source link and capture date before circulating a conclusion. Research question: Does the collected messaging emphasize setup effort or breadth of data? Scope: one advertiser, one region, one seven-day collection window. Finding: two of the three distinct collected creatives discuss setup effort. This is a count of this sample, not a share of impressions or budget. The repeated A1 observation does not make the finding three out of four. Unknown: campaign spend, targeting, conversions, and whether the sample covers every active ad. The creative’s wording is an advertiser claim, not independently verified product performance. Proposed test: compare the current landing-page introduction with a short walkthrough of a real setup task. Keep the audience, offer, and primary CTA constant. Choose the conversion event and evaluation window before starting; do not select a winner from whichever metric moves first. A real brief is ready for review only when A1, A2, and A3 resolve to evidence another researcher can inspect. If access is restricted, retain an authorized capture and mark the access limitation.

​

Turn observations into hypotheses

Write each recommendation as a link between evidence and a test. For example: “The collected ads emphasize implementation time; test whether a concrete setup walkthrough helps our audience evaluate the product.” That is a hypothesis, not proof that copying the competitor will improve results. Keep claims about your own product grounded in actual capabilities. Do not copy a competitor’s numerical promise into your headline. If you cite a specific ad claim, attribute it to the advertiser and retain the source rather than adopting it as fact. Choose a small number of changes that can be evaluated independently. A new landing page, offer and audience introduced together can make the result difficult to interpret.

​

Make a recurring comparison meaningful

For a later research pass, retain the same advertiser identities, regions and collection rules. Compare source records and message themes. A change in filters or source availability can look like a strategic change if it is not recorded. If a source request fails, report that source as unavailable for that run. Do not turn a collection error into “no ads found.” Preserve the distinction between first observed, last observed and source-reported run dates.

​

Use Deepline for the checked execution path

Begin with the quickstart, search the live Play catalog for ad intelligence, and inspect the selected Play before providing inputs. Ask the agent to show the supported sources, required advertiser identifiers, output schema and cost information for the current version. Run an authorized pilot and inspect whether each conclusion links back to evidence. If the available Play does not expose a field the brief requires, label that gap or implement an owned workflow from inspected contracts. Do not invent a field because the report template expects it.

​

Questions operators ask

​

Can public ad research reveal a competitor’s exact spend?

Only report spend when the specific source supplies it with a clear scope and definition. Creative counts or repeated observations are not a substitute for spend data.

​

Does a long-running ad prove it is profitable?

No. Duration is an observation, not access to the advertiser’s economics. Use it to generate a research question, not a performance claim.

​

What makes a brief actionable?

A bounded question, reproducible coverage, linked evidence, explicit unknowns and a proposed test for your own audience. Continue with account research when advertiser context needs deeper investigation.

​

Sources

Google’s official announcement and the current Ads Transparency Center were checked September 14, 2026. This guide’s evidence model and examples are editorial recommendations; no competitor performance measurements are claimed.