Effective Facebook Ads: What Works and the Mistakes to Avoid
Short answer Effective Facebook ads come from a few right settings and fresh creative more than from clever targeting. Pick the objective for the result you want, track it with the Meta pixel and Conversions API, make your service area the only hard limit, run few ad sets, leave them alone while they learn, and compare versions with Meta's A/B test.
Want help applying this to your business? Talk it through on a free 30-minute call.
Book a callThe Facebook ads checklist
Ten checks, in the order you build a campaign. Checks 1–9 rest on Meta's own help pages as of November 2026; the sections below give Meta's wording.
| # | Check | Why |
|---|---|---|
| 1 | Pick the objective for the result you want: Leads or Sales for bookings, not Traffic | Ads Manager delivers toward the performance goal you set |
| 2 | Install the Meta pixel and add the Conversions API | Meta can only find more bookers if it sees who booked |
| 3 | Set your service area as the location; leave age and interests open | Location is a hard limit; the rest are suggestions |
| 4 | Run one campaign with one or two ad sets | Each extra ad set splits what Meta learns from |
| 5 | Budget for about 50 results a week per ad set, or expect "Learning limited" | That's when Meta says delivery turns stable |
| 6 | Put a few clearly different images or videos in each ad set | One creative seen too often gets more expensive |
| 7 | Size for phones: 4:5 feed images, 9:16 vertical video, captions on | Text near the edges gets cropped or covered |
| 8 | Write about your service, never about the viewer's health, money or other personal traits | Meta rejects ads that imply personal attributes |
| 9 | Test with Meta's A/B test; leave a new ad set alone for its first week | Big edits restart learning; on-off tests mislead |
| 10 | Answer every lead within minutes | The ad stops at the lead; the booking needs a reply (speed to lead) |
Facebook ad targeting: what works now
Broad audiences inside a firm location. Meta's AI picks the people; your job is to set the limits it must respect.
"For example, if you choose Locations, we won't target beyond your locations unless you select Reach more people likely to respond to your ads." (Meta Business Help: About Audience controls and Audience suggestions)
Age, gender and interests become suggestions Meta may go past when it predicts a better result (same page). Meta reports 9.7% lower cost per result for Advantage+ audience with the Traffic, Engagement and Leads objectives; that's its own figure, not an independent test (Meta Business Help: About Advantage+ audience). It learns from past conversions and pixel data, which is why check 2 matters. The settings one by one: Meta Advantage+ for a local business.
What still sharpens targeting:
- Your own lists. A custom audience can be a customer list of emails or phone numbers, website visitors seen by the pixel, video viewers, or people who opened your lead form (Meta Business Help: About custom audiences). Exclude current customers from a new-customer offer this way.
- Lookalikes of your best customers. Meta finds people similar to a source list, and notes that a list of your best customers may work better than all of them (Meta Business Help: About lookalike audiences).
- Fewer, bigger ad sets. Meta's example: three ad sets aimed at people interested in dogs, dog toys and dog collars "could decrease performance"; one ad set with all three interests may do better (Meta Business Help: Combine ad sets to reduce audience fragmentation). Serving several towns? Meta suggests one ad set over the larger area, not one per town (Meta Business Help: About managing ad volume).
What you can't do anymore: exclude people by interest. Meta removed detailed targeting exclusions in 2025, citing an advertiser test where median cost per conversion was 22.6% lower without them (Meta for Developers blog, January 21, 2025). Excluding a custom audience still works.
Ads for housing, jobs or credit that reach the US or Canada get limited or no age, gender, postal-code and lookalike targeting (Meta Business Help: How to choose a Special Ad Category). Details: Meta special ad categories.
Facebook ad content: what stops the scroll and gets approved
One message, one action, sized for a phone.
| Item | Meta's recommendation |
|---|---|
| Feed image | 4:5, 1440 x 1800 pixels |
| Feed video | 4:5; captions and sound "Optional, but recommended" |
| 9:16 ads in Stories, Reels and Feed | Keep the top, bottom and side edges free of text and logos |
| Primary text | 50–150 characters |
| Headline | 27 characters |
Sources: Meta Ads Guide (Facebook Feed image and video specs); Meta Business Help: About text overlays and the safe zone.
Meta's current text guidance sets no 20% limit, or any other cap on how much of an image text may cover. It asks for a clean, large font in a contrasting color, nothing covering the picture, and text inside the safe zone. Plus:
"Don't communicate too many messages because ads usually only have one call to action." (Meta Business Help: About text overlays and the safe zone)
One wording rule catches many service businesses:
"In addition, ads must not contain content that asserts or implies personal attributes." (Meta Advertising Standards: Privacy violations and personal attributes)
It covers health, age, religion, "vulnerable financial status," criminal record and more. Meta's own examples:
| Not allowed | Allowed |
|---|---|
| "Do you have diabetes?" | "Bulimia counseling available" |
| "Are you a convicted felon?" | "Services to clean up any previous offenses" |
Name the service, not the viewer's problem: "Back pain treatment in Mesa" rather than "Is your back pain getting worse?"
