# How to Turn Competitor Google Ads Into a Creative Brief

Collect a small matched set of public competitor ads, inspect their messages and offers, then turn those observations into original test ideas. A creative is the ad's visible or spoken content. An offer is what the ad asks the viewer to accept, such as a trial or discount. A hypothesis is a claim you can test and reject with your own campaign data. The public library can show creative choices. It cannot tell you which ad won, how much was spent, or why people bought.

When you plan ad messages, you need a clear way to study what others do without copying them. This article shows how to run a small collection task, code the messages, and build an original test brief. It uses a hosted data collection program called an Actor, which is a tool that gathers public data. A Task is a saved input preset for that tool. You can use these steps even if you do manual research without software.

## How to Collect and Code Competitor Ads

1. Write your own product, target audience, decision, and constraints. Select known competitor accounts after [checking public advertiser identity](/en/articles/google-ads-competitor-account-map/) via the [Apify Google Ads Transparency Scraper](https://apify.com/chronometrica/google-ads-transparency-scraper) and the [Competitor Google Ads Research Task](https://apify.com/chronometrica/google-ads-transparency-scraper/examples/competitor-google-ads-research). Exact identity requires nonblank advertiserId plus creativeId, never creativeId alone. Domains can match several accounts; use verified exact advertiser ID or URL for matched competitors. Preserve each account separately.
2. Collect a small source sample using the task filters: region, platform, adFormat, optional dateFrom and dateTo, includeDeliveryDetails, and maxItems. The maxItems parameter is a shared row cap from 1 to 5,000, with a default of 10. Begin with a small sample such as two creatives with identical region, platform, format, and date filters for each comparison. Separate per-advertiser runs can enforce intended caps, but results still may differ in completeness. Check output run status, partial status, or stop states before stating coverage.
3. Keep required exact fields: advertiserId, advertiserName, creativeId, creativeUrl, format, mediaUrl, previewUrl, variations, variationCount, firstShownAt, lastShownAt, observedAt, requestedRegion, requestedPlatform, requestedFormat, status, and statusReason. Optional pointers and dates may be missing. A small current two-row acceptance checked metadata from two advertiser accounts, not rendered viewing or message taxonomy. Use only manually viewable content for coding. [Unviewable content stays unreviewed](/en/articles/ads-library-evidence/); metadata, HTTP 200 responses, and pointers do not prove viewing. The actor does not transcribe, classify, or score performance.
4. Define a small codebook: primary message, offer, stated benefit, proof shown, format, call to action, and important condition. Each label needs supporting observed words or visible material. Keep "not stated" distinct from "cannot inspect". For video, manually note the relevant spoken passage and scene; the data tool supplies no checked transcript.
5. Count distinct creative identities once per comparison group. When several variations share a message, keep the variation counts beside that grouping. A variation is a child item within a creative; count it separately from creative rows. Repeated text in different creatives is still multiple creatives but not proof of campaign scale or spend.
6. Look for choices you can test with your own product truth. Write a different message and specify which single element will change. Do not copy claims you cannot substantiate. A repeated offer may be an observed pattern; it is not proof of superior results.
7. Build a test brief with objective, audience, original message, offer, proof you can supply, changed variable, held-constant elements, result measure, and decision rule. Choose any threshold from your own baseline and sample plan; never derive it from competitor library counts. No general test-size prescription is supported here.
8. Review the test's content and evidence before launch. The final result is an original creative plan tied to source observations, with an explicit unknown about competitor performance. Actual publishing or ad launching is outside the data collection task.

## Fictional Worked Example

All companies, ads, counts, and text in this section are invented. A fictional repair-tool business studies six creative identities from two checked advertiser accounts. Every creative was manually viewable in the example:

- Creative A/C1 image: "Repair today", no price offer, demonstration photo, shop now.
- Creative A/C2 video: "Simple steps", free guide, three-step demonstration, read guide.
- Creative A/C3 image: "Repair today", no price offer, demonstration photo, shop now.
- Creative B/C4 text: "Start with the right kit", starter bundle, contents list, view kit.
- Creative B/C5 image: "Less waste", refill pack, packaging image, compare packs.
- Creative B/C6 video: "Simple steps", no price offer, demonstration scene, see steps.

| Creative ID | Format | Primary Message | Offer | Proof Shown | Call to Action | Variations |
| --- | --- | --- | --- | --- | --- | --- |
| A/C1 | Image | Repair today | None | Demonstration photo | Shop now | 2 |
| A/C2 | Video | Simple steps | Free guide | Three-step demo | Read guide | 1 |
| A/C3 | Image | Repair today | None | Demonstration photo | Shop now | 1 |
| B/C4 | Text | Start with the right kit | Starter bundle | Contents list | View kit | 1 |
| B/C5 | Image | Less waste | Refill pack | Packaging image | Compare packs | 1 |
| B/C6 | Video | Simple steps | None | Demonstration scene | See steps | 1 |

Creative A/C1 has two asset variations; the other five creatives have one each. Six creative identities therefore have seven variation observations. Preserve the denominator as six. The message groups are urgency/repair today (two), simple steps (two), kit selection (one), and less waste (one). Group shares are two out of six, two out of six, one out of six, and one out of six. State that these are for this invented sample only; they do not measure market share.

The complete original test brief looks like this:

- Objective: Help first-time owners choose the correct kit.
- Audience: The business's eligible beginner audience.
- New headline: "Choose the kit that fits your first repair."
- Offer: Existing starter kit at its real price.
- Proof: Business-owned side-by-side contents photo.
- Changed variable: Headline clarity.
- Held constant: Offer, image, destination, targeting, and test period.
- Measure: Completed kit-selection action divided by eligible measured visits.
- Decision: Use the message only if the planned own-data test supports improvement without more wrong-kit returns.

Source observation suggests clear steps are worth considering. It does not prove the new headline will work. The final artifact is the six-row coding table and the completed single-variable brief. Keep message and proof as separate coding columns.

## Final Action

Open the [Apify Google Ads Transparency Scraper](https://apify.com/chronometrica/google-ads-transparency-scraper) and the [Competitor Google Ads Research Task](https://apify.com/chronometrica/google-ads-transparency-scraper/examples/competitor-google-ads-research), check your advertiser identities, choose a small matched sample, manually view the creatives, and make one original test brief.

## Sources

- [Google Ads Transparency Policy](https://support.google.com/adspolicy/answer/13733850?hl=en-CA)
- [Apify Google Ads Transparency Scraper](https://apify.com/chronometrica/google-ads-transparency-scraper)
- [Competitor Google Ads Research Task](https://apify.com/chronometrica/google-ads-transparency-scraper/examples/competitor-google-ads-research)