Signal First
Conversions API and event quality set up properly. Meta optimizes on what it receives, and browser-only tracking now misses a lot.
Facebook marketing now means Meta ads, because organic reach for business Pages is a fraction of what it was. That shifts the work to Facebook Ads management — signal quality through the Conversions API, audience structure, and creative volume, which is what Advantage+ campaigns actually optimize against. Coordinated with the rest of your social advertising.
Meta rewired targeting and measurement over the last few years. Accounts still run on the old playbook underperform for reasons nobody has diagnosed.
Conversions API and event quality set up properly. Meta optimizes on what it receives, and browser-only tracking now misses a lot.
Broad targeting with strong creative usually beats layered interest stacks now. We test that rather than assume either.
The ad does the targeting. We plan for a flow of concepts, not three assets optimized forever.
Where it fits the account, with the controls it still offers. Not switched on because it is default.
Meta-reported conversions checked against your actual sales data, with the gap stated.
Meta's tooling is the most mature of the social platforms and the most capable of flattering itself. These are the checks that matter.
All figures come from your own Meta account and your analytics. No benchmark or other-client result appears here.
Four inputs. Targeting is no longer the most important of them.
Review My Account →What conversion data Meta actually receives.
The ad itself — now the main targeting mechanism.
Broad, custom and lookalike, plus exclusions.
Campaign setup and budget consolidation.
Tracking, audiences, creative and honest measurement.
Server-side event tracking alongside the browser pixel, with event deduplication and match quality checked. This is the single biggest lever on Meta performance now.
Standard events mapped to what your business actually values, with purchase values passed properly so optimization targets revenue rather than actions.
Custom audiences from your own data, lookalikes where the seed list is strong enough, broad targeting where creative can carry it — and customer exclusions so you stop paying to reach existing buyers.
Budget consolidated enough to exit the learning phase rather than split across ad sets that never gather sufficient data.
An honest read on whether Advantage+ shopping or app campaigns suit your account, and which manual controls to keep.
Multiple concepts per campaign with enough budget per variant to produce a real result, plus a refresh plan for fatigue.
Feed, Stories, Reels and in-stream treated as different creative contexts rather than one asset stretched across all of them.
Meta-attributed conversions checked against your CRM or store data, with the discrepancy reported rather than absorbed.
Facebook and [Instagram](/services/social-media/instagram/) share Ads Manager but not creative expectations. Where the audience is professional rather than consumer, [LinkedIn](/services/social-media/linkedin/) usually costs more per click and converts better.
Facebook's audience has changed substantially, and assumptions from a few years ago no longer describe it.
Groups and local pages remain genuinely active, and often matter more than a brand page.
Broad reach and mature commerce tooling. Still the largest paid audience for many categories.
Where Facebook's audience skew is an advantage rather than the disadvantage it is often assumed to be.
Events remain a real use case with genuine discovery behavior attached.
Categories where Marketplace has become a primary discovery channel in its own right.
Where Facebook is usually the wrong platform, and Instagram is the relevant Meta property.
Fix the signal, then let creative do the targeting.
Pixel and Conversions API status, event match quality, deduplication and value passing. Campaigns optimizing on weak signal will spend confidently in the wrong direction.
Customer lists uploaded, lookalikes rebuilt from strong seeds, exclusions applied so acquisition budget stops reaching existing customers.
Fewer, better-funded campaigns so the learning phase completes rather than restarting every time budgets are split.
Several genuinely different concepts, made for the placements they will run in.
Structured testing, then results checked against your actual sales data before any conclusions are drawn.
Facebook is still the largest advertising audience available for many businesses, and a poor fit for others.
Reach and targeting infrastructure remain the most developed available.
Where community groups and local recommendation genuinely drive inquiries.
Custom audiences from your own data are where Meta advertising is most efficient.
Catalog-based advertising is mature and works well with clean product data.
Narrow B2B audiences and younger demographics are usually served better elsewhere, and we will say so.
Meta advertising changed substantially. Accounts still run the way they were in 2018 underperform for specific, identifiable reasons.
Because Meta's delivery system got considerably better at finding the right person from the creative itself, and because signal loss from privacy changes made narrow audiences less reliable to optimize within.
The old playbook was to layer interests and behaviors into tight audiences and let bidding compete within them. That approach now frequently underperforms a broad audience with strong creative, because the algorithm has more room to find people the interest tags would have excluded.
It is not universal — narrow audiences still make sense for genuinely niche products and for retargeting. What has changed is that broad should be tested rather than dismissed, and many accounts have never tried it.
It sends conversion events from your server to Meta directly, rather than relying only on the browser pixel. With browser tracking increasingly blocked, it is how a meaningful share of your conversions get reported at all.
The practical effect is on optimization rather than reporting. Meta optimizes toward the conversions it knows about; if a large fraction are invisible, it is learning from a partial and biased sample. Accounts that add server-side tracking properly often see performance change without any campaign edit.
It needs implementing carefully — event deduplication so browser and server events are not double-counted, and match quality checked so events actually attach to people. A badly implemented Conversions API can be worse than none.
For most businesses, not as a distribution channel. Organic reach for Pages is a small fraction of followers, and no posting cadence changes that meaningfully.
What the Page still does is act as a credibility check. People look before buying, and a Page with current information, recent activity and answered messages reads differently from one abandoned two years ago.
Groups are the exception worth knowing about. Community-based reach on Facebook still works where there is a genuine community to serve, and it is a different discipline from Page posting — closer to community management than to publishing.
A broad and skewing-older audience compared to Instagram or TikTok, which matters for targeting decisions and is frequently assumed rather than checked.
For many consumer categories that is an advantage — Facebook still reaches the largest and most demographically varied audience of any Meta property, and older audiences often have more purchasing power in categories like home services, healthcare and financial products.
