Choosing the best ad format for your website comes down to matching format to placement and traffic behavior, then measuring the result against a single revenue metric instead of guessing. Banners, native units, video, and interstitials all pay differently depending on where your visitors come from and how they read your pages, so the "best" format isn't universal — it's the one that earns the most from your specific audience without pushing them away. This tutorial walks through a repeatable process for testing formats on your own site and reading the results correctly.
Most publishers skip straight to picking a format because it looked good on someone else's site. That's backwards. The format that works for a news site with fast scroll depth can flop on a tool or utility site where visitors stay on one screen. The steps below start from your traffic pattern, not from a list of ad types.
Before you start
You'll need an ad management setup that can run more than one format at once — most ad servers and networks support this, including header bidding wrappers if you're running one. You also need basic analytics already wired up: pageviews, average session duration, scroll depth if you can get it, and bounce rate. If you don't have an analytics tool connected yet, set one up first — you can't judge a format change without a baseline to compare it against.
Finally, know your traffic mix. A site that's 80% mobile makes different format decisions than one that's mostly desktop, because formats like large leaderboards and multi-unit sidebars don't translate to small screens. Pull that split before you start testing.
Step-by-step: how to choose the best ad format for your site
1. Map your page layout to natural ad slots. Walk through your own site as a visitor would and mark where attention naturally pauses: above the fold, between paragraphs in long content, at the end of an article, in a sidebar next to static content. Don't force a slot where there isn't a natural pause — an ad squeezed between two sentences a reader is mid-thought on gets scrolled past without registering, which wastes the impression even if it technically "loads."
2. Match format to slot type, not to what pays the most on paper. Use in-page display banners (leaderboard, medium rectangle) for above-the-fold and sidebar slots where the layout is stable. Use native ad units — styled to match your site's fonts and card layout — inside content feeds or between article recommendations, where a display banner would look out of place. Use in-content video for long-form content where a visitor is already committed to reading. Reserve interstitials and popunders for moments with a natural break, like a page transition, not mid-scroll on a page someone is actively reading.
3. Check format standards before you commit. The IAB's display advertising guidelines define the standard sizes (300×250, 728×90, 320×50, and others) that most demand sources bid on. Sticking to standard sizes matters practically: non-standard dimensions get fewer bids because fewer advertisers have creatives built for them, which shows up as lower fill rate even before you've tested anything.
4. Screen out formats flagged as bad experiences. The Coalition for Better Ads publishes research-backed standards on ad experiences that measurably drive visitors away — full-page interstitials on entry, ads with countdown timers before content loads, and flashing animated formats are on that list. These formats can still generate revenue, but if your site depends on repeat visits, they're a real long-term cost even when the short-term CPM looks attractive. Decide up front whether you're optimizing for one-time visits or return traffic, because that changes which formats are even worth testing.
5. Run a small, isolated test. Pick one slot and test two formats against each other — for example, a 300×250 display unit versus a native unit in the same position — rather than changing your whole page at once. Split traffic evenly (by session or by day, whichever your ad server supports) and run the test for at least a full week to average out weekday/weekend traffic differences.
<!-- Example: two ad units in the same slot, swapped by a simple traffic split -->
<div id="ad-slot-sidebar">
<!-- Variant A: standard 300x250 display unit -->
<div class="ad-unit" data-format="display" data-size="300x250"></div>
<!-- Variant B: native unit styled to match site cards -->
<!-- <div class="ad-unit" data-format="native"></div> -->
</div>
6. Read revenue per session, not CPM, as your primary number. A format with a higher CPM but lower fill rate can lose to a format with a lower CPM and near-total fill. Revenue per session (or RPM — revenue per thousand sessions) accounts for both at once, which is why it's the number to trust over any single-metric comparison.
7. Roll out the winner, then test the next slot. Once one slot has a clear winner over a full test period, apply it and move to the next slot on your list from step 1. Testing one variable at a time is slower than changing everything simultaneously, but it's the only way to know which change actually caused the result.
How to tell it's working
Compare revenue per session for the tested slot before and after rollout, using the same traffic sources and date range length you used during the test — a holiday week or a traffic spike from an unrelated source will distort the comparison if you're not careful. A genuine improvement should show up as a stable lift over at least two to three weeks, not a one-day spike.
