Anyone can spend on Facebook ads. The difference between spending and running the channel comes down to four signals — and most agencies fail at least two of them.
Anyone can spend on Facebook ads. Plug in a credit card, pick an objective, launch a campaign. Money flows out, some sales come in, the dashboard shows a ROAS number. That's spending.
Running the channel is different. Facebook ads management is the work of structuring the account so it gets more efficient over time, surfaces signal you can act on, and stays connected to the rest of the business. Four signals separate an account being properly managed from one that's just being spent against.
A managed Facebook ads account hits four conditions: it's structured by campaign objective with naming you can read at a glance; creative is on a 4–6 week refresh cadence keyed to fatigue signals; attribution is wired through the Conversions API (CAPI) plus a server-side reconciliation; reporting links spend to gross profit by cohort, not just blended ROAS.
Most accounts hit one of those four — usually the first. Two is decent. Three is rare. Four is what we'd call "run."
Open an account being run properly and the campaign list reads like a sentence. 2026-Q2 · Prospecting · Best-sellers. 2026-Q2 · Retargeting · 30d Cart. 2026-Q2 · Conversion · Repeat-LAL. You know what each one is doing.
Open an account being spent against and you find: 47 paused ad sets with no notes, three campaigns called "Campaign 1," budgets randomly bumped at 10pm last Friday, and at least one campaign launched in 2022 still running because nobody wanted to kill it.
The structure check: pick any campaign at random and answer (1) what's its job, (2) when was it last reviewed, (3) what's the success threshold. If you can't answer all three for 80% of active campaigns, it's not structured.
Meta's algorithm rewards creative variety. The same creative running 8 weeks loses 30–50% of efficiency from peak. Most accounts run creatives until they crater rather than refreshing on a schedule.
What "managed" looks like:
What "spent against" looks like: the same hero video running 12 weeks, CTR halved, CPM up 60%, but still running because the team is busy.
After iOS 14.5, the Meta pixel alone is unreliable. Properly run accounts use Conversions API (CAPI) sending server-side events deduplicated with the browser pixel. Without that, you're missing 20–35% of conversions and Meta's algorithm is learning on a partial signal.
| Attribution layer | What it does | Required? |
|---|---|---|
| