How do you audit a site list a client hands you?
Part of Lookalike websites
The short answer
Audit a client supplied site list by putting every site through the same gates as any candidate: real publisher status, an authorized seller, made-for-advertising traits, readers in the target market and fit with the brand's suitability settings. Record a verdict and a reason for each site before any of it is used.
Key takeaways
- A client's favorite sites earn no exemption: familiarity says nothing about who sells the inventory today or how the pages look on a phone.
- Each site gets one of three verdicts: keep on the plan, keep only as a content reference, or remove, each with a written reason.
- Selling path problems and content problems lead to different outcomes, because a site with good articles but no authorized seller can still describe what the client wants.
- Removals land better with evidence the client can open in a minute, such as a screenshot of the ad layout or the missing ads.txt line.
Why does a client supplied site list need an audit at all?
A client list can come from memory: sites the marketing director reads, sites a previous agency used, sites a competitor appeared on. Each is a real signal about taste, and each can be out of date. Publishers change owners, add ad slots, move to new sellers and drift in subject. A domain that was a fine magazine three years ago may now be a different business under the same name.
Google's spam policies (opens in a new tab), updated August 28, 2026, describe expired domain abuse as an expired domain name purchased and repurposed primarily to manipulate search rankings. A favorite site that lapsed and was bought by someone else can keep its name and its reputation in a client's memory while publishing something else entirely. The audit is how the agency finds out before the money does.
Definition
- Client site list audit
- A structured review of every site a client asks to include, using the same evidence the agency applies to any new candidate, ending in a recorded verdict and reason per site.
Which checks should every client supplied site pass?
The checks are the same for a client's favorite and for a stranger's blog. What changes is the record: a client list deserves a written verdict on every line, because each removal will be questioned.
| Gate | Question | Evidence to record | Typical failure |
|---|---|---|---|
| Publisher status | Is this an editorial publisher that sells ad space? | Three recent article URLs and a visible ad slot | A brand site, store, forum or a site with no advertising |
| Selling path | Is the seller the buyer will use authorized by the site? | The matching ads.txt line and the sellers.json entry for that seller ID | No ads.txt file, or the buyer's seller missing from it |
| Made-for-advertising risk | Does the site show the traits trade groups listed in 2023? | Phone and desktop screenshots of two articles, notes on refresh and traffic links | Ads crowd the content, slots reload on a timer, slideshows split each story |
| Audience in market | Do readers in the target country actually visit? | Country popularity evidence, with the month | Popular abroad and absent from the target country's data |
| Suitability | Do sampled articles sit within the brand's tolerance per category? | Five recent articles read against the client's settings | A topic the client avoids now fills a large share of the site |
| Ownership and network | Who runs the site now, and does it share an owner with others on the list? | About page, shared ad or analytics account IDs | Several listed sites are one network, or the owner changed |
The selling path gate uses two public files. The ads.txt specification (opens in a new tab) defines DIRECT as an account the publisher directly controls and RESELLER as an account the publisher has authorized another entity to control. The sellers.json specification (opens in a new tab) says the seller_id is the same ID that appears in an ads.txt file, so the two can be matched line by line. A reseller line is not a failure on its own; a missing line for the seller the buyer intends to use is.
For audience, the Chrome UX Report (opens in a new tab) infers country from users' IP addresses and publishes country tables in BigQuery, not in the CrUX APIs. A favorite site missing from that data may simply be small, so treat absence as a question, not a verdict.
How do you run the audit, step by step?
Auditing a client list in seven steps
- Normalize the list. Reduce every entry to its domain or the exact section the client means, remove duplicates and note any entry that is an app or a YouTube channel rather than a website.
- Ask why each site is there. One line per site from the client, such as the subject, the tone or the local readership, tells you what to keep if the site fails.
- Open each site as a reader. Read three recent articles on a phone and a desktop and save screenshots of the first screen.
- Match the selling files. Find the buyer's seller in the site's ads.txt and the matching seller ID in that exchange's sellers.json.
- Check audience and language. Record country popularity for the target market and the language of the sampled articles.
- Read for suitability. Compare the sampled articles with the client's category settings, not with a general idea of what is safe.
- Write the verdict. Keep, reference only or remove, with the gate that decided it and the evidence link.
What do you do when a favorite site fails a check?