Keep creative fresh. When people have seen an ad too often, Ads Manager shows "Creative limited" if its cost per result is above your past ads, and "Creative fatigue" at twice or more. Meta's first fix is a new ad with a "materially different" image or video, and it suggests keeping the old ad running rather than pausing it (Meta Business Help: About creative fatigue recommendations). One ad can hold up to 10 creative assets, so you get variety without piling up ads (Meta Business Help: About managing ad volume).
Common Facebook ad mistakes
| Mistake | What happens | Do instead |
|---|---|---|
| Boosting a post to get bookings | A boost may optimize for likes, comments, shares or awareness | Ads Manager with Leads or Sales (boost or Ads Manager) |
| Picking Traffic because clicks are cheap | Meta looks for people likely to click, not to book | Optimize for the lead or booking (lead quality) |
| One ad set per audience, town or placement | Each learns from too few results | Combine them; see age or region splits in breakdowns |
| Editing in the first week | A significant edit sends the ad set back into learning | Wait for about 50 results or a week, then change one thing |
| A big budget jump | $100 to $1,000 may restart learning; $100 to $101 likely won't (Meta's example) | Raise in moderate steps |
| Testing by switching ads on and off | Audiences overlap and results mislead | Meta's A/B test, same budget for both versions |
| Pixel on pages with health or financial details | Meta prohibits sending that data | Read the HIPAA risk for clinics |
Sources: Meta Business Help pages on boosted posts, significant edits, ad volume and Meta Pixel.
The learning phase is behind most of these. An ad set leaves it once it delivers stably:
"This usually occurs after about 50 results in the week after the ad set's last significant edit." (Meta Business Help: About the learning phase)
Significant edits include any change to targeting, creative or optimization event, adding an ad, a new bid strategy, or a pause of 7 days or more (Meta Business Help: Significant edits and learning phase). Below about 50 results a week, the status reads "Learning limited," which Meta says "isn't a penalty." Its fixes: combine ad sets, widen the audience, raise the budget, or pick a more frequent optimization event (Meta Business Help: About learning limited). If you change the event, stay close to a booking: a lead, not a page view.
On testing, Meta is blunt:
"We do not recommend testing informally, such as by turning ad sets or campaigns on and off manually." (Meta Business Help: About A/B testing for Meta Ads)
Its A/B test splits the audience so nobody sees both versions (same page).
Outdated Facebook ads advice to ignore
| You may read | What's true now |
|---|---|
| Objectives come in three groups: awareness, consideration, conversion | Six objectives: Awareness, Traffic, Engagement, Leads, App promotion, Sales |
| Research people in Audience Insights | "Audience Insights is no longer available"; use Meta Business Suite Insights |
| Watch your Relevance Score | Replaced by quality, engagement rate and conversion rate rankings |
| Keep image text under 20% | No percentage cap; keep text legible and in the safe zone |
| Exclude interests to tighten targeting | Detailed targeting exclusions were removed in 2025 |
| Split audiences into many small ad sets | Meta recommends combining them |
Sources: Meta Business Help on ad objectives, Business Suite Insights, ad relevance diagnostics and text overlays; Meta for Developers blog.
The three rankings appear only after 500 impressions, and Meta says they "aren't inputs into the ad auction" (Meta Business Help: About ad relevance diagnostics). Use them to see whether the creative, the page after the click, or the audience needs work. Judge the ad by cost per lead or booking.
Example
Illustrative; round numbers.
A dog groomer with $20 a day
About $600 a month. At $25 a booking that's 24 bookings, about 6 a week, far below the 50 Meta's learning wants. So: one campaign, one ad set, Leads objective, a 10-mile radius, three different videos in one ad. Expect "Learning limited" and judge the month on bookings, not the status.
A Toronto physiotherapy clinic with four ad sets
Four ad sets at C$10 a day, one per neighborhood, each learning from a handful of leads. Combined into one ad set at C$40 a day covering all four, every lead feeds one pool. The ads say "Physiotherapy for running injuries," not "Is your knee pain back?"
A landscaper whose ad wore out
One spring video ran for three months. Cost per lead rose from $30 to $65 and Ads Manager showed "Creative fatigue." A materially different video, a finished patio walkthrough, went into the ad set beside the old one.
FAQ
How much should I spend on Facebook ads?
Meta sets no single minimum; it says minimums "may vary by country and objective." A useful yardstick: enough for about 50 results a week per ad set. Typical prices: what Facebook and Instagram ads cost.
How long before I judge a Facebook ad?
About a week, or 50 results, after its last significant edit. Relevance rankings need at least 500 impressions.
Is the 20% text rule still in place?
No. Meta's current guidance sets no percentage. It asks for legible text inside the safe zone and one message per ad.
How many ads should one ad set have?
Meta gives no fixed number. It warns that too many ads each deliver less often, and suggests fewer ads with varied creative, up to 10 assets in one ad.
Different for you?
- You're choosing between boosting and Ads Manager: boost post or Ads Manager.
- You run a local service and get weak form leads: Facebook ads for a local service business.
- You advertise homes, jobs or credit: Meta special ad categories.
- You're in health or wellness: Meta's health and wellness restrictions.
- Your ad account was shut down: Facebook ad account disabled.
- Your leads are in Canada and you text them back: CASL consent.
- You'd rather have the ads built and judged on bookings: Paid Ads Agent.