ARTICLE SUMMARY

We deliberately left placements wide open on instant form lead campaigns across multiple accounts to find out where the budget actually went. Audience Network and Messenger received close to nothing. The more useful second finding was that non-feed placements frequently beat feed on cost per lead.

"Turn off Audience Network" is one of the most repeated pieces of Facebook advice on the internet. It has been repeated for about a decade. It gets recommended in every audit template, every checklist, every teardown video.

We stopped doing it, on purpose, to find out whether it mattered. Across multiple lead generation accounts we left placements fully open on instant form lead campaigns and watched where the budget actually went.

It went almost nowhere near Audience Network.

What we did and what we saw

The test was not clever. We simply stopped intervening.

On instant form lead campaigns, across several client accounts running through 2026, we set Advantage+ placements and left every placement eligible, including Audience Network and the Messenger placements. Then we pulled the placement breakdown and looked at spend rather than at impressions or reported results.

Audience Network and Messenger received close to zero delivery. Not a small amount that we then wanted to remove. Effectively nothing, consistently, across accounts and over months. Meta's own auction simply was not routing instant form lead budget to those surfaces.

There was a second finding that we did not expect, and it is the more useful one.

Non-feed placements frequently beat feed on cost per lead. Reels, Stories and the various in-stream and explore surfaces regularly came in cheaper per lead than the main Facebook and Instagram feeds in the accounts where we looked. Not always, not by a fixed margin, but often enough that a habit of restricting to feed would have been costing money.

Near zero Share of instant form lead budget our accounts routed to Audience Network
10 yrs Roughly how long “turn off Audience Network” has been repeated as standard advice
2 min Time it takes to check the placement breakdown in your own account

Why the auction behaves this way

This is worth understanding rather than just filing as a fact, because the mechanism tells you when the finding will and will not hold.

An instant form lead ad keeps the entire interaction inside Meta's own surfaces. The user taps, a native form slides up already populated from their profile, they confirm and submit. There is no site visit and no external page load.

Audience Network is different in kind. It is inventory on third party apps and mobile websites. Meta can serve an ad there, but the native instant form experience is not the same object in that context, and Meta's delivery system is optimising for a conversion that it predicts poorly on that inventory. Faced with an ad set optimising for on-Facebook lead submissions, the auction routes budget to the surfaces where those submissions actually happen.

Messenger placements have a related issue. The Messenger inbox and Messenger stories placements are built around click-to-message ad experiences. An instant form campaign is not that, so the fit is weak and the predicted conversion rate is low, so the bid loses.

In short, Meta is already doing the thing the checklist tells you to do, and it is doing it per impression opportunity with far more information than you have.

What this means for placement editing

Here is the uncomfortable conclusion, and we will say it plainly rather than hedge it.

On instant form lead campaigns, manually excluding Audience Network is mostly ceremony. It removes something the auction was not buying anyway. It makes the setup screenshot look tidier. It does not move cost per lead in a way we have been able to see.

Worse, the habit has a real cost, because people rarely stop at Audience Network. The same checklist mentality goes on to strip out Reels, Stories, Explore, in-stream and search, on the grounds that those are "low quality." In our accounts that would have removed several of the cheapest sources of leads we had.

There are three things placement editing is still genuinely for:

Brand safety requirements. If a client has a contractual or regulatory reason to stay off third party app inventory, exclude it. That is a business constraint, not an optimisation, and it is fine.

Creative that only works in one aspect ratio. If you only have a 1:1 static and no vertical asset, a Stories placement will render it with letterboxing or crop it badly. The honest fix is producing the vertical asset, not excluding the placement, but excluding is a legitimate stopgap. If you are short on assets, that is an argument for the systematic approach in AI real estate ad creative testing rather than for narrowing where the ads can run.

Genuine measured harm. If you pull the placement breakdown and one surface has real spend and demonstrably worse downstream outcomes over a period long enough to mean something, act on it. That is different from acting on a rule you read.

Note the ordering. Look at the breakdown first, then decide. The default in most audits is the reverse.

KEY TAKEAWAY

Look at the placement breakdown first, then decide. The default in most audits is the reverse, and the habit rarely stops at Audience Network. It goes on to strip out Reels, Stories and Explore, which in our accounts would have removed several of the cheapest sources of leads we had.

The trap in judging placements by reported cost per lead

One caution on the "non-feed beat feed" finding, because it would be easy to over-read.

