How to ask a client for a bigger ad budget
A bigger ad budget gets approved when you price the next increment: its cost, what the first days buy, and when the money goes back.
A bigger ad budget gets approved when you price the next increment: its cost, what the first days buy, and when the money goes back.
The account is profitable. CPA sits comfortably under target, the client is happy, and the obvious next move is more money. So you write the email: "Results are strong, we recommend doubling the budget." Then nothing happens for three weeks. The finance lead has a question you didn't answer, and the question is always the same one: what exactly do we get for it?
I have written that email. It stalls because it asks for money without saying what the money buys.
Short answer: To get a bigger ad budget approved, write the request as a purchase. State the price of the next increment from your own last step-up, admit the first days after a big budget change mostly buy information, and name the number and date at which you hand the money back. That makes it a bounded bet a client can sign.
The takeaways
Budget increase requests stall because they quote the wrong number. The typical ask says "our CPA is €30, so double the spend and get double the customers." The client's finance person knows that's not how buying works anywhere else, and they're right to doubt it.
€30 is the average price of every conversion you've bought so far, cheap early ones included. The money you're asking for buys the next increment, and the auction sells that increment at a worse price: the platform has to reach people who were always less likely to convert. If you've read why ads look better on a lower budget, it's the same mechanism from the other side.
So the request that sounds confident ("same CPA, twice the volume") is the one most likely to miss. When it misses, the next request gets harder.
Your previous step-up is the best evidence you have, because elasticity is specific to each account. Pull the weeks right before and right after it. Ignore the campaign's lifetime average.
A made-up example to show the arithmetic: at €150 a day you got 35 conversions a week, a €30 CPA. You moved to €300 a day. The following weeks averaged 50 conversions on €2,100. The blended CPA reads €42, which looks like a mild dip.
The marginal view is harsher. The extra €1,050 a week bought 15 extra conversions, so €70 each. That €70 is the honest price of growth in this account, and it's the number to put in front of the client, next to what a customer is worth to them. If €70 still pays, you have a strong case. If it doesn't, you've just saved yourself an awkward month.
No. Promising today's CPA at double the spend sets up a failure you can see coming. Meta's documentation on the learning phase names budget changes among the significant edits that can send an ad set back into learning, and during that stretch the delivery system explores and results swing.
So the first days at a new spend level mostly buy information about that level. Budget for it like you would budget for a test.
An honest proposal says that up front: "Expect the first week to run above our current CPA while the campaigns relearn. Judge the increase on weeks two to four." A warned-about cost is easy to live with. Say it before the client sees it in the dashboard, and the bad first week becomes proof that you know the account.
Present the new level as a range and say where the range comes from. "At €300 a day we expect a marginal CPA between €60 and €80, based on what the last step-up cost" is a sentence a finance lead can check against your history, and argue with on specifics.
Then name what the increment buys, because three options carry three different risk profiles:
Mixing all three into one number hides which part carries the risk.
A stop condition makes a bigger budget safe to approve. Before any money moves, write down the number and the date: "If marginal CPA is above €80 on 4 November, we return to €150 a day." That one sentence turns an open-ended spend into a bounded bet with a refund condition, and bounded bets get signed.
The stop condition only works if something enforces it. A promise to keep an eye on spend fails the first week you're busy. That enforcement is where tooling earns its place.
In Adscalr, the budget intelligence layer builds 3 to 5 prioritized campaign plans from the last 12 weeks of real performance, each with a conversion goal and a CPI marked as an AI estimate. Staged pacing alerts check every 5 minutes and fire on runaway spend (150% of cap), overspend (110%) and underspend (below 70% after midday). None of that predicts what a new spend level will return. Your own step-ups are still the only evidence for that. It just makes sure the limits you agreed on hold.
This is the thinking behind Adscalr.
See the product →