Detailed targeting for cold audiences has largely been retired. The algorithm now picks the audience, which means your creative is the targeting. Reporting shows brands testing 20 or more new ads a month achieve around 65% higher return than those testing fewer than ten. Most accounts ship three.
A Facebook ads agency plans, produces and manages social media advertising across Facebook, Instagram and the wider Meta placement network. In 2026 the balance of that work has shifted decisively: campaign structure and audience building have been substantially automated, while creative production, testing volume and measurement have become the levers that actually decide performance.
For roughly a decade, skill in this channel meant knowing how to build audiences. You layered interests, excluded segments, split-tested demographics, and the person who could assemble the cleverest audience won. That era is essentially over. Detailed targeting for cold audiences has been progressively deprecated, broad targeting now routinely outperforms narrow segments, and the algorithm decides who sees what.
What it decides with is your creative. Each ad you upload is effectively a hypothesis about who might respond, and the system finds the people it thinks will. Give it three ads and it explores three narrow possibilities. Give it thirty and it maps a far larger space. This is why creative volume has become the single strongest predictor of account performance, and why an agency that produces three static images a month cannot compete with one producing thirty varied concepts — regardless of how skilled either is at the settings.
That is a genuinely uncomfortable shift for agencies, because a Facebook marketing strategy built on creative production is harder and more expensive than one built on adjusting targeting. It is worth knowing which kind of agency you are hiring.
This is what a creative testing cycle actually looks like from the algorithm's side. Every ad gets a small share of budget, most are judged and set aside within days, and spend concentrates on the handful that clear the bar. Press run and watch a batch get sorted.
An illustration of the testing cycle, not a simulation of any specific account. Win rates vary enormously by offer, vertical and creative standard. The shape is what matters: a small share of concepts carries almost all the performance, which is why testing fewer than ten a month leaves you dependent on luck.
If roughly one in ten concepts is a genuine winner, then producing three a month means most months you produce none. You then run the least-bad of the three until it fatigues — which reporting suggests happens within five to ten days — and performance decays for the remaining three weeks while everyone wonders what changed in the algorithm.
Nothing changed in the algorithm. The account simply ran out of things to say. This is the most common pattern we inherit, and it is a production problem wearing the costume of a media buying problem.
Creative volume is not a matter of enthusiasm. Every ad needs enough impressions to be judged fairly, and your budget divided by that threshold is a hard ceiling on how many concepts you can meaningfully test in a month. Testing more than the ceiling allows produces noise rather than learning.
A planning model. It assumes roughly a tenth of concepts prove worth scaling and that testing budget is a portion of total spend rather than all of it. The impressions-to-judge figure is the honest variable — a click-through test needs far fewer than a purchase test, so set it to match what you are actually optimizing toward.
Not twenty color variations of the same image. Twenty different concepts — different hooks, different formats, different problems addressed, different people speaking. Variations of one idea test the same hypothesis twenty times and teach you almost nothing.
Reporting consistently finds user-generated and creator-style content outperforming polished brand creative on this platform, by meaningful margins on both click-through and cost per acquisition. Authenticity signals trust more efficiently than production value does, which is genuinely good news for budgets. We produce both through video production, and the mix matters more than the polish.
You will see it on nearly every agency page selling this service: automated campaigns deliver 32% lower cost per acquisition than manual ones. The figure is real. It is also being used to mean something other than what it measures, and the difference matters if you are deciding where to put money.
The 32% figure describes cost per incremental conversion when automated campaigns run alongside manual ones — not a like-for-like comparison of one against the other. Independent analysis of the like-for-like question puts the genuine improvement nearer 12% to 22%, with third-party data showing automated shopping campaigns averaging around 4.5x return against 3.7x for manual.
Twelve to twenty-two percent is a good result. It is simply not thirty-two, and an agency that quotes the bigger number either has not read the source or is hoping you will not.
The caveat that matters more: the advantage is data-dependent. It is widest for accounts spending above roughly $10,000 a month and thin for accounts under about $2,000, because the system needs conversion volume before it can optimize. Below that threshold, automation is being asked to learn from too little signal, and manual structure frequently does better.
So the honest answer to "should I use automated campaigns" is: probably, if you have the spend and the conversion volume to feed them, and probably not as your only campaign regardless. We run both, and which one we lead with depends on your numbers rather than on which is currently fashionable.
Most accounts leave placement selection for their social media ads entirely automatic, which is usually correct and occasionally expensive. Knowing the underlying economics tells you whether the algorithm's choices are serving you.
Placement CPM ranges compiled from 2026 reporting including Revealbot placement-level data and Varos cross-account analysis. Directional rather than a rate card — your costs depend on vertical, geography and creative format.
Reels clicks run materially cheaper than Feed. Reported cost per click sits around 26% lower, because inventory has expanded faster than advertiser demand. That gap will close as more advertisers notice it, which is a reason to exploit it now rather than later.
Marketplace is the interesting one. At roughly $6.90 CPM with a click-through rate near 1.68%, it carries genuine buying intent at close to half the cost of Feed. People on Marketplace are there to purchase something, which is not true of most placements.
Cheapest is not best. Audience Network delivers the lowest CPMs and frequently the lowest-quality traffic, with inflated click counts that flatter a report and produce nothing. This is the same lesson as the true-cost argument on our programmatic advertising page: a CPM describes what you paid, not what you received.
Alongside running paid campaigns, we own and administer a network of private and public local buy, sell and trade groups across the United States — the community spaces where neighbors ask each other for recommendations every day. That is an unusual position for a marketing agency, and it changes what we can offer local businesses.
