Case studies
Every number here is a screenshot.
Unedited Meta and Google exports from live accounts, with the baseline left in frame. Client names withheld as standard, the totals are the platform’s arithmetic, not ours, and we will walk any of these through on a call.
£2.33m
Donation value from paid social, lifetime of one account
9.25×
Return on ad spend, all time, as the platform reports it
£7.83
Cost per donation, Qurbani 2026, from £14.20 in 2024
+81%
More donations, same appeal, smaller budget
Three years, one target, one account
£251,862 of paid social run to a cost per donation instead of a reach figure. 107 campaigns consolidated into three funnels; prospecting kept wide, retargeting capped so it stopped buying gifts the email list would have collected for nothing.
changedBudget moved onto the two creative angles carrying the account. Reporting re-based on purchase value, never reach.
is goodCharity paid social typically returns 3 to 4×. This account has held 9.25× across its entire lifetime, not a single strong month.
Ramadan, three years running
The busiest six weeks in the charity calendar, run three times on a near-identical budget: £74.3k, £75.8k, £79.4k. Donations went 3,489 to 4,566 to 5,176 while the cost of each one fell from £21.30 to £15.33.
changedBudget locked before the season rather than chased during it, donor file suppressed from prospecting, creative pre-built so testing happened in week one instead of week four.
is goodRepeatable, not lucky, the same appeal getting cheaper three seasons in a row is the only proof an account is compounding.
£7.83 a gift on a 9× account
Six weeks across the Qurbani window: £18,463 spent, £170,206 raised, 2,358 donations. Every gift cost £7.83 and the account returned 9.22× on spend.
changedSpend front-loaded into the ten days when Qurbani intent peaks rather than spread across the window. Donor file suppressed from prospecting, broad targeting left to find new givers.
is good9.22× on a five-figure budget, with cost per gift under £8. Search-style efficiency on paid social, where most charity accounts sit at 3–4×.
11.58× across Ramadan
£75,831 spent, £878,054 raised, 4,566 donations at £16.61 a gift. The account returned 11.58× on spend across the Ramadan window.
changedBroad targeting with the donor file excluded, spend weighted toward the last ten nights when giving intent peaks. Creative refreshed fortnightly rather than rebuilt per appeal.
is good11.58× on a five-figure Ramadan budget, against a charity paid-social norm of 3–4×. 4,566 gifts from a single six-week window.
70× on search, on a modest budget
$3.67k of cost against 258k of conversion value from 2.43k clicks. Search is where the intent already exists, this is what it looks like when the keywords are the ones that carry the cause.
changedBrand and cause terms separated. Everything else earns its place from search-term data, reviewed weekly.
is good70× return on a four-figure budget. Search rewards precision over spend, proof that a small charity budget is not the limiting factor.
19× across a full year of search
$11.6k of cost, 221k of conversion value, 6.6k clicks. Cost and value plotted on the same axis, twelve months, no smoothing applied.
changedBudget held flat and spent on fewer terms, two-thirds of the inherited keyword list paused in month one.
is good19× sustained over a full year on a flat budget, achieved by removing keywords rather than adding money.
The baseline we started from
22.7m impressions and £17.47 a gift. Reach was never the problem, the account was buying attention it could not convert, across 107 live campaigns competing with each other in the same auction.
changedFirst ninety days: kill the overlap, fix the pixel and purchase events, rebuild the funnel, then touch budget.
is goodWe show the weakest year on purpose. Everything above it is measured from this line, so the improvement is verifiable rather than asserted.
ACoS run hot on purpose
More than double the other account’s TACoS, on purpose. Click-through ran at 0.54% against 0.33%, and the brand banked $473,835 in a month.
changedTarget ACoS raised on purpose to take share in a live category, tracked against TACoS so the true cost against total sales stayed visible.
is goodA single TACoS number means nothing without the account it sits inside. This one ran hot by design: $93,275 of ad spend returned $362,936 in ad sales, and total sales still reached $473,835.
