AdSense gives you two paste-and-go choices: let Google decide where ads appear (Auto Ads), or pick the spots yourself (manual units). The marketing copy says Auto is easier and earns more. In practice, well-placed manual units on an intent-heavy content page often pull 15-25% more RPM in publisher audits, because each impression sits next to high-value content instead of being scattered for sheer volume. This is the decision that actually matters, plus the Auto Ads control changes that landed in April-June 2026. If you have already mixed both and are seeing layout shift or double-stacked ads, see the troubleshooting page for that specific mess.
TL;DR
- Default for an indie content site: manual. You reserve space (no CLS), you put ads where intent is highest, and you keep full design control.
- Auto Ads makes sense if you have 100+ pages, can monitor Core Web Vitals weekly, and value time saved over fine-grained control.
- Hybrid is the 2026 consensus for sites past ~20k monthly pageviews: 2-3 manual units in prime spots, Auto Ads filling the margins with anchor/vignette OFF.
- The old Auto Ads “ad load slider” was retired on April 16, 2026. You now set a maximum number of in-page ads and a minimum distance between them — use those instead of dumping a generic slider at 80%.
How Auto Ads and manual placements actually differ
Auto Ads is Google’s machine-learning placement system: you add one script tag and Google decides location, format, and density per page. It earns by maximizing impressions — more ad slots across more pages. Manual placements require you to create individual ad units and embed them in fixed spots, and they earn by maximizing the value of each impression — putting ads next to comparison tables, reviews, and decision-moment content that advertisers bid up.
The two also behave differently for page speed. Manual units render in space you reserved in the template, so they do not move content. Auto Ads injects slots dynamically as the model decides, which is the single most common cause of a Cumulative Layout Shift (CLS) regression on indie sites. In 2026, a CLS slip from “Good” to “Needs Improvement” can be what keeps a page out of Google Discover, so it is not a cosmetic concern.
| Dimension | Auto Ads | Manual placements |
|---|---|---|
| Setup effort | One script tag, done | Create + embed each unit |
| Revenue model | More impressions (volume) | Higher value per impression |
| Typical RPM (well-tuned) | Baseline | Often +15-25% on intent pages |
| CLS / layout shift | Higher risk (dynamic injection) | Low if you reserve heights |
| Design control | Low — Google places | Full |
| Best fit | 100+ low-touch pages | Curated content, brand-sensitive |
What changed in Auto Ads (April-June 2026)
If your last mental model of Auto Ads is the old slider, refresh it:
- Ad load slider removed (April 16, 2026). It was replaced by three banner controls: a maximum number of in-page ads per page, a minimum distance between ads, and a toggle to find additional placements on article pages. Existing slider users were auto-migrated to approximate values, so check yours — the migrated cap is often higher than you want.
- Vignette ads now have six triggers (since February 9, 2026), including end-of-article and 30-second-inactivity triggers. The browser back-button trigger is being dropped on June 15, 2026. Vignettes still respect your frequency cap, but they spike bounce rate on mobile — keep them OFF until you have data.
- Dynamic anchors come to desktop. The “collapsible anchor ads” toggle was renamed “dynamic anchors,” and on screens wider than 1000px the new desktop anchor activates automatically for opted-in publishers unless you change it before June 20, 2026.
- New Auto Ads formats worth knowing: side rail ads (widescreen left/right), multiplex native grids, and ad intent links/chips that inject ads into existing text. Side rails and multiplex can earn well; ad intent is more intrusive — test deliberately.
The practical takeaway: Auto Ads is more configurable than it was, but every one of those toggles is a lever you now own. “Set it and forget it” is no longer the honest description.
Step by step: run the comparison on your own site
- Inventory what you have. List deployed ad units, or decide which mode to try first if you are starting fresh.
- Manual baseline: create 2-4 in-article units in AdSense, embed them with reserved heights (set
min-heighton the container so the slot never shifts content), run for 14 days. Record RPM, LCP, CLS, INP. - Auto Ads baseline: enable Auto Ads, set maximum number of ads to 3-4 and a sensible minimum distance, turn anchor and vignette OFF, deploy. Wait 7-14 days for the model to converge, then measure the same metrics.
- Compare after 30 days. Look at RPM, session duration, bounce rate, the three Core Web Vitals, and AdSense’s own page-experience signals side by side. A 10% RPM win that costs you 0.15 of CLS is usually a net loss.
- Build the hybrid: keep 2-3 manual units in the highest-intent spots (in-article anchor, end-of-article), enable Auto Ads with anchor/vignette OFF and a modest in-page cap so it only fills secondary space.
- Re-evaluate quarterly. As traffic grows and content evolves, the optimal mix shifts. Treating ad strategy as a one-time decision is the biggest mistake.
Common pitfalls
- Enabling Auto Ads and never checking CWV again. Dynamic injection commonly regresses CLS by 0.1-0.3 — enough to flip “Good” to “Needs Improvement” silently. Pull a PageSpeed Insights field-data check weekly for the first month.
- Running Auto Ads and many manual units without coordinating. They compete for the same slots and you get duplicate ads stacked in one viewport. Use Auto Ads’ excluded areas to fence off zones your manual units already own.
- Leaving anchor and vignette ON by default. Both are bounce-rate risks on mobile. Watch per-device fill rate while you test — if mobile and desktop diverge sharply, the network is usually telling you something about the placement, not the demand (why mobile vs desktop fill rates diverge).
- Never A/B testing manual positions. The first placement is rarely the best. Quarterly testing is realistic and worth it.
- Assuming Auto Ads is always lazier. The CWV fixes Auto Ads often forces can cost more total time than placing 3 manual units thoughtfully once.
When to skip manual entirely
Sites with strict design or brand constraints stay manual-only. Sites with hundreds of low-traffic pages where hand-placing units is impractical lean Auto Ads — just cap the in-page count and keep overlays off. For everyone in the middle, hybrid wins.
FAQ
- Which earns more, Auto Ads or manual? On intent-heavy content (comparisons, reviews, how-to), a well-tuned manual or hybrid setup typically beats pure Auto Ads by 15-25% RPM in audits, because value-per-impression outweighs raw volume. Auto Ads can edge ahead on sprawling, low-curation sites where you would never hand-place ads anyway. Net economics for indie sites under ~50k MAU usually favor manual or hybrid.
- Can I mix Auto Ads and manual placements? Yes, and it is the 2026 consensus. Use manual for premium spots (in-content, end-of-article), let Auto Ads fill the margins. Use excluded areas so you never double-show in the same viewport.
- What replaced the Auto Ads ad-load slider? As of April 16, 2026, three banner controls: maximum number of in-page ads, minimum distance between ads, and a toggle for finding extra placements on article pages. Migrated accounts inherited approximate values — review them.
- Should I turn on vignette and anchor ads? Start with both OFF. Vignettes gained new triggers in February 2026 (the back-button trigger is removed June 15, 2026) and can spike mobile bounce. Desktop dynamic anchors auto-activate for opted-in publishers by June 20, 2026 unless you change the setting. Enable only after you have measured the bounce-rate cost.
- Does Auto Ads adapt over time? Yes — Google’s model optimizes for revenue from your traffic patterns. The first 7-14 days are learning; data after that is more reliable. Major content changes can reset the learning period.