Open the spreadsheet your sales team keeps. Not the one in Ads Manager. The one with last month's inquiries in it, where somebody has gone down the list and marked which ones turned into work.
Forty names. Eleven of them worth calling back. Two that became jobs. One that is still deciding.
That grading took somebody an afternoon, and it is the most valuable marketing data your business produced all month. It is also, almost certainly, still sitting in that file. Meta has never seen it. It spent your budget last month, it will spend your budget again this month, and it is doing that with no idea which of the forty names it found were the good ones.
This is not a story about Meta being blind. It is a story about a door Meta left open that almost nobody walks through.
Your Ad Account and Your Sales Process Are Not Speaking
Most lead-gen accounts are optimizing on the wrong finish line, and the reason is structural rather than careless.
When somebody fills in your form, an event fires. Meta writes it down, counts it as a result, and goes looking for more people who look like the person who filled it in. That is the entire feedback loop. It is fast, it is reliable, and it stops precisely at the moment the lead arrives.
Everything that happens after that, the part you actually care about, is invisible. Whether the number was real. Whether they picked up. Whether they were in your service area, or had a budget, or were a competitor doing research, or wanted the one thing you do not sell. Your team learns all of that within a few days. Meta learns none of it, ever, and keeps optimizing toward whatever produced the cheapest form fill.
We have written before about why Meta has no idea what your leads are worth, and about the two settings that deal with it: value optimization, which needs 100 valued conversions with 5 distinct values inside 14 days and rules out most local businesses on volume alone, and value rules, which anybody can use today.
There is a third one. It did not make that article, and for a business that sells its time rather than a product, it is the most interesting of the three.
The Loop You Are Already Halfway Through Building
Meta calls it the Conversion Leads performance goal, and the idea is refreshingly blunt: stop guessing what a good lead looks like and tell it.
Here is the shape of the thing. Leads already travel in one direction, out of Meta and into wherever you keep them. Somebody fills in an instant form, and through an integration you set up once, that name lands in your CRM. That half of the loop exists in most accounts already.
The Conversion Leads goal builds the return leg. As each lead moves through your sales process, your CRM sends that movement back to Meta through the Conversions API. Not the money, necessarily. The stage. This one was qualified. This one booked a consultation. This one signed. Meta takes those outcomes, looks back at who those people were and which ads brought them in, and starts optimizing for the pattern that produced customers instead of the pattern that produced form fills.
One detail catches people who think they have already done this. If you run the Conversions API on your website today, for purchases or page views or anything else, that is a different integration. Meta treats it as a separate build, because the fields it expects are not the same ones your website sends and the records originate in your sales system. Having one does not give you the other. You are laying a second pipe, not reusing the first.
What ties the two halves together is an identifier. Every lead Meta sends you carries a Meta Lead ID, a 15 to 17 digit number. Store it against the record in your CRM, send it back with each stage update, and the match is exact. If you do not have it, Meta will work with customer parameters instead, a click ID or a phone number or an email, which functions but matches less reliably. Storing the lead ID is a ten-minute decision in your CRM setup that quietly determines how well the whole thing works.
Two Hundred Leads a Month Is the Door
This is where most articles about this feature get vague, so here are the actual requirements, from Meta's own integration documentation.
You need to be running Facebook or Instagram lead ads using instant forms. You need to generate at least 200 leads a month. You need to be able to upload data back at least once a day. The stage you optimize toward has to happen within 28 days of the lead arriving. And that stage has to convert at somewhere between 1% and 40%.
Worth pausing on the first number, because the internet has it wrong. Search this feature and you will find guide after guide repeating 250 leads a month. Meta's documentation says 200. That is a difference of fifty leads a month, which is exactly the width of the gap that decides whether a busy home-services company is eligible or not.
Two hundred leads a month is roughly seven a day. That is a real bar, and it will exclude plenty of businesses. But notice what it is counting. Not sales. Not valued purchases. Leads. A pest control company running a steady budget, a driving school, a mid-sized dental group, a roofing contractor in season: these businesses clear 200 leads a month comfortably while being nowhere near the 100 valued purchases in 14 days that value optimization demands. The feature that looked out of reach and the feature that is within reach are measuring completely different things.
The daily upload requirement is the one that quietly fails. It does not mean a developer sitting at a keyboard. It means the connection has to be automatic, because a sync that happens when somebody remembers is a sync that stops happening in about three weeks.
Meta Wants Your Stage to Fail Most of the Time
That 1% to 40% conversion rate is a window with two walls, and the top wall is the one people misread.
The floor makes intuitive sense. If your chosen stage converts at half a percent, then out of 200 leads you are handing Meta one success a month. There is nothing to learn from one example. The signal is too sparse to train anything.
