How do you run a website ad layout audit on phone and desktop?
Part of MFA avoidance
The short answer
A website ad layout audit is a timed, recorded visit to a fixed sample of a site's pages on a phone and a desktop. The auditor counts slots per screen, logs sticky units, interstitials and autoplay video, watches for refresh with the browser's network log, and writes each observation on a sheet with the URL, device, time and a screenshot.
Key takeaways
- Plan the sample before the visit: page types, devices and repeat visits, written down so another person can repeat the audit.
- Count by screen, not by page, because what a reader meets at each scroll depth is what the audit is meant to capture.
- Use the browser's network log with Preserve log on to catch refresh that a glance at the screen would miss.
- Classify formats with the Coalition for Better Ads names, so findings from different auditors line up.
- Confirm phone findings on a real handset, since Chrome's own documentation calls device emulation a first-order approximation.
How do you audit a website's ad layout?
An ad layout audit answers one narrow question: what does a reader of this site meet, screen by screen, while reading? It is narrower than a full made-for-advertising review, which also covers traffic, content and the selling path, and it produces evidence the other checks can lean on. The signal checklist for identifying MFA sites says which signals matter; this guide covers how to collect layout evidence in a repeatable way.
Prepare the browser before the first page
- Use a fresh browser profile with no ad blocker. Extensions that hide ads or change pages would make the audit describe a page no reader sees.
- Open the Network panel and turn on Preserve log. Chrome's Network reference (opens in a new tab) says this keeps requests across page loads, so a refresh after navigation stays on record.
- Turn on screenshots in the Network panel settings. Chrome describes this as a way to see what users see while a page loads, which captures units that appear and then move.
- Write the start time and the device on the sheet. A clock visible in screen recordings makes later timing checks easy.
- Arrive the way a reader would. Open the article from the site's own section page or a search result, not from a bookmarked deep link, and note which route you used.
Which ad formats should the audit classify?
Shared names keep audits comparable. The Coalition for Better Ads standards (opens in a new tab) name six desktop experiences and ten mobile ones that fall beneath what the Coalition calls a threshold of consumer acceptability. Google's Ad Experience Report help pages define each one, and the definitions give the auditor a test to apply on the screen.
| Format | Device | The test Google's definition gives |
|---|---|---|
| Pop-up ad | Phone and desktop | Appears on top of the main page after it loads and blocks the content |
| Prestitial ad | Phone; desktop only with a countdown | Appears before the page content loads |
| Postitial ad with countdown | Phone | Appears after the reader follows a link and forces a wait before it can be dismissed |
| Large sticky ad | Phone and desktop | Sticks to the page regardless of scrolling and takes up more than 30 percent of the screen |
| Full screen scrollover | Phone | Forces the reader to scroll through an ad at a different speed than the content |
| Autoplay video with sound | Phone and desktop | Plays with sound without any user interaction |
| Sticky pop-out video | Phone | Starts as an inline autoplay video, then moves to the bottom of the viewport once the inline player is out of view |
What should be counted on a phone screen?
Count by screen, from the top of the article to its last paragraph. A screen is one viewport of scroll. For each screen, write down how many ad units are fully or partly visible, whether one of them is sticky, and whether any article text is visible at all. Screens with no article text are the finding readers remember, and they are easy to lose in a page level total.
Density has its own method, measured against the main content area, which the ad density guide covers. A layout audit records the count per screen and the screen numbers where text disappears, then leaves the percentage to that method. On a phone, Google's mobile definitions (opens in a new tab) also flag a large inline unit of 300x600 or greater combined with sticky video, so note unit sizes where the page source or the creative shows them.
Per screen counts on a phone
- Screen number, counted from the top of the article.
- Ad units visible on the screen, including partial units.
- Sticky unit present, with its position (top or bottom) and whether it has a close control.
- Article text visible on the screen: yes or no.
- Units labeled as sponsored content or recommendations, counted apart from display units.
- Video players on the screen, and whether they started without a tap.
Chrome's device mode documentation (opens in a new tab) calls emulation a first-order approximation of a phone and says that when in doubt, run the page on a mobile device. Use emulation to plan the audit and a real handset for the counts that end up on the sheet.
How do you detect ad refresh during an audit?
Refresh is invisible when two creatives look alike, so the network log is the better witness. Google Ad Manager's refresh documentation (opens in a new tab) describes three triggers: user action refreshes based on user-guided navigation, event-driven changes initiated by the publisher, and time interval refreshes at set intervals, with a 30 second minimum for event and time based refresh on desktop and mobile web. A new ad request for the same slot with no scroll or click in between points to a timed trigger.
A refresh check with the network log
- Load the article with Preserve log on and note the time. Let the page settle for ten seconds.
- Filter the request list for the ad server's requests. Chrome's filter matches text in the request URL, domain or hostname; type the ad server domain you see in the first requests.
