Should you optimize for leads or purchases on Meta? If you sell something expensive and you can send real sales back to Meta, optimise for purchases, or for the event closest to a sale that you can track reliably. Lead optimisation finds people who fill in forms. Purchase optimisation finds people who pay. Those are different people, and on high value services the gap is large. Lead optimisation still has a place, mainly when you are new, low on data or selling something cheap with a short decision.
Why does the optimisation event matter so much?
Because it tells Meta who to look for. Meta’s explanation of how ad auctions work says each auction weighs your bid, the estimated action rate and ad quality. The estimated action rate is the probability that a given person takes the action you optimise for. Choose leads, and Meta hunts for people who are likely to submit a form. Choose purchases, and it hunts for people likely to pay.
Some people fill in every form they see. They are cheap to reach and they will flood your sheet. Meta is doing exactly what you asked. The mistake is asking for the wrong thing.
What happens when you optimize for leads or purchases on a high value service?
Here is the clearest example I have. A multi specialty clinic in Dubai came to me with high spend, plenty of messages and almost no return. On paper the account was busy. In the clinic, the calendar was not. The campaigns were rewarding volume of enquiries rather than people who booked and showed up.
We rebuilt the account around conversion focused campaigns, split the budget by department so each service had its own clean signal, and measured bookings rather than messages. Booked appointments went from 52 in the first month to 70 in the second and 93 in the third. Over those three months the clinic produced AED 322,500 in estimated revenue at a 4x return, and we did it while reducing the overall ad budget.
Nothing about the creative was magic. The change was telling the system what a good customer looks like and counting only what mattered.
When is lead optimisation the right choice?
- You have no purchase data yet. A brand new pixel or a business that closes sales offline with no way to report them back cannot feed purchase optimisation. Start with leads and build the tracking.
- Your offer is cheap and quick to decide. A free consultation or low price trial with a short sales cycle often converts fine from lead optimised traffic.
- Your budget cannot produce enough purchases. Meta needs enough conversion events to learn. If purchases are rare, optimise for a deeper event that happens more often, such as a booking, rather than a raw form fill.
On the last point, the middle ground is usually best. A confirmed booking, a qualified lead marked in your CRM, or a deposit paid are all closer to revenue than a form fill and more frequent than a final sale. Send that event to Meta and optimise for it.
How do you switch without losing results?
Do not switch the live campaign in place. Build a new campaign optimised for the deeper event, give it part of the budget, and let it run long enough to learn before you judge it. Expect cost per result to look worse at first, because you are paying for a rarer, more valuable action. Judge both campaigns on the same business number: bookings, sales or revenue from your own records.
Make sure the event fires reliably first. If sales happen on WhatsApp or at the front desk, you need a way to report them back, or purchase optimisation will learn from a fraction of the truth. I explain that setup in how to track WhatsApp leads from Facebook ads.
This is also where AI helps in an unglamorous way: matching leads to bookings, flagging junk and sending outcomes back on a schedule. That is the practical side of AI-powered marketing. If you want help deciding which event your account should chase, it is a standard first fix in my work as a digital marketing consultant in Dubai.
FAQ: optimize for leads or purchases
Why are my Meta leads cheap but never buy?
Because the campaign is optimised to find form fillers, not buyers. The causes and fixes are in why Facebook lead quality is so bad.
Can I optimize for leads or purchases in the same campaign?
No. Each ad set optimises for one event. Run them as separate campaigns and compare them on your own sales data.
How long should I test purchase optimisation?
Long enough to leave the learning phase and see real sales, usually a few weeks on a small budget. Judging it after three days rewards the cheaper, worse option.
What if I sell on WhatsApp only?
Then your purchase event has to come from the chat or your CRM. Until it does, optimise for the best signal you can track and measure bookings yourself.
I share one complete workflow a week from live accounts in The AI-Powered Marketer newsletter.