The first banner on a competitor's homepage is the shortest answer to what that brand most wants to sell that week. A team puts a single campaign in that slot after long discussions. The second and third slides of the slider are a ranking too: what first, what next.
That is why banner tracking is a different job from tracking campaign pages. A campaign page shows what exists, the homepage shows what is being pushed. This post covers how to record banners and how to read their order.
The offer is written inside the image
The first trap in banner tracking is looking for text. In the page source you usually find an image URL, a link and perhaps a short alt text. The real information, such as the percentage, the last day or the minimum amount, is printed on the image. Tracking that compares HTML sees only a changed file name when a new image arrives.
So the banner has to be read as an image: what does it say, which product or category is shown, is it an offer or a brand message? If you do this by hand, taking a screenshot and noting those three answers is enough. If you automate it, you use a model that reads images, on the condition that it leaves unreadable fields empty. A rate filled in by guessing does more harm than a rate never written down.
Mobile is a separate shop window
Many sites show different banners on mobile. Sometimes a narrow crop of the same campaign, sometimes a different campaign altogether. The order can differ too: an offer on the third desktop slide can be the first slide on mobile. If most traffic comes from mobile, the competitor's real priority is the mobile order.
Set up tracking separately for the two views. If you merge desktop and mobile records into one row, you lose which offer was put in front of which audience.
What to record for each banner
- The image itself and the text read from it.
- Position: which slide of the slider.
- The target link, meaning which page a click leads to.
- When it was first seen and when it was last seen.
- View: desktop or mobile.
The target link is an underrated field. If the banner leads to a campaign page, you can start tracking that page too; if it leads to a product listing, you learn which category the competitor is pushing. If the same image now points somewhere else, the offer stayed the same and its destination changed.
Recognising the same banner
Did the banner change between two checks? The question is harder than it looks. Sites serve the same image at different sizes or under different URLs for caching. Tracking that decides by file name records the same banner as new every day. The opposite happens too: the competitor refreshes the image but keeps the same file name.
The practical answer is to recognise a banner by its content: the image itself, the text on it and the target link together. If all three match, it is the same banner, only its position may have changed. If the text is the same but the image was refreshed, that is an update, not a new campaign.
Reading the order
When a banner disappears matters at least as much as when it arrived. An offer that sits on the first slide for three days and then vanishes is a short, fast campaign. An offer that holds the first slide for weeks is the competitor's main message for that period.
Then there are moves. A banner that climbs from the third slide to the first shows a change in the competitor's priorities. A banner that drops from the first slide says something too: the campaign is still running but it is no longer the priority. Seeing this needs the history of the order over time, not a single moment. A weekly screenshot shows that moment, not the order.
It also helps to read banner changes together with changes on campaign pages. For the full setup: How to track competitor campaigns?
Taking the finding to the team
A banner record usually travels around the team as a screenshot and is forgotten two days later. What works better is a short sentence written next to the image: the competitor moved its new-customer offer to the first slide on mobile, it is still on the third slide on desktop, and the banner leads to a newly opened campaign page. The reader understands what changed and why it matters without opening a tab.
How often to check
Homepages turn over faster than campaign pages. Catching short-lived banners takes more than one check a day. A flash offer that opens at night and closes in the morning never shows up in tracking that checks once a day. Mark the gaps you could not check as well: you do not know what happened in those hours, so you cannot record that there was no banner.
With Adversee
Adversee reads the homepage banners of the competitors you track on desktop and mobile roughly every 2 hours. Added, removed and updated banners appear as separate events, with position, target link and first and last seen times on record. The image is read with AI: type, category and a campaign match suggestion with a confidence score, and any field it is not sure about stays empty.
Start tracking competitor moves automatically.