Advanced · Lesson 4 of 4

Building a multi-channel performance plan

About 12 minutes · Yogender Kumar, Managing Partner

By the end of this lesson you'll be able to

  • Give each channel a clear role and a cost target
  • Move budget between channels using marginal CPA
  • Write a measurement and testing plan for the year

Why this matters

Many budgets are set by last year's split, plus or minus a bit. A plan built on targets, marginal costs and tests aims to put each pound where it adds the most. This lesson brings the whole course together.

The core idea

A performance plan answers five questions:

  1. What's the target? A target CPA, or a target per new customer, from your margins (Track 1).
  2. What does each channel do? Some find new people, some convince, some close. Judge each on its job.
  3. Where does the next pound go? To the channel with the lowest marginal CPA: the cost of the next orders, not the average.
  4. What are we testing? A fixed share of budget for new channels and ideas. A common rule of thumb is 10% to 15%, but set yours by what you can afford to lose.
  5. How do we know it works? One source of truth for orders, and incrementality tests for the big channels (Lesson 1).

Worked example

Kestrel Home is an imaginary UK homeware shop, so the figures are illustrative. It spends £20,000 a month. It keeps £2,000 (10%) for tests, so four channels share £18,000. Its limit is £40 per new customer.

Each channel's role and target:

  • Paid search, non-brand: finds people searching for homeware. £7,000.
  • Meta prospecting: finds new people who aren't searching yet. £5,000.
  • Affiliates: content and cashback partners, paid per sale. About £3,000 in commission.
  • Retargeting: closes sales from recent visitors. £3,000.

The latest incrementality tests and spend changes give each channel's marginal CPA, the cost of its last orders:

  • Search: the last £2,000 brought 50 new customers: £40.
  • Meta: the last £2,000 brought 80 new customers: £25.
  • Retargeting: the last £1,000 brought 20 incremental new customers: £50.

Affiliates stay as they are until a test gives them a marginal figure.

Retargeting's last £1,000 is over the limit. Meta's is well under. So Kestrel moves £1,000 from retargeting to Meta.

Meta's cost will rise as it spends more, so Kestrel assumes the extra £1,000 buys new customers at about £28: around 36. It loses 20 from retargeting. That's about 16 more new customers a month for the same budget.

It then checks the move after a month, rather than trusting the forecast.

Finally the testing and measurement plan:

  • Test budget: the £2,000 kept aside, for one new idea at a time. This quarter it's a creator programme.
  • Measurement: the order system is the total. Each big channel gets an incrementality test twice a year.
  • Review: marginal CPA by channel every month, the full plan every quarter.

My view: a plan that moves money every month on evidence beats a perfect annual split that never changes.

Kestrel's marginal CPA: what the last pounds bought

Limit £40

Search: last £2,000 for 50

£40

Meta: last £2,000 for 80

£25

Retargeting: last £1,000 for 20

£50
Retargeting's last £1,000 is over the limit and Meta's is well under, so £1,000 moves from one to the other.

In practice

  • Set one target from your margins, and write it at the top of the plan.
  • Give each channel a role, and judge it on that role.
  • Move budget by marginal CPA, in small steps, and check each move.
  • Keep a fixed test budget you can afford to lose, and run one test at a time.
  • Use your order system as the total, and test the big channels for incrementality.
  • Review monthly, and rebuild the plan quarterly.

Common mistakes

  1. Splitting budget by last year's habit. Last year's mix isn't evidence. Move money on marginal CPA.
  2. Judging every channel on last click. Channels that find new customers rarely get the last click. Judge them on their role and on tests.
  3. No test budget. Without one, you never find the next channel. Protect it.

Try it yourself

Your limit is £35. Search's last £3,000 brought 60 orders. Social's last £3,000 brought 120. You move £1,000 from search to social, and expect social's extra orders to cost £30 each. How many orders do you gain or lose overall?

Show answer

Search's marginal CPA: £3,000 ÷ 60 = £50, over the £35 limit, so £1,000 less loses about 20 orders. Social's last £3,000 bought 120 orders at £25, under the limit. Social's extra £1,000 at £30 an order buys about 33 orders. Net gain: about 33 − 20 = 13 orders a month, for the same budget.

Quick check

Not scored. Pick an answer to see the explanation.

  1. Which channel should get the next pound of budget?

  2. A plan has £25,000 a month and keeps 12% for tests. How much is the test budget?

Key takeaways

  • Start from one target set by your margins.
  • Give each channel a role, and judge it on that role.
  • Move budget by marginal CPA, then check the result.
  • Protect a test budget, and test the big channels for incrementality.