It matters for two reasons. The first is practical: we can place local businesses in front of genuinely local, genuinely engaged audiences inside communities we run, with competition limited per category so you are not sitting beside four rivals. Details, ground rules and pricing are on Social My Business.
The second is that administering communities teaches you things running ads never will. We see what actually gets engagement in a local group and what gets a member reported. We know which posts read as helpful and which read as advertising, because we moderate the difference daily. That informs the paid work too — the creative that performs in a community is frequently the creative that performs in a feed.
A necessary note: community groups have their own rules, set by their administrators, and platform policies apply to everything. We operate inside both. Any agency promising to bypass either is describing a short-lived arrangement, and the fastest way to lose access to a community is to treat it as ad inventory.
We look at what you have been running, how often it changes, and where the budget is going. Then we tell you whether your problem is the account or the creative pipeline feeding it.
Get my free auditSocial media advertising is demand creation rather than demand capture. It suits businesses that can explain why someone should want the thing, and suits pure comparison shopping less than search does.
The natural home. Creator content and product video carry most of the performance.
🍽RestaurantsLocal radius, strong offer, food that photographs well. One of the better fits.
🏠Real EstateListing video and local reach. Special ad category rules apply here.
🦷DentalNew patient offers and before-and-after content, within advertising policy limits.
🏥HealthcareRestricted category with real targeting limits. Educational creative works best.
💻SaaSWorks for problem-aware audiences. Optimize toward paid conversion, not signups.
⚖Law FirmsSpecial ad category constraints. Awareness and education rather than direct response.
🏢B2BViable with strong creative and long nurture. Rarely beats search on intent.
Far more creative production and far less audience building than it used to. Detailed targeting for cold audiences has been progressively deprecated and broad targeting now routinely outperforms narrow segments, so the algorithm decides who sees your ads. What it decides with is your creative, which makes concept production, testing volume and measurement the levers that actually move performance. A Facebook ads agency should be running a repeatable creative pipeline, verifying conversion tracking, keeping structure simple enough for the system to learn, and reporting on cost per customer rather than cost per click.
More than you currently are, almost certainly. Reporting indicates brands testing 20 or more new ads a month achieve around 65% higher return than those testing fewer than ten. Any Facebook marketing strategy has to reckon with the arithmetic rather than enthusiasm: if roughly one concept in ten proves worth scaling, producing three a month means most months produce nothing, and you end up running the least-bad option until it fatigues. Your practical ceiling is set by budget, since each ad needs enough impressions to be judged fairly — the volume calculator on this page works out yours.
Not as that figure is usually presented. The 32% number describes cost per incremental conversion when automated campaigns run alongside manual ones, rather than a like-for-like comparison. Independent analysis of the like-for-like question puts the genuine improvement nearer 12% to 22%, with third-party data showing automated shopping campaigns averaging around 4.5x return against 3.7x manual. That is still a good result. The more important caveat is that the advantage is data-dependent: widest above roughly $10,000 a month in spend and thin below about $2,000, because the system needs conversion volume before it can optimize.
Usually creative fatigue rather than an algorithm change. Reporting suggests audiences tire of an ad within five to ten days, after which performance decays regardless of how well it started. If your replacement pipeline produces a few concepts a month, you spend most of each month running a decaying ad while a new one is made. Other common causes are seasonal auction pressure, tracking degradation, and prospecting being blended with retargeting so a decline in the former is masked by the latter. We check those in that order.
2026 reporting puts the cross-industry median CPM near $15.06 and median cost per acquisition around $39, with average cost per click on Facebook near $1.72, up roughly 11% year over year. Instagram Reels runs materially cheaper at around $1.28 per click. Costs also swing seasonally — January reportedly sits well below average while November runs sharply above during holiday competition. Management fees are separate, and we price them to the work rather than as a percentage of ad spend.
For social media ads, Reels and Marketplace are the two worth knowing about. Reels clicks run around 26% cheaper than Feed because inventory has grown faster than advertiser demand. Marketplace sits near $6.90 CPM with a click-through rate around 1.68%, which is underpriced relative to the buying intent people bring to it. Audience Network offers the cheapest impressions and frequently the worst traffic quality, which is a reminder that a CPM describes what you paid rather than what you received. In most accounts automatic placement is the right default, but knowing the economics tells you whether the algorithm's choices are serving you.
Both, with more of the latter than most brands are comfortable with. Reporting consistently finds user-generated and creator-style content outperforming polished brand creative on this platform, by meaningful margins on click-through and cost per acquisition. Authenticity appears to signal trust more efficiently than production value does. That is good news for budgets, since it means the constraint is ideas rather than production spend. We produce both through video production and the mix matters more than the polish.
Yes. We own and administer a network of private and public local buy, sell and trade communities across the United States, which lets us place local businesses in front of genuinely local audiences with competition limited per category. Details and pricing are on Social My Business. Community groups have their own rules set by their administrators and platform policies apply throughout, so this works alongside paid advertising rather than replacing it.
Yes, and you should require this of any agency. You retain ownership of your business account, your ad account, your pixel and your audience data throughout the relationship and after it ends. An agency running your advertising inside their own account is holding assets that are genuinely yours, and you discover the consequences at the worst possible moment. Ask the question before signing anything, with us or with anyone else.
They solve opposite problems. Google Ads captures demand that already exists — someone is searching for what you sell right now. Meta creates demand among people who were not looking, which means lower intent, lower cost per click and a much heavier dependence on creative. Most businesses that can afford both should run both, with search capturing the demand that social creates. Running only one is usually a budget constraint rather than a strategic choice, and if it is a constraint, search is the safer first channel.
Tell us what you have been running. We come back within one business day with what we would change and why.
Your information is never sold or shared. We respond within one business day.