Amazon PPC · lifetime account view
Every day the account has run: forty five million impressions, $1,018,489 of advertising revenue at 17.84% ACoS, and 5.61x return on every dollar in.
changedFull account history measured rather than a favourable window, so the average ACoS and ROAS are what the account actually does.
is goodLifetime numbers cannot be cherry-picked. 5.61x held across every day the account has run, not a single strong quarter.
Appeal seasons, year on year
The same appeal, three years running, getting cheaper every time.
One good campaign is luck. The test is the same appeal, same season, three years apart, budget held flat, the only way to see whether an account is compounding or just spending. Each row links to its export.
Ramadan appeal
Meta · Facebook and InstagramThe busiest six weeks in the charity calendar, run three times on a near-identical budget. Donations up 48% while the cost of each one fell 28%.
| Season | Spend | Donations | Cost per donation | Raised | Screen |
|---|---|---|---|---|---|
| Ramadan 2024 1 Mar – 15 Apr 2024 | £74,315 | 3,489 | £21.30 | £653,407 | view ⤢ |
| Ramadan 2025 15 Feb – 5 Apr 2025 | £75,831 | 4,566 | £16.61 | £878,054 | view ⤢ |
| Ramadan 2026 1 Feb – 21 Mar 2026 | £79,356 | 5,176 | £15.33 | £684,645 | view ⤢ |
changedBudget locked before the season rather than chased during it, existing donors suppressed from prospecting, and creative pre-built so testing happened in week one instead of week four. 2026 shows a softer average gift than 2025, more donors giving slightly less, which is why we report cost per donation and gift count separately rather than hiding both inside a return figure.
Qurbani appeal
Meta · Facebook and InstagramThe clearest proof on the account: 2026 spent slightly less than 2024 and brought in 81% more donations at 45% less each.
| Season | Spend | Donations | Cost per donation | Raised | Screen |
|---|---|---|---|---|---|
| Qurbani 2024 15 May – 18 Jun 2024 | £18,476 | 1,301 | £14.20 | £162,699 | view ⤢ |
| Qurbani 2025 1 May – 9 Jun 2025 | £22,834 | 1,687 | £13.54 | £212,598 | view ⤢ |
| Qurbani 2026 1 May – 6 Jun 2026 | £18,463 | 2,358 | £7.83 | £170,206 | view ⤢ |
changedTwo years of Qurbani search and creative data reused instead of restarted. Audiences narrowed to the segments that gave in previous years, and the spend front-loaded into the ten days when the intent peaks rather than spread evenly across the window.
Strategy brief
How those numbers were actually made.
The same six moves, in the same order, on every account we take over. None of it is proprietary, we publish it because knowing the method and having run it forty times are different things.
The one rule
If a change cannot be traced to a donation in the platform’s own reporting, we do not make it and we do not bill for it.
Fix the measurement first
Pixel, conversions API and purchase events rebuilt so a donation is counted once, with its real value attached. Until this is right, every other number is a guess.
→ Made the −37% cost per gift readable at allConsolidate, do not add
107 live campaigns collapsed into three funnels. Overlapping ad sets stop bidding against each other and the algorithm gets enough data per unit to learn.
→ 7.84× to 10.62× on a flat budgetExclude the donors you already have
The existing donor file is suppressed from prospecting and retargeting is capped, so paid media stops buying gifts email and SMS would have collected for free.
→ Cost per gift £17.47 to £10.94Creative on a cycle, not per appeal
New angles tested on a fixed fortnightly rhythm so fatigue is caught before performance drops, rather than scrambling when an appeal underperforms.
→ 9.25× held across three yearsWeight the budget to the calendar
60% of the annual spend deliberately concentrated into the periods when giving intent already exists, Ramadan, Qurbani, Christmas, emergencies.
→ £975k in a single year from £91.8kSearch earns its keywords
Brand and cause terms separated, then two-thirds of the inherited keyword list paused. What remains is re-earned from search-term data every week.
→ 70× conversion value against costWant your account read the same way?
Give us read-only access. Three working days later you get a written answer: where the money is going, what it is buying, and the three things we would change first, creative included. Yours to keep either way.