Also check viewability alongside revenue. The industry standard, set by the Media Rating Council, counts a display ad as viewable when at least 50% of its pixels are in view for a minimum of one continuous second (Google's viewability measurement guidance uses this same threshold). A format that technically pays well but sits below 50% viewability is earning on impressions nobody actually saw, which tends to catch up with you later as demand partners reprice or drop the placement. If viewability drops after a format change, that's a signal worth acting on even if revenue looks flat or slightly up.
Bounce rate and session duration are your guardrails. If a new format lifts revenue per session but bounce rate climbs sharply on the pages carrying it, you've likely traded short-term revenue for audience retention — worth knowing even if you decide the trade is acceptable.
Troubleshooting
Revenue looks flat even though the new format should pay more. Check fill rate first. A format that pays a higher rate per impression is worthless if only a fraction of requests get filled — few demand partners may be bidding on that size or placement yet. Give it more time or widen the demand sources bidding on that slot before concluding the format doesn't work.
Viewability dropped after switching formats. This usually means the new unit is placed lower on the page than visitors typically scroll, or it's sized in a way that gets cut off on mobile. Recheck the slot's actual visible position on the devices your traffic actually uses, not just on desktop.
One format wins on paper but visitors are visibly annoyed. If you're getting direct complaints, a spike in ad-block installs from your own traffic, or a sharp bounce rate increase, the format is probably too aggressive for your audience even if the test data looks favorable. Formats like interstitials and heavy autoplay video carry this risk — test the softer version (delayed interstitial, muted video) before ruling the format out entirely.
Test results are noisy and inconclusive. Low-traffic slots take longer to reach a statistically meaningful result. If a slot gets under a few thousand impressions a week, extend the test window rather than trusting an early lead — small sample sizes swing wildly day to day.
Most publishers running this process end up managing several formats across several partners by hand, which is exactly the coordination problem a pooled-demand network like Adsy is built to simplify — one tag, multiple formats, bids compared automatically instead of managed slot by slot.
FAQ
What's the single best ad format for a website?
There isn't one — the best ad format for website revenue depends on your layout, traffic device mix, and whether you're optimizing for one-time visits or repeat traffic. A native unit inside content and a standard display unit in the sidebar cover most sites reasonably well as a starting point.
How many ad formats should a site run at once?
Enough to cover your natural ad slots without crowding the page — most sites do fine with two to four formats total (for example, one display size, one native unit, and video if content supports it). More formats than that usually adds latency without adding proportional revenue.
Do intrusive formats like popunders and interstitials pay more?
Often yes, per impression, but they carry a real cost in bounce rate and repeat visits according to Coalition for Better Ads research on ad experiences visitors actively avoid. Whether the trade is worth it depends on whether your revenue model depends on return traffic.
How long should I test an ad format before deciding?
At least one full week to average out day-of-week traffic patterns, and longer for lower-traffic slots. Watch for a stable lift over two to three weeks before rolling a change out permanently.
Does mobile need different ad formats than desktop?
Yes. Large leaderboard and multi-unit layouts that work on desktop often get cropped or pushed below the fold on mobile. Test formats separately by device where your traffic split justifies it.
Conclusion
Finding the best ad format for your site is a process, not a single choice: map your natural ad slots, match formats to how visitors actually read the page, screen out formats known to hurt the experience, test one slot at a time, and read revenue per session alongside viewability rather than either number alone. The format that wins on your site depends on your layout and audience, so treat any list of "top ad formats" as a starting point for testing, not a final answer.
Key takeaways
- Match format to how visitors behave in each slot — native for content feeds, display for stable above-the-fold and sidebar positions.
- Stick to IAB standard sizes where possible; non-standard dimensions get fewer bids and lower fill.
- Judge formats by revenue per session, not CPM alone, since fill rate changes the real payout.
- Track viewability (50% of pixels for one second) alongside revenue — a high-paying format nobody sees isn't actually winning.
- Test one slot and one variable at a time, for at least a week, before rolling a format out site-wide.