Where it is a disadvantage is younger consumer brands, where the audience has largely moved to Instagram and TikTok. Running the same budget across both without acknowledging that split is a common and expensive assumption.
It is a server-side way of sending conversion events to Meta, alongside or instead of the browser pixel. For most businesses running paid social it is now closer to necessary than optional.
The reason is that browser-based tracking has been progressively restricted — by browser privacy features, by operating-system prompts and by ad blockers. Events that never reach the platform cannot be optimized toward, and the system makes worse decisions as a result.
Sending events from your server bypasses much of that, because the event does not depend on the browser cooperating. Running both together, with correct deduplication, gives the fullest picture.
The setup is not trivial and deduplication is where it goes wrong — the same conversion counted twice inflates reported performance and misleads the optimization. It is worth doing carefully rather than quickly.
Partly because many interest options were removed, and partly because the delivery system now finds relevant people better than manual definitions do.
Meta has withdrawn a large number of detailed targeting options over recent years, particularly those touching sensitive categories. Campaigns built on narrow interest stacks have progressively lost the specificity they were built around.
The larger change is that broad targeting with automated delivery now generally performs at least as well. The system has vastly more signal than an advertiser can specify, and constraining it with a narrow audience often prevents it finding people it would have found.
What still works is your own data — customer lists, site visitors, engagement audiences — and lookalikes built from them. Those are first-party signals rather than platform-inferred interests, and they have held up considerably better.
For most brand pages, as maintenance rather than as a growth channel. For local businesses and groups, genuinely more than that.
Organic reach from a brand page has declined over years as the feed prioritized friends, groups and recommended content. Expecting a page to build an audience through posting is expecting something the platform stopped offering.
What the page still does is answer questions people ask about you. Someone checking whether a business is real and active will look, and an abandoned page with a two-year-old post and an unanswered review answers badly.
Groups are the genuine exception. Community and local groups still produce real reach and real conversation, and for many local businesses that is where the actual activity is — not on the page they have been maintaining.
More consolidated than the older approach taught, because the delivery system learns better with concentrated data.
The historical pattern was many ad sets, each with a tightly defined audience and its own budget. That divided conversion data across ad sets that each saw too little, and none exited the learning phase properly.
Consolidation means fewer ad sets with broader audiences and budgets large enough to accumulate events. The system then distributes across placements and people itself, which it does better than manual splitting.
What still justifies separation is a genuine difference in economics or creative — a distinct product line, a different geography with different value, or an audience needing a fundamentally different message. Separation for tidiness costs performance.
Take them seriously, because for many local and community businesses that is where the actual activity is — not on the page they have been maintaining.
Facebook's feed has prioritized groups and personal connections over brand pages for years. A recommendation in a local group reaches more relevant people than most page posts, and it carries the trust a brand post never will.
The rule is participation rather than promotion. Groups have moderators, most have explicit rules about self-promotion, and a business that only appears to advertise gets removed. One that answers questions usefully becomes the obvious recommendation when someone asks.
Running your own group is a larger commitment than it appears and only works where there is a genuine reason for people to gather. A group created because a business wanted one produces an empty space that reads worse than not having it.
What is worth doing in almost every case is knowing which local groups discuss your category, and having a real person from the business participating in them honestly, identified as who they are.
The information someone checking you out is looking for, current, and easy to find — because verification is most of what a business Page now does.
That means the basics being right: hours including exceptions, address, phone number, and a link that goes somewhere useful. Those are checked more often than any post is read, and they are frequently out of date.
The About section should say what the business does in plain language rather than repeating a marketing line. Someone reading it is trying to establish whether you do the thing they need.
Reviews and recommendations need answering. A Page with unanswered negative recommendations is visible to everyone considering you, and a measured reply is read by far more people than the original complaint.
And the pinned post controls the first impression. Whatever is most useful to a first-time visitor belongs there, rather than whatever happened to be published most recently.
Directly, in ways that catch businesses out. The Page is the identity ads run under, and its state affects how people respond to them.
Someone who sees an ad and wants to check whether the business is real will click the Page name. An abandoned Page with a two-year-old post and unanswered comments answers that question badly, and it undoes the ad.
The Page also accumulates the comments on your ads, which are public and which most advertisers never read. Ad comment sections routinely contain questions, complaints and occasionally competitors, all visible to everyone the ad reaches.
Moderating those is real work with real return. An unanswered objection in the comments is read by people deciding whether to click, and a helpful reply is the cheapest persuasion available in the channel.
And the Page carries the account's standing. Policy issues, restrictions and quality signals attach to the identity running the ads, which makes the Page an advertising asset rather than an organic one.







For years, Facebook advertising expertise meant knowing the audience tools. Which interests to stack, how to layer behaviors, where the overlap was. The person who understood Ads Manager best generally got the best results, and that made the job legible.
Most of that skill has been automated away. Meta's delivery system now finds the right person from the creative more reliably than a hand-built interest stack does, and the privacy changes made narrow audiences less dependable to optimize within anyway.
What is left is less comfortable to sell. Performance now depends on the quality of the signal you feed the system and on producing a continuous flow of genuinely different creative. Neither is a monthly optimization task; both are ongoing production work.
Accounts that stall almost never stall because they were configured badly. They stall because nobody had anything new to say and the same three ads kept running.
Give us access to your Business Manager. We will check event match quality, audience structure and creative coverage, and reconcile what Meta reports against what you actually sold.

In practice, Meta ads management — pixel and Conversions API setup, event configuration, audience architecture, campaign structure, creative testing and reconciled reporting. Organic Page work is included but is a credibility exercise rather than a distribution channel.