Episode 25: Budget, bidding and A/B testing

Daily vs lifetime budget

With a campaign budget (Advantage campaign budget), Meta distributes money between ad sets itself. With an ad set budget, you’re in control. Both can work in a small account; for tests, ad set budgets are cleaner.

Bid strategies, in simple words

• Highest volume (default): as many results as the budget allows. Start here.
• Cost per result goal: “I want one order at around ₹250”. After some data comes in.
• Bid cap: the maximum bid in the auction. Advanced; risky for small budgets.
• ROAS goal: a target for revenue ÷ spend. With a website + purchase tracking.

StyleStreet: highest volume first, try a cost goal after 2–3 weeks.

The learning phase: let the ad learn

After a new ad set or a big change, Meta goes into “learning”. Results jump up and down during this time.

• Meta’s guidance: an ad set needs roughly 50 optimization events in a week to stabilise
• Don’t edit repeatedly during learning. Every big edit restarts learning
• 50 purchases is hard on a small budget, so sometimes you optimise for a higher-funnel event like “messages”

A/B testing: one thing at a time

Meta Ads Manager also has a built-in A/B test feature that splits the audience evenly.

When do you declare a winner?

Picking a winner after 2 days and ₹200 = tossing a coin.

• At least 5–7 days (weekdays + weekend)
• Enough spend on each version for some results (e.g. 15–20+ messages/orders each)
• The difference is big and consistent, not one day’s

What if both are equal?

Then that variable probably doesn’t matter much. Keep whichever is cheaper/easier, and test the next big variable (like the offer or hook).

StyleStreet’s real problem

Practical example: Hook test: wash test vs POV

Same audience, same offer, same budget (₹150/day each), 7 days:

Winner: B. Next test: B with two different offers.

Your mission: Build a test plan

For StyleStreet, the next 4 weeks:

1. Choose a budget type (daily/lifetime) and explain why
2. Decide the starting bid strategy
3. List 4 tests for 4 weeks (one variable at a time)
4. Write a hypothesis for each test: “I think B will be better because…”
5. Write the rules for declaring a winner (days, spend, results)
6. Design a sheet for test results

Quiz

1. The most important rule of an A/B test?

   • A. Change as many things at once as possible
   • B. Change only one variable at a time
   • C. Test only on weekends
   • D. Keep budgets different
   Show answer

   Correct answer: B. Change only one variable at a time

   Only then do you know what caused the difference.

2. What happens if you edit repeatedly during the learning phase?

   • A. Results come faster
   • B. Learning restarts and results stay unstable
   • C. No difference
   • D. The ad stops
   Show answer

   Correct answer: B. Learning restarts and results stay unstable

   Give the system time to learn.

3. The starting bid strategy for a new account?

   • A. Bid cap
   • B. Highest volume
   • C. ROAS goal
   • D. Manual CPM
   Show answer

   Correct answer: B. Highest volume

   Let data come in first, then set cost goals.

4. Ad B costs ₹150/order, Ad A ₹350 (7 days, equal spend). What do you do?

   • A. Stop both
   • B. Treat B as the winner, scale it, run the next test
   • C. Give A the budget
   • D. Wait a year
   Show answer

   Correct answer: B. Treat B as the winner, scale it, run the next test

   The difference is clear and consistent. Keep the winner and move the testing forward.

Output for the client project

Your project file now has a Budget & Testing Plan:

• Budget type + bid strategy
• 4-week test roadmap
• Winner rules
• First result: the “Washed 10 times” hook wins (₹150/order)

Golden rule: Decide by test, not opinion. One thing at a time, enough data, and let the learning breathe.

In the next episode: In Chapter 26 we’ll bring back those who came, looked and left without buying: retargeting and remarketing.