Brand versus performance
Split a budget between brand and performance by comparing the marginal dollar over a stated horizon, and name the brand return the split turns on.
On this page
Performance spend reaches people ready to buy this month and is judged within weeks, on clicks and sign-ups a platform can count. Brand spend reaches the larger group who will buy later and is judged, if at all, in years. The split between them is usually set by whoever owns the shortest report. Worked with numbers, the right split depends on the horizon you choose and on one input nobody in the room has measured.
Two kinds of spend, two clocks
Les Binet and Peter Field put numbers on the difference in The Long and the Short of It (IPA, 2013), a meta-analysis of 996 campaigns for 700 brands in the IPA's effectiveness databank, which is built from entries to the IPA's effectiveness awards. Sales activation, their term for performance, produces a response that is large within weeks and fades within months of the spend stopping. Brand building produces one that is small at first and grows over years, because it works through memory rather than a click. Across those campaigns, the number of very large business effects peaked when about 60% of the budget went on brand and 40% on activation. Source: The long and the short of it: 10 key principles of success (opens in a new tab) (IPA and Thinkbox).
That 60:40 now travels as a rule. It is the peak of a curve fitted to award entries, not a measurement of any one company. Binet and Field's own Effectiveness in Context (opens in a new tab) (IPA, 2018) shows how to adjust it by market category, business model, brand life stage and size, and their business-to-business analysis, The 5 Principles of Growth in B2B Marketing (B2B Institute, 2019), put the balance nearer 46:54, as How B2B marketing really drives growth (opens in a new tab) reports. For a company outside the dataset, 60:40 is a hypothesis to test.
What brand spend buys
Byron Sharp's How Brands Grow (2010) supplies the mechanism. Brands grow mainly by adding buyers, most of them light buyers who rarely think about the category. Whether such a buyer picks you at the moment of need depends on mental availability, the chance the brand comes to mind then, and physical availability, how easy it is to find and buy once it has.
For the invoicing app from the funnel and AARRR, the moment is usually a freelancer's first late payment. Mental availability is whether "I need an invoicing app" retrieves our name. Physical availability is whether we appear in the search results and the accounting tool's integration directory, and whether the free plan sends an invoice in five minutes. Brand spend builds the first, product and distribution the second, and performance harvests both. So the budgets are not independent: cut brand, and a year later the performance report shows a rising cost per customer and blames the auction.
A $1M split, three ways
Assume the invoicing app spends $1M a year, every year, and keeps at least $100,000 on brand. Value is the lifetime margin of the customers each split wins, at the $800 a customer that CAC, LTV and payback derives from the course's constants in the arithmetic a marketer carries. It is not the 12-month $368 the channel lessons use, because this lesson compares what each split earns over horizons of up to three years; payback inside 12 months is a separate test, which the exercise at the end applies tier by tier.
- Performance comes in tiers, because each extra dollar reaches the same audiences more often and bids against more competitors, which paid media basics prices. The first $300,000 a year buys customers at $333, the course's $2-click chain, returning $800 ÷ $333 = $2.40 per dollar. The next $300,000 buys them at $667, $1.20 per dollar, and the next at $1,143, $0.70 per dollar.
- Brand returns $0.40 per dollar from customers won in the year of the spend, $0.80 the year after and $0.60 in the third: $1.80 over three years. The shape, small first and larger later, follows Binet and Field; the figures are assumptions.
The tiers are assumptions too. Module 2's funnel buys broad traffic at $0.50 a visit and pays $521 a customer; priced there, the first tier returns $800 ÷ $521 = $1.54 per dollar. That lowers every total below and moves only the steady-year answer, because the 12-month and 36-month answers turn on the second and third tiers.
split, brand:performance first 12 months over 36 months still to come year 3 alone
10:90 $1.33M $4.21M $0.20M $1.47M
40:60 $1.24M $4.60M $0.80M $1.80M
60:40 $1.08M $4.56M $1.20M $1.92M
The working for 40:60. Performance brings $300,000 × 2.40 + $300,000 × 1.20 = $1.08M a year, and brand adds $400,000 × 0.40 = $0.16M in year one, so $1.24M. Over 36 months performance brings $3.24M. Year one's brand dollars have earned all of their $1.80, year two's $1.20 and year three's $0.40, so brand brings $400,000 × 3.40 = $1.36M, and the total is $4.60M.