- Leave the page untouched for two minutes. Write down the time of each new ad request and which slot changed on screen, if any.
- Scroll once, then wait again. New requests right after the scroll fit a user action trigger; requests at even spacing without input fit a timer.
- Export the log. Chrome's default HAR export is sanitized and omits cookie and authorization headers, which keeps the file safe to share with a seller.
The ad refresh guide covers what refresh means for viewability and how a seller should declare it. For the audit, the job is narrower: a dated log that shows requests, times and the input that came before them.
How many pages should an ad layout audit cover?
No standard sets a number. Google's Ad Experience Report help (opens in a new tab) says its own reviews are not exhaustive, look at a sample of pages, and cover the desktop and mobile versions of a site separately; it does not say how many pages it samples. A buyer's audit needs a sample that covers the page types a campaign would buy and is written down before the visit, so the result cannot be steered.
| Page type | Why it is in the sample | Visits |
|---|---|---|
| Two recent articles from different sections | Articles carry the slots a plan buys | Phone and desktop, once each |
| One older article from the archive | Older templates can carry heavier ad loads | Phone and desktop, once each |
| One gallery or list page, if the site has them | Multi page formats change the count per visit | Phone, once |
| A repeat of the first article | Shows whether interstitials or pop-ups depend on the visit | Phone, a second visit after navigating within the site |
That sample produces eight or nine observations per site. If one observation shows a format from the table above, add two more articles of the same type before writing a finding, so a single unusual page does not decide the result.
How should audit evidence be recorded?
One row per observation, one observation per page and device. A sheet built this way can be sorted by site, filtered by format and handed to a seller without rewriting.
| Column | What to enter |
|---|---|
| Observation ID | Site, page type, device and visit number, for example site 4, article A, phone, visit 1 |
| URL and time | The full address and the local start time |
| Device and viewport | Handset model or desktop window size, and the browser version |
| Screens and units | Number of screens in the article and the per screen unit counts |
| Text free screens | Screen numbers with no article text visible |
| Formats observed | Format names from the classification table, with the screen number |
| Refresh | New ad requests with times and the input before them, or none seen |
| Evidence files | Screenshot names and the HAR file name |
| Benign reading | Any innocent explanation considered, such as a live page updating |
What does an ad layout audit look like on a real brief?
Fictional example
A flooring brief in Maine and New Hampshire
A fictional Portland, Maine retailer of hardwood flooring asks its agency for a display plan in Maine and New Hampshire, beside home renovation guides, design features and local home and garden coverage. Twelve sites pass the content review and go to a layout audit with the sample above.
A Manchester home design blog shows a sticky bottom unit with a close control and two inline units per article, with text on each screen. It goes on the list. A renovation tips site shows a postitial with a countdown after the reader taps to the second article on a phone, on two of three repeat visits, and screens six to nine of its gallery carry no text. Both observations go on the sheet with screenshots.
The network log on a Bangor lifestyle site shows new ad requests at even spacing with no input. The planner sends the sanitized HAR file to the seller and asks which slots are declared as refreshing before deciding, instead of excluding the site from one log.
How does PlacementGraph check ad layout on candidate sites?
Frequently asked questions
Should the audit be done logged in or logged out?
Logged out, unless the plan buys a subscriber only section. A signed in reader's page can differ from what an anonymous visitor gets, and a campaign on open web inventory is bought against the anonymous view. Note the sign in state on the sheet so a later audit can match it.
Can screenshots alone prove an ad layout problem?
They show what was on the screen at one moment, which settles counts and formats. They cannot show refresh or timing. Pair each screenshot with the URL, the device, the time and, where refresh matters, the network log export, so the finding can be checked by someone who was not there.
How long should an auditor stay on each page?
Long enough to read the article at a normal pace and then wait two minutes untouched at the end. Google Ad Manager sets a 30 second minimum for timed refresh on web, so a two minute wait gives a timed slot room to fire more than once and shows the spacing.
Should the publisher be told about an audit?
Not before the visit, since the aim is to see what readers see. After the audit, sharing the sheet rows and the sanitized network log with the seller is a fair way to ask whether a lighter layout exists for a deal, and it gives the publisher a chance to explain a finding.
Sources
- The Better Ads Standards (opens in a new tab)Coalition for Better Ads
- Mobile ad experiences (opens in a new tab)Google Web Tools Help (Ad Experience Report)
- Desktop ad experiences (opens in a new tab)Google Web Tools Help (Ad Experience Report)
- About the Ad Experience Report review process (opens in a new tab)Google Web Tools Help
- Ad refresh (opens in a new tab)Google Ad Manager Help
- Network features reference (opens in a new tab)Chrome for Developers (updated July 16, 2024)
- Simulate mobile devices with device mode (opens in a new tab)Chrome for Developers (updated February 20, 2024)
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.