Yes — for most small businesses, Advantage+ is now the sensible default. Meta’s AI decides targeting, placements and budget split better than a hand-built audience in almost every account under $10,000 a month. Keep manual controls only where you have a real constraint: a service area, a licence limit, or an audience you must exclude.
What does Advantage+ actually do differently?
Advantage+ hands three decisions to Meta’s AI that you used to make yourself: who sees the ad, where it appears, and how budget moves between audiences.
In a manual campaign you pick interests, behaviours and demographics, then choose placements. In an Advantage+ campaign you give Meta a conversion goal, a budget, a broad geography and your creative, and the system finds the buyers. Since February 2025, Meta stopped making advertisers choose between the two modes at setup — the two campaign types became Advantage+ sales campaigns and Advantage+ leads campaigns, with an "Advantage+ on" label showing which elements are automated, according to Social Media Today’s coverage of that rollout on 6 February 2025.
The practical difference for a shop owner: you spend your time on offers and creative instead of audience-building. That is not a downgrade. A 50,000-person interest audience in the Greater Toronto Area was always too small for the system to learn from quickly.
Why does Meta keep pushing automation on small accounts?
Because Meta’s own delivery engine now predicts intent better than a manual audience can describe it.
Two systems sit underneath every Facebook and Instagram auction. Andromeda is the retrieval engine that decides which ads are even eligible to show you; it rolled out in late 2024 and became core to the platform through 2025. GEM, Meta’s generative ads recommendation model, feeds predictions into Andromeda and began rolling out in mid-2025. Meta says GEM is "4x more efficient at driving ad performance gains" than its earlier ranking models, as reported by Search Engine Land on 28 January 2026.
Treat that 4x as Meta’s number, not an independent audit. But it is why segmented interest stacks stopped beating broad targeting around 2024.
Meta’s own figures for automation, reported in February 2025, are more modest than the marketing suggests: a 10% lower cost per qualified lead on leads campaigns with Advantage+ switched on, and a 5% median decrease in cost per result for advertisers who adopted Opportunity Score recommendations. Advantage+ shopping campaigns grew 70% year over year in that quarter — that is adoption, not performance.
Advantage+ vs manual: which controls do you actually lose?
You lose granular audience and placement control; you keep geography, exclusions, budget, schedule and creative.
| Control | Advantage+ | Manual campaign |
|---|---|---|
| Interest and behaviour targeting | Suggestion only — Meta can go beyond it | Hard limit |
| Placements (Feed, Reels, Stories, Audience Network) | Automatic across all eligible | You choose each one |
| Geography / service area | You set it, and it is respected | You set it |
| Audience exclusions (existing customers, past leads) | You set it, and it is respected | You set it |
| Budget split between audiences | Meta reallocates continuously | You split by ad set |
| Creative testing | Meta picks winners inside the campaign | You isolate by ad set |
| Best fit | Most accounts under $10k/month | Licence, compliance or tight-radius constraints |
The row that matters most is exclusions. Owners often assume automation ignores them. It does not — an excluded customer list stays excluded, which is why uploading your customer list is the first thing we do on any new Meta ads management account. A GTA mortgage brokerage we run sits at 1,113 leads a month at $7.69 per lead as of August 2026, on broad Advantage+ delivery with a clean exclusion list and a tight service-area radius.
When should a small business still build the campaign manually?
Only when a business rule, not a preference, requires it.
Three cases come up repeatedly. First, a genuinely small service radius: a GTA mobile detailing business that will not drive past one city needs tight geography, and a very small radius starves the learning phase, so one clear manual audience sometimes stabilises faster. Second, regulated categories, where Meta’s special ad category rules already strip targeting options and you need to know exactly what is running. Third, diagnostic testing — if you cannot tell which of four offers is working, isolating them in ad sets for two weeks buys a clean read that Advantage+ will not give you.
Outside those three, manual campaigns in a small account usually mean higher costs and slower learning.
Why do my Advantage+ numbers look different from my CRM?
Partly because Meta changed how it counts conversions in March 2026, and partly because Meta and your CRM have never counted the same thing.
From late March 2026, only link clicks count toward click-through attribution for website and in-store conversions; likes, shares and saves moved into a renamed "engage-through" metric. Meta also cut the video engaged-view window from 10 seconds to 5, saying that 46% of Reels purchase conversions happen within the first two seconds of attention, according to Search Engine Land on 3 March 2026.
If your reported conversions dropped in spring 2026 without a matching drop in actual booked jobs, that is the likely cause. The fix is not a new campaign. It is picking one number — booked jobs, or signed clients — and judging Meta against that, using Meta’s reporting only to compare ads against each other.
What to do this week
- Upload or refresh your customer and past-lead lists as exclusions on every active campaign. This is the single change that most often lifts lead quality in an automated account.
- Check your Opportunity Score in Ads Manager and action only the recommendations that do not conflict with your service area or exclusions. Meta reported a 5% median decrease in cost per result for advertisers who adopted them (February 2025).
- Pick your one true number — booked jobs, signed clients, or paid orders — and compare it against Meta’s reported conversions for the last 90 days. If the gap is wide, the attribution change above is probably why.
- If you are running four or more ad sets on a budget under $3,000 a month, consolidate into one Advantage+ campaign and let the budget pool.
Frequently asked questions
Is Advantage+ cheaper than a manual Meta campaign?
Sometimes, and by less than the marketing implies. Meta’s own figure, reported in February 2025, was a 10% lower cost per qualified lead on leads campaigns with Advantage+ enabled. That is a real improvement but not a transformation, and it is Meta’s number rather than an independent study. The bigger saving is usually operational: one pooled budget learns faster than five starved ad sets, and you spend your hours on creative instead of audience research.
Can I still exclude my existing customers in Advantage+?
Yes. Exclusions are one of the controls Advantage+ keeps. You upload a customer list or use a website custom audience and exclude it at the campaign level, and Meta respects it. This matters most for home services and mortgage or real estate businesses, where showing a lead-generation ad to someone who already signed wastes budget and irritates the customer. Refresh the list monthly — a stale exclusion list is almost as bad as none.
How much budget does an Advantage+ campaign need to work?
Enough to produce roughly 30 to 50 conversion events a week, which is what the delivery system needs to exit learning. For a home services business at a $40 cost per lead that is around $1,700 a month; for an e-commerce brand optimising to purchases it is usually higher. If your budget cannot reach that at your current cost per result, optimise to a cheaper upstream event — a lead form submit rather than a booked appointment — and judge quality separately.
Does Advantage+ work for local service businesses, not just e-commerce?
Yes, and Advantage+ leads campaigns exist specifically for it. The constraint is geography rather than the automation: a broad audience inside a realistic drive time works well, while a 10-kilometre radius may be too small for the system to learn quickly. Set the radius to the area you will genuinely travel to, exclude your existing customers, and give the campaign at least two weeks before judging it.
Want a second opinion on whether your Meta account should be automated or rebuilt? Book a free 48-hour audit and we will tell you which of the two it is.
Sources: Social Media Today, "Meta Rolls Out Updates to Advantage+ AI Ad Targeting Tools," 6 February 2025; Search Engine Land, "Inside Meta’s AI-driven advertising system: How Andromeda and GEM work together," 28 January 2026; Search Engine Land, "Meta introduces click and engage-through attribution updates," 3 March 2026. Figures attributed to Meta are Meta’s own reported numbers, not independent studies.