Channel Mix
Date: 2026-08-16
The portfolio view of where customers come from. Channels differ not just in cost but in the kind of customer they bring — so a mix optimised on first-order CAC can systematically acquire worse customers.
What it is
Channel mix is the distribution of acquisition across sources, and the economics of each.
The comparison that matters isn’t CAC alone:
channel share CAC first-order repeat contribution
contribution rate LTV
paid search 30% £28.00 £15.00 28% £22.40
paid social 25% £42.00 £15.00 18% £19.20
organic search 25% £0.00 £15.00 35% £24.80
email 8% £2.00 £15.00 52% £31.60
referral 7% £12.00 £15.00 40% £27.00
affiliate 5% £18.00 £15.00 22% £20.40
[CHECK: illustrative figures — the shape is what matters, not the numbers. Compute your own.]
Paid social looks affordable on CAC and is the worst on lifetime value, because the customers it brings repeat least. Ranking channels by CAC alone gets this backwards.
Channels bring different customers
The point the table makes, and it’s the one most budget decisions miss. A channel isn’t a tap delivering identical customers at a price — it’s a selection mechanism.
- Branded and organic search capture people who already decided. High intent, high repeat, and largely not incremental — Incrementality Testing
- Paid social creates demand among people who weren’t looking. Genuinely incremental, lower intent, lower repeat
- Discount and deal channels bring price-sensitive customers who repeat only for the next discount
- Referral brings customers who resemble their referrer, which is why referral LTV is usually high
In plain terms: cheap customers are often cheap because they were nearly free to get, which usually means they were coming anyway — or because they’re low-commitment, which shows up later as poor retention.
Concentration risk
A mix dominated by one channel is fragile in a specific way: the channel owner can change your economics unilaterally. An algorithm update, a policy change, a bidding change, a measurement change. There’s no negotiation and no notice.
A rough working position: no single channel above around 40% of new customers, and at least one channel you own outright — email, direct, organic. That’s not a growth optimisation, it’s insurance, and it costs efficiency in the short run.
Reading it properly
- Segment LTV by acquisition channel, always. A blended LTV justifying a channel’s CAC is wrong in both directions simultaneously — Customer Lifetime Value
- Compare payback, not just LTV:CAC. A channel with excellent lifetime value and twelve-month payback constrains growth more than a mediocre one paying back in three — Payback Period
- Use marginal CAC for spend decisions. Average CAC tells you what you’ve paid, not what the next customer costs — Marginal Analysis
- Discount attributed figures. Channel-level attribution is a modelling output, not a measurement, and it flatters bottom-funnel channels systematically — Attribution Models, Walled Garden Reporting
Where it’s decided badly
- Optimising the mix on last-click return, which reliably shifts budget towards channels that harvest existing demand and away from those creating it
- Judging channels on first-order contribution and never checking repeat rate
- Treating organic as free. It has a cost — content, technical work, time — it’s just not in the ad account
- Blended CAC in channel decisions, where organic customers subsidise the arithmetic — Customer Acquisition Cost
- No incrementality view at all, so the least incremental channels keep winning budget on the strength of their attribution