The ceiling is less obvious and more interesting. Meta does not publish its reasoning here, but the arithmetic is not hard to reconstruct. If the stage you picked converts at 80%, you have not really selected anything. Almost everyone reaches it, which means reaching it does not distinguish a good lead from a bad one, which means telling Meta about it adds no information it did not already have from the form fill. A stage that nearly everybody passes is just the form fill wearing a different name.
So the useful range sits in the middle, and the practical reading is this: you are looking for a stage that most leads fail. That feels wrong the first time you hear it. It is the whole mechanism. The value of the signal comes from its selectivity.
Pick the Furthest Stage That Still Lands Inside the Month
Now the judgment call, and it is genuinely the part that decides whether this works for you.
The instinct is to optimize for the last thing: signed, paid, closed won. That is the outcome you care about, so that must be the right target. Often it is not, because of the 28-day rule.
Meta needs the stage to occur within 28 days of the lead arriving. If your average deal closes in six weeks, "closed won" is outside the window, and the events that do land inside it will be an unrepresentative sample of your fastest, easiest, probably smallest jobs. You will have trained the system to go and find more small fast jobs. That is the opposite of the reason you started.
So the rule is to pick the furthest-down stage that still reliably happens inside 28 days. For a business that closes quickly, that might genuinely be the sale. For a longer sales cycle it might be "quote issued" or "site visit booked" or "spoke to a decision maker." Each of those is a real filter and each one sits comfortably inside a month.
Run the numbers before you choose. Take last quarter's leads, pick a candidate stage, and answer two questions: what percentage reached it, and how long did that take on average. If the answer is 12% within nine days, you have your stage. If it is 60% within two days, it is too easy to be worth reporting. If it is 8% within 51 days, it is the right idea and the wrong stage, and you should look one step earlier in your funnel.
The Catch Nobody Wants to Talk About
Here is the awkward part, and we would rather name it than let you find it halfway through a setup.
The Conversion Leads goal works only with instant forms. Meta says so plainly: it is currently compatible with Facebook and Instagram lead ads. Not your landing page. Not a form on your own site.
That matters here because we have argued, in some detail, that instant forms get you cheap leads while a landing page gets you customers. The lower friction of a form that fills itself in is exactly what lets people submit without really meaning to.
So the single best tool for fixing instant-form lead quality requires you to use instant forms. That is a genuine tension and we are not going to pretend it resolves neatly.
What it does is change the shape of the decision. The case against instant forms was always about quality: you get volume, and a chunk of it is noise. The Conversion Leads goal attacks precisely that weakness, by teaching the delivery system which of those easy submissions turned into work. It does not make instant forms better at filtering. It makes Meta better at choosing who sees them.
The honest read is that instant forms plus a closed CRM loop is a stronger position than instant forms alone, and that a landing page with a serious form still beats both for businesses whose problem is qualification rather than volume. If you are already running instant forms and drowning in mediocre leads, this is your fix. If you moved to a landing page deliberately and it is working, this feature is not a reason to move back.
Nothing Happens for the First Month
Expectations, because this is not a switch.
Meta's own project timeline puts total time to value at roughly three to four weeks. The connection itself is the fast part: through an integration partner it is under a day of work, while a custom build by a developer is estimated at three to four weeks. Then Meta validates that your data is arriving correctly, which takes a day or two. Then there is a training period of two to four weeks while the system learns your funnel.
You can run Conversion Leads campaigns during that training window, but Meta is upfront that you will not see the full benefit until it finishes. Which means the first month is mostly the system watching your outcomes accumulate.
You also need three kinds of people available, and the one that gets forgotten is the middle one. Somebody from sales who genuinely understands how your funnel works and can define the stages. Somebody who knows your CRM well enough to add fields and build the flow. And a developer, but only if you are building it manually rather than going through a partner integration.
The stage definitions are the hard part, and they are not a technical problem. If your team uses "qualified" to mean four different things depending on who is typing, that ambiguity goes straight into the training data.
Who Should Close This Tab
Worth being direct about who this is not for, because the setup cost is real.
If you generate fewer than 200 leads a month, you do not qualify, and no amount of enthusiasm changes it. If you have no CRM, or your CRM is a shared inbox and a notebook, the prerequisite is not this feature. It is a CRM. If nothing in your sales process reliably happens within 28 days, the window does not fit your business and you will be training on a biased sample. And if you have no automated way to push data back daily, build that first or do not start.
If you are in that group, the earlier article's answer still stands: value rules carry no volume requirement at all and you can set one this afternoon.
For everybody else, the thing worth sitting with is that you are not being asked to generate any new information. You already know which leads were good. Somebody on your team wrote it down. The only question is whether that knowledge gets back to the system that decides who sees your ads tomorrow, or whether it stays in a spreadsheet.
We set these loops up for clients constantly and are happy to look at whether your account clears the bar. If you want a hand with it, that is what we do.
