If you want to know what’s working in meta ads right now, here is the September 2026 answer from someone spending real money daily: lead quality beats lead volume, cutting spend is outperforming scaling on broken funnels, and follow-up speed is deciding more outcomes than creative. I run paid ads for 30+ beauty, wellness and clinic brands, so I see patterns across many accounts at once. These are my field notes for September. Not theory. What is actually converting, what quietly stopped, and what I changed on live accounts over the past quarter.
This is the September edition of a monthly series. The July field notes are here if you want to compare what changed.
What’s working in meta ads right now: the short version
Three shifts define this edition. First, judging campaigns on cost per booking instead of cost per lead is changing which campaigns survive. Second, the bravest move I made this quarter was telling a client to spend less, and it is working. Third, everything after the click, meaning reply speed, inbox placement and follow-up, is where accounts are winning or bleeding. Lo-fi talking videos still beat polished brand films on cost per lead, often by 30 to 50 percent, same as the last edition. That one is not news anymore. The rest is.
Is cheap lead volume still worth chasing?
No, and this is the biggest mindset shift on my accounts this quarter. A flood of cheap WhatsApp leads that never book costs you more than fewer, pricier leads that show up. I rebuilt lead capture on a clinic account that was drowning in unqualified WhatsApp messages: we moved the campaign to Meta lead forms with qualifying questions and cleaned the targeting. Lead cost went up. Lead quality went from a guessing game to roughly 70 percent qualified, and the front desk stopped wasting hours on people who were never going to buy. If you only track one metric right now, make it cost per booked appointment, not cost per lead.
When should you cut Meta ads spend instead of scaling?
When the leak is after the click. Recently I told one of my longest-running clients, a wellness brand, to reduce ad spend. Their landing page was converting at 8 percent, and pushing more budget through a page that leaks 92 out of every 100 visitors is just paying Meta to amplify the problem. We paused the scale-up, fixed the page first, and the account gets a compounding return on every dirham once traffic resumes at full volume. Whoever runs your marketing should be willing to make that call. If your agency only ever recommends spending more, that tells you whose revenue they are optimizing.
The exception: do not confuse a broken funnel with a campaign still in learning. Meta’s own guidance is to let ad sets exit the learning phase before judging them. Week-one numbers are noise. An 8 percent landing page after three months of data is signal.
Why are good campaigns still losing money after the click?
Because delivered is not the same as seen, and captured is not the same as called. Two audits from my own accounts this quarter. One: an email follow-up workflow showing 92 percent delivery with near-zero engagement. Delivery means a server accepted the message, not that it reached the inbox. The sequence was almost certainly landing in spam and promotions folders, which means the client was paying for leads and then going silent on them. Two: a clinic where form leads were quietly skipped by the follow-up automation while WhatsApp leads got instant replies. The leads they had already paid for were sitting uncalled. If your ads “aren’t working,” check your follow-up before you touch the campaign.
What should you fix in your tracking now?
Source tracking on conversation channels. On one account I reviewed, around 80 percent of leads arrived by WhatsApp with no UTM or source data attached. That means every scale-or-kill decision on those campaigns was a guess. You cannot scale what you cannot trace. Before you add budget anywhere, make sure a booking can be traced back to the campaign that produced it, through lead forms, tagged links or CRM automation. This is also where AI genuinely earns its keep: not writing another caption, but reading messy lead data faster than any junior media buyer. That is the difference between AI-powered marketing and AI-flavored content.
FAQ: what’s working in meta ads right now
What ad formats are converting best in September 2026?
Lo-fi talking-head video and single-claim static ads with real numbers are still beating polished production for cold audiences, by 30 to 50 percent on cost per lead across my accounts. The bigger gains right now are not in formats at all. They are in lead quality and follow-up.
Should I use WhatsApp campaigns or lead forms?
WhatsApp still dominates for GCC service businesses, but only with fast first replies and proper source tagging. If your WhatsApp volume is high and your booking rate is low, test Meta lead forms with qualifying questions. On my accounts that trade raised lead quality to roughly 70 percent qualified.
How much should I be spending on Meta ads?
Enough to exit the learning phase, split roughly 70 percent on proven angles and 30 percent on testing. I published real minimum budget numbers here. And sometimes the right answer is to spend less until the funnel deserves the traffic.
How do I keep up with what changes each month?
This series updates monthly from live accounts. The weekly version, with the workflows behind each fix, goes out in The AI-Powered Marketer newsletter, free.
The one thing to do this week
Pick your highest-spend campaign and trace one booking backwards: which ad, which form, how fast was the first reply, where did the follow-up land. If you cannot complete that trace, that is your leak, and no amount of new creative will fix it. That audit is what’s working in meta ads right now, more than any format or hack.