How should I split my Facebook ads budget for lead generation?

Give every ad its own ad set and its own budget, then judge it after four days. Your daily spend doesn't change that rule. It only changes how many ads you can test at once, and how fast you can add new ones.

This is one of the topics people ask my AI bot, Revy, about most. Some are just starting out and don't know what they can afford. Others are spending a few hundred dollars a day and want the right structure. Here's how I split it at each level. It works the same on Meta's instant forms or a landing page with a video sales letter. I go through every level in this YouTube video.

$30 a day: one ad, four days

If you're just starting out, keep it simple. One campaign, one ad set at $30 a day, one ad. I wouldn't go lower than that.

Then wait four days. You need enough impressions to say with some certainty that the ad is good or bad. At this stage you're judging it on cost per lead and the number of leads.

After four days you do one of two things. If it's performing, let it run and watch for booked calls, appointments and sales. If it isn't, and most of the time it won't be, rotate a new ad in and run the same test again. That's all it is. I go through testing on a small daily budget in this video.

$50 a day: room for statics if you run a landing page

On lead forms, I'd keep it at one ad on $50 a day. Keep or kill it every four days.

If you're sending traffic to a landing page or a video sales letter, you have an option. Split it into two ad sets. Put $30 a day behind one video ad, and the other $20 behind up to three static image ads. You don't have to do this, but it lets you test the image ads that often work well in front of a sales video.

The process doesn't change. Every four days you rotate the losers out and let the winners run.

$100 a day: three ads judged against your cost per lead ceiling

This is for a business selling something worth $4,000 or more. A sale like that needs a booked call. I'd run three ad sets at $30 a day, each with one video ad. That's $90 a day. I'm not even using the full $100.

After four days, each ad gets judged against a cost per lead ceiling. If you've run ads before, that ceiling is already sitting in your ad account history or your CRM. For one business coaching client, the account's history told me the ceiling was $65. Any ad under that keeps running.

Then I look in the CRM. Have the leads from those three ads made any sales, and how much cash has actually been collected? Say you spend $90 a day for eight days. That's $720. If those leads produce one $5,000 sale, the campaign has earned more budget.

Only then do I add another $30 ad set with a new ad, and run the same process again.

Why I add new ads instead of raising budgets

I could take that $30 ad set to $36. That's a 20% bump, which is about as far as I'd ever push a winner in a day. But I've been scaling horizontally a lot more this year.

There's been a lot of turbulence with Meta's updates. So I want to stack the account with as many proven ads as I possibly can. With plenty of proven ads, lead flow stays consistent, and one ad dropping off doesn't take your leads to zero.

Once lead flow is consistent, you have a buffer. I tell clients to take about 10% of their monthly ad budget, or add it on top, and use it for experiments. That might be a new optimisation event or some heavily experimental ad types.

$200 a day: four ads, and a newer optimisation event

At $200 a day I split it into four ad sets at $50, one ad each. Nothing new there.

What's new for me is Meta's sequence optimisation event, which I started using a few weeks ago. Normally you optimise for either a lead or a booked call. From what I've seen, this lets Meta optimise for both, as long as your CRM sends booked calls back through the Conversions API.

For one client, I'm aiming for booked calls under $100, and they're coming in between $67 and $77. In another account, I ran it side by side with lead optimisation on the same ads, landing page and budget. Cost per booked call was $264 on lead optimisation and $160 on sequenced events. Those aren't normal results, so treat them as what I've seen, not a promise.

It also solves a small budget problem. At $30 a day with a $300 booked call, you'd have to wait ten days per ad to judge it on booked calls. With this event you still see leads coming in, so you can spot the drop-off between leads and booked calls much sooner. Then you can ask the right questions. Too many qualifying questions? Ads that don't match the landing page? I break down how I'd run $200 a day for a higher priced service in this video.

I give these ads five days instead of four, because a booked call costs more than a lead. Your unit economics decide how long you wait.

$1,000 or more a day: 20 ads, one per ad set

These are accounts with a sales team behind them. A closer or two, an appointment setter, follow-up that's dialled in. They close somewhere between 30% and 40%, and they sell something worth $4,000 or more.

For one of these clients I'm running 20 ads, each in its own ad set at $50 a day. Every ad is a 15-second direct response ad. An assumptive hook that calls out the pain, his offer, and a call to action. The video sales letter and the landing page do the heavy lifting, with five or six qualifying questions on the form.

The people who love building funnel stages won't like this. But it works. Leads come in at about $55. Self-booked calls cost $164, down from $260 over the last month.

When to turn ads off

Usually every four days. I'll make the call sooner only when a client's numbers allow it. A client spending $1,000 a day with years of history and cheap qualified leads gives me enough data to judge an ad faster.

The opposite is a small audience. If you only sell to businesses doing a few hundred thousand a month, there are far fewer of them than people trying to lose weight. Your cost per lead will usually be higher, and each ad needs longer to judge.

If nothing has worked yet, you're in discovery

If you've never had success with lead gen ads, you're in the discovery phase. Everyone goes through it. It feels like burning cash, but you're paying to find out what doesn't work, and that's how you find what does.

Say you launch 20 thoughtfully scripted ads and make no sales. If you've followed the structures above, you'll probably still find two or so potential winners. Work out what they're doing that the rest aren't, and do more of it.

We did this with the solar business I co-own. We paid about $20,000 to shoot just over 30 ads. Our worst case was that one ad worked for two weeks and made two sales, which would pay for the shoot. I upcycled those 30 ads into different ad types, and they became just over 220 ads. A few of them are still running.

FAQ

How should I split my Facebook ads budget for lead generation?

Give every ad its own ad set and its own budget, then judge it after four days. Your total budget decides how many ads you can test at once, not the structure.

Is $30 a day enough for Facebook lead ads?

It's the minimum I'd start with. Run one ad in one ad set at $30 a day, judge it after four days, and rotate in a new ad if it doesn't perform.

How many ads should I run at $50 a day?

On lead forms, one ad at $50 a day. On a landing page, you can split it into one video ad at $30 and up to three static image ads on the rest.

Should I put several ads in one ad set?

Not when you're testing video ads. Meta favours one or two ads and starves the rest, so give each video ad its own ad set and budget. A few static images can share one.

Should I raise the budget on a winning ad or launch new ads?

Both, but I lean on new ads. Raise a winner by 20% at most in a day. Stacking the account with proven ads keeps lead flow steady if one drops off.

How long should I run a Facebook ad before turning it off?

Usually four days. Go longer when a booked call costs much more than a lead, and shorter only when years of account history make the call clear.