On this page
1. Consolidate campaigns and ad sets
Each ad set learns on its own, so splitting a small budget across many sets starves them all. Run fewer, broader ad sets and let the delivery system find the buyers.
The budget side is on CBO vs ABO. Overlapping audiences compete with each other, so merge them where you can.
A practical starting point is one campaign with two or three ad sets and several ads in each. Merge ad sets whose audiences overlap heavily and avoid launching a new ad set for every idea. More budget per ad set means faster, steadier learning and easier reading of results.
2. Choose the objective that matches the goal
Optimize for the action you want: purchases, leads or messages. Optimizing for clicks gets clickers. If you lack the volume for purchases, a step higher in the funnel is a legitimate start.
See how to advertise on Facebook for the setup order.
Sales, leads and engagement objectives optimize for different people. If you want purchases and have a working pixel, use the sales objective. If you want contacts, use leads. Check your result column matches the real business action, and move the objective down the funnel only if the data is too thin.
If you run lead ads, send leads to a CRM or inbox that someone checks quickly, and feed the outcome back so Meta can learn which leads were good. Optimizing for form fills alone often brings in the cheapest and least serious people, so quality signals matter once you have enough volume.
Want this checked in your Meta account?
Reads your campaigns, ad sets, ads and pixel health. Every change waits for your approval.
3. Fix tracking before spending
Meta can only optimize toward events it sees. Send them from the browser pixel and from the server, and check that they are deduplicated.
See Meta Conversions API and pixel not tracking.
In Events Manager, look at which events arrive from the browser and which from the server, and whether match quality is acceptable. Fix duplicate counting before you scale. Without good signals, every other tip on this list performs worse.
Test by completing a purchase or form yourself and watching the event arrive in Events Manager. Check that the value and currency are sent for purchases, since value-based optimization needs them. A short monthly check of event quality keeps problems from building up silently after a site change or a plugin update.
4. Test different ideas, not small edits
The creative is the targeting now. Test distinct hooks, formats and angles, not a different button color. Retire an ad when frequency climbs and results fall.
See Facebook ad fatigue for the signs.
Brief three angles: a problem, a result and a proof. Make each in at least two formats, such as a short video and a static image. Refresh when the same people see the ad many times and click-through falls. A steady rhythm of new ideas matters more than any single ad.
5. Try Advantage+ where it fits
Advantage+ campaigns automate audience, placement and budget choices. They suit stores with a catalog and good tracking. Run one alongside a manual campaign and compare cost per result.
See Advantage+ shopping.
Feed it varied creative and clean conversion data, and give it time. Compare against your manual setup for at least a couple of weeks. If cost per result is similar, the automated option saves effort. If it is worse, keep it as a test rather than the core.
6. Do not reset the learning phase
Big edits to budget, audience or creative restart learning and make results noisy. Make one change at a time, and wait for results to stabilize before the next.
See the learning phase.
If you must change something, change one thing, small, and then wait. Avoid edits in the first few days. Note each change in a log with the date so you can explain jumps in results later.
7. Start with automatic placements
Automatic placements give the system the most room to find cheap results. Check the breakdown by placement and exclude only those that clearly lose money.
Make sure each creative is sized for Reels and Stories: placements.
Review the breakdown by placement after you have enough spend to judge. Make vertical versions for Reels and Stories, since a square image can look wrong there. Keep safe zones in mind so text is not covered by the interface.
8. Retarget and exclude buyers
Warm audiences convert cheapest. Exclude recent purchasers from prospecting so you do not pay to show ads to your own customers.
Create custom audiences from site visitors, video viewers and customers. Exclude buyers from prospecting, and run a separate retargeting ad set with a short list of messages. Check frequency so people do not see the same ad too many times.
The setup is on Facebook retargeting campaigns. Sprites can check pixel and Conversions API health, and propose audience and budget changes on Meta as approval cards you approve before they go live.
Questions
How do I create ads on Facebook for free?
Creating an ad in Ads Manager costs nothing, and you can build a campaign without paying. Showing the ad to people costs money from your budget. For unpaid reach, post organically from your Page.
What is a good Facebook ad budget?
Enough to reach a handful of results per ad set per week. Work back from your target cost per result. See Facebook ads cost.
Which Facebook ads are most successful?
Those that show the product being used or a customer saying why, in a format native to the placement. Short video usually leads, and static images with a clear offer still work.
When should I scale?
Scale when results are steady over several days at your target cost, and raise budget gradually. See scaling Facebook ads.