DM ads on Instagram are one of the best and one of the absolute worst funnels you can possibly run
Both things are true at the same time, and most people only find out about the second one after they have already built their whole business on top of it.
4 acts18 sections16 min read
ACT I
Why Your DM Ads Fill Up With The Wrong People
Before strengthening anything, expose the flaws. You cannot fix a funnel whose failure mode you have never named.
THE HONEST VERSION
The best and the worst funnel you can run
DM ads are phenomenal. There are operators cracking million-dollar months with them. (Individual results, not typical.) They are also very scary to run:
They are unstable. Performance moves in ways you did not cause and cannot always explain.
They are easily manipulated. Someone can pay people to get your ability to DM disabled.
You can break rules without knowing it. So can your setters. Either one can get your DM ability shut off.
It is a very easy-to-rug-pull kind of funnel. Some of the risk is outside your control, some comes from rules you or your setters break without realizing, and either way Instagram gives you very little protection from the fallout. What follows is how to strengthen performance anyway.
MINDSET
Fear is a signal, not a stop sign
A lot of people misread fear. It should not inhibit action; it should encourage it. When you are fearful about something, that is an opportunity to strengthen a weak area. The useful move is to know what causes the fear before it shows up in your chest, so you can attack it and stop it happening at all.
Name the flaw early and it is a checklist. Name it late and it is a crisis.
FLAW #1
Two things control your targeting
Flaw number one is pixel conditioning, and most people treat it as a near impossibility with DM ads.
Two dials decide who sees the ad. Only one keeps moving on its own.
Messaging is who the ads are most likely to reach at first. Before the standard event returns anything, messaging controls targeting, so you can run classic “who it's for and who it's not for” messaging with creative that is genuinely aligned to your ideal buyer.
Pixel conditioning takes over once the standard event starts reporting. If you optimize around message, whether under an engagement campaign or a lead campaign optimized for people messaging the page, you are going to have a bad time.
THE DONKEYS
Why the wrong people always get in
Picture content made for people doing million-dollar months. Ideally everyone watching is doing at least $100K a month, but when you talk at that level, everyone below it pays attention too. Even though you are talking to the people you want, a bunch of donkeys will always come anyway.
All it takes is one donkey to kick the door in.
Your standard event might start out conditioned on the right people. One gets in, the door stays open, and the rest are already coming. There are far more donkeys out there than right people, and the pixel does not know the difference. It just learns.
THE PATTERN
The wave, and then the decay
Once the wrong people are in the results column, you start moving in waves: the right people half the time, rubbish the other half.
Shape only. Every reset buys back less than the one before it.
The wave is the good scenario. If it holds, you can endure it and scale through it, as long as connection, booking, show and close rates and AOV are good. Usually it does not hold: you start high, dip, pull back up a little, then it keeps going down. You can reset a campaign or an ad set to get back to the first pocket, but after enough conditioning, every relaunch has a higher chance of getting donkeys.
That is the big downside of optimizing around message. The reset stops working.
THE TRICKS
The tricks do not escape it
Instead of a message-optimized campaign, people optimize around engagement, tell people to comment, or run follow campaigns. They all have the same issue: the pixel does not care which cheap action you optimized for, only who took it.
Three ways in, one hole at the bottom of all of them.
Brace yourself: the fix will probably raise the cost Ads Manager shows you. It replaces a cheap cost per DM with your real cost per qualified prospect.
ACT II
The Number You Have Been Hiding From Yourself
The fix makes a number on your screen bigger. If you do not understand this part, you will turn the fix off.
MOVE THE SLIDERS
Comparison bias and the hallucinated number
You have a bunch of bad leads in your results column. You see 50 people technically sent you a DM, at $10 a result. But only 10 of them are qualified.
Total spend, which never moves: $500
What the column says50 at $10The number you have been reporting on
What you actually paid10 at $50The number that was always true
Now a fix makes your results column say 10 instead of 50, and you panic: “my cost is $50 now.” No. It has been $50 all along. That is the type of person you want, not the other 40.
The cheap number was never real. Comparison bias will make you an idiot.
THE TRAINING PROBLEM
What 80% unqualified teaches the algorithm
50 people report in and 40 are bad: 80% of the leads in your results column are donkeys. That is not just a reporting problem, it is a training problem. The algorithm now has a much higher chance of finding you more donkeys, because that is what you told it to find.
Every unqualified DM in that column is a vote.
So the question is simple: how do you get only qualified people to report into the results column? There are two ways, in the next two acts.