Cost per lead by placement is a shallow metric. A placement can produce cheap form fills that convert badly downstream, and the placement report will not tell you that. We have written about why the reported number and the real number diverge in what a manufactured home dealer lead actually costs.

The way to check is to carry the placement breakout through to your CRM outcomes rather than stopping at the platform. That requires the lead to arrive tagged, which requires the attribution plumbing to be in place before you need it. If your leads land in the CRM with no ad or placement information attached, you cannot answer this question about your own account and you are back to following somebody else's checklist.

The broader point about default settings

There is a general lesson here that applies well beyond Audience Network.

A lot of paid social advice is fossilised. It was true about a version of the platform that existed when the advice was written, it was repeated until it became a convention, and nobody re-ran the test because the advice was free to follow and appeared harmless.

Meta's auction and delivery systems changed substantially over that period. Placement controls in particular went from a meaningful lever to something the system largely handles, and the accounts we see that are still running tight manual placement lists are usually carrying a set of restrictions somebody applied in 2021 and never revisited.

The general habit worth building: for any setting you change out of habit, go and look at what it would have done if you had left it alone. Most of the time the answer is "nothing," which is a useful thing to know because it frees up attention for the levers that do move numbers, like offer, creative and geography. That is the same conclusion we reached about most of the 2021-era tactics still circulating, in Facebook ads in 2026.

For any setting you change out of habit, go and look at what it would have done if you had left it alone. Most of the time the answer is nothing, which frees up attention for the levers that do move numbers.

Caveats, stated clearly

This is our own measurement, across our own client accounts, on instant form lead campaigns, observed through 2026. It is not a controlled experiment and we are not publishing account-level numbers.

Three boundaries on it:

If you want to know what your account does, the check takes about two minutes. Open any instant form lead campaign, switch the breakdown to placement, look at the spend column rather than the results column, and see for yourself.

Our methodology

We ran instant form lead campaigns with Advantage+ placements and no manual placement exclusions across multiple client ad accounts, then reviewed the placement breakdown by amount spent over rolling periods rather than by a single window. We compared cost per lead by placement within each account rather than pooling across accounts, because pooling across accounts with different budgets and offers produces an average that describes none of them. Where we could, we carried the placement breakout through to CRM stage outcomes rather than stopping at the platform's reported lead count.

Share this article

Frequently Asked Questions

Should I turn off Audience Network on Facebook lead ads?

On instant form lead campaigns, in our accounts, it makes very little difference because the auction routes almost no budget there anyway. Excluding it removes something that was not being bought. The exception is a genuine brand safety requirement, which is a business constraint rather than an optimisation.

Why does Audience Network get so little spend on lead ads?

Because an instant form keeps the whole interaction inside Meta's own surfaces, and Audience Network is inventory on third party apps and mobile websites. Meta's delivery system predicts that conversion poorly on that inventory, so the bid loses and the budget goes where the submissions actually happen.

Do Messenger placements work for instant form lead ads?

In our accounts they received close to nothing. Messenger inbox and Messenger stories placements are built around click to message experiences, and an instant form campaign is not that, so the fit is weak and the predicted conversion rate is low.

Are Reels and Stories placements low quality for lead generation?

That is the common assumption and it did not hold in our accounts. Reels, Stories and the in-stream and explore surfaces regularly came in cheaper per lead than the main feeds. Not always and not by a fixed margin, but often enough that restricting to feed would have been costing money.

When should I edit placements manually?

Three cases. A contractual or regulatory brand safety requirement. Creative that only exists in one aspect ratio, where excluding is a legitimate stopgap while you produce the vertical asset. And genuine measured harm, where a surface has real spend and demonstrably worse downstream outcomes over a meaningful period.

Is cost per lead by placement a reliable way to judge a placement?

Only partly. A placement can produce cheap form fills that convert badly downstream, and the placement report will not tell you that. To answer it properly you need the placement breakout carried through to CRM outcomes, which requires the leads to arrive tagged in the first place.

Keep reading

Want Someone to Read Your Account Instead of a Checklist?

We start every account takeover by looking at what the delivery system is actually doing, not by applying a template of settings somebody wrote in 2021. Book a session and we will walk your placement and spend breakdown with you.

GET YOUR FREE STRATEGY SESSION

Or call us: 512-877-5541