Dr. Matt Shiver

Never do this with a winning ad.

Most people find a winner and duplicate it across five or six ad sets to test audiences. That worked in 2015. It doesn't now.

Break the winner down instead. Figure out whether it was the hook, the offer, or the format that did the work.

If it's the offer, that talking head video is still only reaching a slice of your market. Rebuild the same offer as a static, a carousel, B-roll. Same messaging, different packaging.

Creative diversity scales. Duplication doesn't.

Big spenders can filter what Facebook counts as a result. Route booked calls through an automation, enrich and qualify with AI, and only send qualified leads back as conversions.

I don't think the traditional funnel describes how people actually buy anymore.

The old model says top of funnel, then middle, then bottom, in order.

What I actually see: someone hits a direct offer ad first, goes to YouTube, watches a how-to, watches something that shifts a belief, comes back to Instagram for a few reels, then applies.

They ran it backwards.

It's less a funnel and more a sphere of touch points consumed in whatever order they choose. One ad isn't doing the heavy lifting anymore.

Adding a CTA to every post cut our lead volume by more than half, from 30/day down to 12.

Constant pitching feels like constant pitching. Some posts should just build trust, no ask attached.

These are the placements I keep on when I want quality leads.

In the ad set, go to placements, then show more settings.

Platforms: Facebook and Instagram only. Audience Network, Messenger and WhatsApp off. Threads off if quality is the priority.

In placement controls, turn off in-stream ads for reels, search results, and Facebook Reels. Lead quality on Facebook Reels has been poor for me, though it's worth testing on your own account.

In the feed section, turn off Facebook notifications.

Then scroll down and turn off "allow limited spending to excluded placements." If you excluded a placement, you meant it.

100,000 views and zero calls. 3,000 views and 3 calls, from a different post two days earlier.

Reach doesn't drive results on its own. Repeating your core message does.

Every winning ad dies, and it's not because the ad got worse.

When you launch, Facebook serves the hottest people first. Followers, people who've seen your other ads. Once that bucket is exhausted it moves to warm, then to cold.

That's why you can open at a 4-5x ROAS and watch it drift toward 1-2x. You ran out of easy people.

The fix is refreshing creative, speaking to problem-aware buyers, and knowing your lifetime value well enough that 1-2x on cold still works.

B2C funnel choice comes down to two questions: how big is your budget, and does your offer need explaining?

Education-heavy + mass market = webinar funnel. Simple + mass market = low ticket funnel, if you can test for 3-4 months.

Here's the script I use for direct offer ads.

Open by calling out your avatar and naming the problem or goal. Only speak to that one person.

Then the promise, early. Put it up front in case they don't make it to the end.

Then qualify. Add some emotion and filter for only the people you actually want coming through.

Then social proof, so they see it worked for someone like them.

Then the CTA, and restate the risk reversal you already made at the top.

That whole thing runs about 60 seconds.

For colder traffic, add a minute in the middle to introduce yourself and your method and why it's different.

The 80/20 rule for scaling ads: 80% of your effort should go toward doing more of what's already proven. Only 20% toward testing something brand new.

Test offer → test pain points → go wide on creative formats. In that order.