Paid Social
Date: 2026-08-16
Buying attention from people who weren’t looking. The mirror image of paid search — genuinely incremental, and structurally disadvantaged by every attribution model, because it creates demand rather than capturing it.
What it is
Paid social places ads in feeds based on audience characteristics rather than expressed intent.
The customer wasn’t searching for you. That’s the whole difference, and everything else follows from it.
Why it looks worse than it is
Attribution rewards the last touch before conversion. Paid social is rarely that touch:
sees an ad on social day 0 ← created the demand
searches your brand day 3
clicks the paid brand ad day 3
buys day 3
last-click credit → paid search (branded)
actual cause → paid social
Demand creation gets credited to demand capture, systematically — Demand Creation vs Demand Capture. Which means a business optimising on last-click return reliably shifts budget away from the channel that generated the demand and towards the one that harvested it — until demand dries up and nobody can explain why — Attribution Models, Multi-Touch Attribution.
The compounding problem: paid social’s own reporting overstates in the opposite direction, using view-through and modelled conversions. So you have one source under-crediting and another over-crediting the same channel, and neither is usable for budget decisions.
Only a holdout settles it — Incrementality Testing, Geo Holdout Tests.
Prospecting and retargeting are different products
Frequently blended in reporting, and they behave nothing alike:
| Prospecting | Retargeting | |
|---|---|---|
| Audience | Hasn’t visited | Has visited |
| Intent | Created by the ad | Already present |
| Apparent efficiency | Poor | Excellent |
| Actual incrementality | Higher | Often low |
Retargeting shows ads to people who were already coming back, then claims their conversions. It’s the single most over-credited tactic in digital marketing, and the one where a holdout most reliably changes minds.
Split the budgets, split the reporting, and hold out retargeting first.
Creative is the lever
Unlike search, where the query constrains everything, paid social performance is dominated by creative. Targeting has largely been automated away by the platforms; what’s left under your control is what the ad says and shows.
Which makes Creative Testing the main optimisation surface — and it’s a testing discipline with the same statistical requirements as any other, usually applied without them.
The economics
Paid social typically has higher CAC and worse first-order economics than search, offset by being genuinely incremental. On the running model:
CAC (paid social) £42.00
first-order contribution £15.00
────────
first order −£27.00
needs ~2.8 orders before contributing
Which makes retention the deciding factor. A paid social customer who repeats twice is profitable; one who doesn’t is a substantial loss. And paid social customers typically repeat less than search-acquired ones — Channel Mix, Repeat Purchase Rate.
That combination — high CAC, low repeat — is why paid social budgets need cohort-level scrutiny rather than campaign-level ROAS.
Managing it
- Judge prospecting on incrementality, not platform ROAS
- Judge retargeting on a holdout, or don’t run it
- Feed contribution, not revenue, to the optimiser
- Exclude existing customers from prospecting audiences — New vs Returning Customer Acquisition
- Track cohort value by channel, since first-order economics understate or overstate depending on repeat behaviour — Cohort Revenue