STRAIGHT TALK
Before the fix, the actual recommendation
DM ads are scary. Among operators who run them, the businesses are some of the most unstable you will ever see, and the volatility they endure is insane.
Yes, you can do it, and the two paths below are likely to get you better results than what you have now. But point blank: something more stable, consistent and predictable is the better choice. This is one of the scariest things you could run; here is how to make it a little more stable anyway.
If a funnel needs two workarounds before it behaves, ask what it is costing you to keep it.
ACT III
Path One: The Quiz Funnel
Separate people out before they ever touch anything that matters.
THE FIRST WAY
Brokie bait
The first way is a small quiz funnel with one to two qualifiers, no more. We call it brokie bait, because that is exactly what it is for.
We do not want the qualified coming anywhere near the unqualified, or the other way round. Above all, we do not want the unqualified touching the results column.
The questions are not there to inform you. They are there to turn people away.
THE BUILD
From question one to the DM
The ad goes to a very simple landing page, essentially just one question.
Question one forwards to question two.
Answered the right way, conditional logic pushes them to a third page.
That page has a ManyChat button that sends them back into Instagram DMs, with the same trigger-word prompt they would get clicking straight from an ad.
Use a trigger word specific to this funnel, so you know the conversation came from ads and not organic. Do not be lazy about this one.
Then the standard DM funnel: setter conversation, booked call, show, back-end selling systems, close.
Push AOV as high as humanly possible, so you make something real out of it.
Further along, you might run an AI tool for the conversation instead of setters. That is not covered here.
THE KEY
The most important part: where the pixel fires
On that third page, the one people only reach after answering the qualifiers correctly, add a web pixel standard event that fires. Use a lead-based event: Schedule, Lead or Submit Application.
Keep one thing straight: campaign type and standard event are two different settings. Path two runs a Lead campaign while optimizing for the Schedule event. Same word, different dial.
That is the whole trick. The unqualified never reach that page, so they never fire the event, never enter the results column, and never teach the algorithm anything.
Move the event behind the qualifier and the pixel learns from qualified prospects instead of everyone who clicks.
PRO TIP
How much of the audience each event can reach
Of everyone available to target, only a slice is actually reachable for the standard event you optimize around.
The audience you can buy is a fraction of the audience you can target.
Say you have a million-person audience. Select Lead and you will not reach 900,000 of them; you reach roughly 100,000 likely converters before frequency gets jacked up. So choose the lead-based event that best represents this funnel's qualified action and fire it on the post-qualifier page. If people answer honestly, you get a dramatically better ratio.
Will some people lie? Yes. It matters less than you think, and the next act shows why.
ACT IV
Path Two: The Scheduler Link
Same destination, no quiz funnel. This one is genuinely pro level.
THE SETUP
Lead campaign, schedule ad set
Campaign type
Lead
Ad set optimization
Schedule
What do you actually want out of DM ads? Not just people who DM you: qualified people who DM you and book a call. Optimize for the thing you actually want, not the thing that is easiest to get. Schedule reaches its own slice of the audience, separate from what Lead would reach.
THE TRICK
An ad that says “DM us”
Pay close attention, because this is where people get lost.
The ad's destination URL is the scheduler page your setters push people to.
But the creative is entirely about DMing the business a specific word.
People see it and DM you. The conversation happens with your setter, who sends them that same scheduler URL.
They book the call and are officially locked in.
An ad that says DM us, whose click destination is the scheduler.
THE LEAK
Where the pixel fires, and what slips through
When they book, they land on a confirmation page. That is where the Schedule standard event fires, and you can use the Conversions API there too. It pushes back into your results column, so what you see is a cost per qualified call, and pixel conditioning gets heavily biased toward people who actually qualify and book.
The leak: occasionally someone ignores the “DM us”, clicks the link, books anyway and turns out unqualified. Yes, absolutely. But will it be the majority, the way message-optimized ads have been? No.
SWITCH THE PATH
The same funnel, with the pixel moved
Both routes end at a booked call. What changes is where the pixel is allowed to fire, and that decides who your ads chase next.
SWITCH THE STATE
Both paths do the same job: they flip the ratio
QualifiedUnqualified
Even at the lower end, 70% qualified and 30% not, that still dramatically increases the chance your ads keep going after the right people instead of a pile of donkeys. Get the junk out of the punch bowl, and the juice gets pure.
You are not buying cheaper DMs. You are buying a pixel that knows who your buyer is.
THE PLAYBOOK LIBRARY
Keep going: fix the next thing
9 free playbooks on the problems we fix for clients.