Each horizon crowns a different split: 10:90 inside 12 months, 40:60 over 36 months, and 60:40 in year three alone, once earlier brand spend is paying out. The best split for a steady year goes further, to 70:30.
The mistake: reading the split off one report
A finance team that reviews twelve-month results picks 10:90 every time, because a brand dollar returns $0.40 inside the year and the third performance tier returns $0.70. But the third tier never earns more than $0.70. It buys $800 customers for $1,143 and loses $90,000 a year at every horizon. The twelve-month report ranks a permanent loss above a delayed gain because the loss arrives sooner.
The opposite mistake is adopting 60:40 because a book's fitted curve peaked there. Against 10:90 it gives up $250,000 of first-year margin ($1.33M − $1.08M), pulls ahead only in year three, and rests on a brand return nobody at this company has measured.
The fix: compare the marginal dollar over a stated horizon
Move a dollar from performance to brand while brand's return over the horizon beats that of performance's last tier over the same horizon. With the budget recurring each year:
- Over 12 months, brand's $0.40 meets the third tier's $0.70. Nothing moves; 10:90 is the floor.
- Over 36 months, a dollar a year on brand earns 1.80 + 1.20 + 0.40 = $3.40 inside the window, against 3 × 0.70 = $2.10 in the third tier and 3 × 1.20 = $3.60 in the second. Brand displaces the third tier and stops at the second: 40:60.
- In a steady year, a year's brand spend earns its full $1.80 against $1.20 in the second tier and $2.40 in the first. Brand displaces two tiers: 70:30.
This is the rule the five decisions states for channels, that a budget is decided at the margin. The split is an output of the horizon and the returns.
The input the split is most sensitive to
It is brand's three-year return. Keep its shape and call its total B, so a dollar a year on brand earns B × 3.40 ÷ 1.80 inside 36 months. That beats the third tier's $2.10 when B is above $1.11, and the second tier's $3.60 when B is above $1.91. So the 36-month answer is 10:90 below $1.11, 40:60 between $1.11 and $1.91, and 70:30 above. The assumed $1.80 sits 6% below the upper threshold.
The performance tiers show up in platform data within weeks, flattered by attribution but visible; the brand return shows up in no report, so it is the input to measure. Put the brand spend in half the regions, hold the other half out, and compare branded search, direct sign-ups and the performance cost per customer for six to twelve months. A slow loop needs a long-running holdout, not a two-week test, as argued in when an A/B test is the wrong instrument.
The table also counts margin, not cash, and the brand return lands in years two and three. Runway therefore sets the horizon before the marginal rule sets the split, the constraint the exercise split $1M between brand and performance adds.
In practice
To the board, show the split at two horizons, with the brand return written as an assumption beside the test that will measure it: "40:60 earns $90,000 less margin than 10:90 in the first year and $390,000 more over three, if brand returns $1.80 a dollar; the geo test reads by month nine." A board can argue with $1.80; it cannot argue with a ratio from a book. To the finance team: cut the $1,143 tier first, which needs no argument about brand. To an agency: ask for the returns its split assumes, by year.
The mistake that costs money is letting one horizon's report set the split. On a twelve-month review the third performance tier, which returns $0.70 per dollar for ever, outranks a brand dollar that returns $1.80 over three years.
Check yourself
1. A first geo read suggests brand returns $1.00 per dollar over three years, not $1.80. What do you spend next year, on a 36-month horizon?
Brand now earns 1.00 × 3.40 ÷ 1.80 ≈ $1.89 inside 36 months, under the third tier's $2.10, so the marginal rule says 10:90. But the third tier returns $0.70 per dollar and brand only breaks even after three years, so neither earns the last $300,000. Spend $600,000 on the first two tiers, keep $100,000 on brand so the measurement continues, and hold the rest.
2. The free plan now needs a sales call before a user can send an invoice. The CEO wants more brand spend to lift sign-ups. Do you agree?
No. The call damages physical availability: a freelancer who remembers the name at the moment of a late payment cannot buy in that moment, and goes to the next result. More brand spend raises mental availability that the product now converts less of, so every brand dollar returns less than the $1.80 assumed. Remove the call first, then fund and measure brand.