Meta needs about 50 of your optimisation event per ad set per week before the bidding settles. At the trade's filed average, a bath remodeler spending $29,700 a month gets 39 leads a week. That is under the floor, at nearly $30,000 a month.
"What should I be spending on Meta?" tends to mean "what can I afford". That is the wrong end to start from. The number you can afford has nothing to do with the number that works.
There are two real numbers. Meta sets the floor. Your calendar sets the ceiling. Your budget wants to sit between them, and on most accounts it does not.
Meta's bidding gets better at finding a result by seeing that result happen. Below a certain rate it has too little to go on, delivery goes uneven and costs swing. Meta's own guidance puts that at about 50 optimisation events per ad set per week.
So the floor is not a dollar figure. It is a count. Whatever event you told Meta to buy, you need about 50 a week of it, in one ad set.
Now put the trade's own numbers against that. Re-Bath's 2026 filing covers 122 units. The average one spends $356,326 a year on marketing and gets 2,030 leads. That is $29,700 a month, 169 leads, and $176 a lead.
169 leads a month is 39 a week.
So the average unit in this trade spends close to $30,000 a month and still does not reach Meta's floor on leads with one ad set. Fifty leads a week at $176 would take about $38,000 a month.
Two things, and both go against the usual advice.
Do not split the money across many ad sets. Four ad sets at 10 events a week each learn nothing. One ad set at 39 learns slowly. Splitting a budget that is already under the floor is the most common way to make an account worse.
Do not optimise on an event you cannot feed. If 39 leads a week is under the floor, booked estimates are far under it and visits that happened are further still. The deeper event is the better lesson. You get to use it once the volume is there.
Which event your account can learn from at your volume
The ceiling is not about money either. It is slots.
Take the visits your estimators can do in a day. Times the days you sell on. Times your cost per visit that happened. That is your ceiling.
Spend past it and the damage does not show up in the spend report. It shows up in the calendar. A homeowner who taps the ad this week gets a slot two weeks out, because the near ones are gone.
An appointment two weeks away is a worse appointment than one three days away. The homeowner has had two weeks to get three other quotes, to change his mind, or to forget. You paid the same $176 for it.
Most accounts sit in one of three places.
Under the floor. The money is real but the account never settles, and every week looks different from the last. The fix is fewer ad sets, a shallower event, or more money. Not better creative.
Over the ceiling. Bookings keep coming and land further out each week. The fix is more estimator days or less spend, and most often it is the first.
Between them. This is where you want to be. It is also the only place where changing the creative is the biggest thing you can do.
Work out which of the three you are in before you change anything else. It costs an afternoon and it decides what the next three months are about.
Your market is not the filed average. A metro with four competitors bidding costs a different amount per lead from one with fourteen. The filing's own range runs from $5 to $889.
Use the arithmetic, not the numbers. The floor is a count of events. The ceiling is a count of slots. That holds in any market and any trade.
Write down three things. How many of your optimisation event you got last week. How many ad sets that was split across. How many visits your estimators could have done.
If the first divided by the second is under 50, you have a floor problem. No amount of creative fixes it.
If your bookings land more than a week out, you have a ceiling problem. More budget makes it worse.
Meta's own guidance on the learning phase. It is about 50 optimisation events per ad set per week, for the event you told the campaign to buy. Google's equivalent for Target CPA is about 30 conversions per campaign in 30 days. Both are the platforms' published figures, not ours.
Ask them how many of your optimisation event each one gets in a week. If the answer is under 50, the money is spread too thin to learn. Ask it in writing.
On the deepest event your volume can feed. At 39 leads a week the answer is leads, and the booking becomes right later. The whole point of measuring is to move down that list as the account grows.
Then run one ad set and stop splitting. Accept that the account will take longer to settle. The alternative most owners take is to keep changing things, and that guarantees it never settles at all.
In small steps, and from the calendar rather than the dashboard. A budget that jumps resets what the bidding has learned. A calendar that fills faster than your estimators do costs you show rate, not money.
Lovro Lucic, on US hours from Zagreb. Your page, your number and your calendar stay in your time zone. Call +1 (650) 643-4497, or pick a time on the calendar.