The honest answer is “less than you can afford to lose, and more than a number that produces no data.” That is not useful on its own, so here is how to turn it into a figure for your product.

Start with the value of a customer

Before choosing a budget, know roughly what one customer is worth. You do not need a perfect model. A range is enough.

Two ways to estimate:

  1. Revenue-based: average monthly price × average months retained. If you charge €19/month and people stay six months, a customer is worth about €114.
  2. Replacement cost: what would you pay someone else to acquire this customer? That number is your ceiling, not your target.

Write the number down. Every budget decision is relative to it.

The three budget tiers

Tier 1 — learning budget. Enough to get 1,000–2,000 impressions or 100–200 clicks in your target placement. For many indie products this is €100–€300. The goal is not profit; it is a signal.

Tier 2 — validation budget. Enough to repeat the test two or three times with different copy or audiences. Usually 3–5× the learning budget. The goal is to see whether the first result was luck.

Tier 3 — acquisition budget. Spend you can justify from unit economics: cost per signup below the value of a customer, with a margin for churn. This is where ads become a line item, not an experiment.

Most indie products should not skip from tier 1 to tier 3. The middle tier is where you learn whether the channel deserves to scale.

Set the test budget as a fixed number

Decide the amount before you launch and do not adjust it mid-test. A moving budget makes the result uninterpretable.

A practical rule: if losing the entire test budget would affect your ability to pay for something important this month, the budget is too high. If it is so small that you cannot get 100 clicks, it is too low to learn anything.

What “enough data” means

You do not need statistical significance to make a decision. You need enough volume that a single unusual day does not define the result.

A rough guide:

  • Under 50 clicks: too early. Do not conclude anything.
  • 50–200 clicks: direction, not proof.
  • 200+ clicks with stable CTR: a usable read.

If your placement cannot deliver that volume within your test window, the channel is too small for a paid test — treat it as a sponsorship or skip it.

When to increase spend

Increase only when all three are true:

  1. Cost per signup is below the value of a customer.
  2. The result has held for at least two weeks.
  3. The landing page and onboarding can handle more volume without breaking.

If any of those is false, more budget amplifies a problem instead of a result.

When to stop

Stop when:

  • You have spent the test budget with no conversions and no promising signal.
  • Signups do not activate after several attempts to fix onboarding.
  • The cost per signup is more than the customer is worth and the trend is not improving.

Stopping is not failure. Continuing a channel that does not work is the expensive choice.

The spreadsheet version

If you like numbers, this is the whole model:

max cost per signup = customer value × margin
test budget        = max cost per signup × target signups
target signups     = enough to judge the channel (often 5–20)

If the resulting budget is too small to produce 100 clicks, your customer value is too low for that channel, or the channel is too expensive. Both are useful discoveries.

Budgeting is not about finding the perfect number. It is about making the experiment small enough to run and large enough to mean something.

When campaigns open on RoketAd, you will be able to start at the learning tier without a sales call. Join the waitlist to be notified.