The answer depends on which gate it failed. Some failures are about the business behind the site and cannot be fixed by the agency. Others are about the buying setup and can be fixed with a call. A short decision table keeps the response consistent across clients.
| Failed gate | On the plan? | As a content reference? | Next step |
|---|---|---|---|
| Not a publisher | No | Limited; a brand site describes products, not reading | Ask the client which publications cover that brand's subject |
| No authorized path for the buyer's seller | Not until fixed | Yes, if the articles are strong | Ask whether the site sells through another authorized seller or a direct deal |
| Made-for-advertising traits | No | No, because a search seeded with it can find sites with the same traits | Show the screenshots and propose a publisher with the same subject |
| No readers in the target market | No, for this market | Yes, if the subject fits | Keep it on file for a brief in the market where it is read |
| Outside the brand's suitability tolerance | No, or only a fitting section | Only the fitting section | Offer the section URL if one part of the site fits |
| Duplicate network member | One entry for the network | One entry | Keep the strongest site and note the others |
Can a site that fails the audit still be used as a seed?
Sometimes, and the table above draws the line. A seed tells a search what kind of content to look for, so it only has to be a good example of the content. A site with excellent articles that no buyer can reach through an authorized seller still describes the reading the client wants. A site whose problem is the content itself, such as rewritten articles or a layout built around ads, would teach the search to find more of the same.
Keep the two roles apart on the audited list: a column for on the plan and a column for content reference. Choosing which references to use, and how many, is covered in the seed site guide, and the full search from those references is described in the lookalike websites guide.
How does PlacementGraph check the sites a client already likes?
How do you explain removals to the client?
Lead with what was kept, then show each removal with one line of reason and one piece of evidence. A screenshot of a phone screen covered in ads is easier for a client to check than a score. Where a site was removed for a fixable reason, say what would bring it back.
What the audit summary should give the client
- A count of sites kept, kept as content references and removed.
- For every removal, the gate it failed, written in plain words.
- A link or screenshot the client can check without help.
- For fixable failures, the action that would restore the site, such as an authorized seller.
- For each removed site, a proposed replacement that covers the same subject.
- The date of the audit, since seller files and layouts change.
What does a client list audit look like on a real brief?
Fictional example
A used car dealership group in Tennessee and Kentucky
A fictional dealership group with lots in Nashville and Louisville sends its agency 18 favorite sites for a display plan: car review magazines, two regional news sites, a motorsport blog and several sites the owner remembers from a previous campaign.
Two entries are a carmaker's site and a parts store, so they come off as non-publishers. One remembered review site now runs slideshows with ads above every image and outbound sponsored links on each page; it fails the made-for-advertising review and is not kept as a reference either. A strong independent car blog has no ads.txt line for the agency's exchange, so it moves to content reference only, with a note asking the publisher about an authorized seller.
Three sites share one analytics account and collapse into one entry. A national motoring site is kept, but only its used car buying section, because its main pages focus on racing. The client receives the audited list with every verdict, the evidence links and a replacement for each removal.
Frequently asked questions
Should the audit be repeated when a client adds sites mid campaign?
Yes, for the new sites. Run each added site through the same gates before it goes live and add its verdict to the existing record. The earlier verdicts stand unless something about those sites has changed, such as a new owner or a different seller in ads.txt.
What if the client insists on keeping a site that failed?
Record the decision as the client's, with the gate it failed and the evidence they reviewed, and date it. Keeping the site on its own labeled line in the plan lets its delivery and context be watched apart from the audited sites.
Can past campaign reports replace the audit?
No. A placement report shows where ads ran and how they delivered, not whether the site still has the same owner, layout or authorized sellers today. Use old reports to explain why the client trusts a site, then run the current gates. A site that delivered well last year can fail this year's review.
Who should sign off on the audited list?
The planner who ran the audit and the client contact who supplied the list. Both signatures, with the date, make the record useful later when someone asks why a remembered site never ran.
Sources
- Spam policies for Google web search (opens in a new tab)Google Search Central, updated August 28, 2026
- ads.txt Specification Version 1.1 (opens in a new tab)IAB Tech Lab, GitHub
- sellers.json specification (opens in a new tab)IAB Tech Lab, GitHub
- Leading Trade Groups Define Made For Advertising Websites (opens in a new tab)ANA, September 26, 2023
- CrUX methodology: dimensions (opens in a new tab)Google Chrome for Developers
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.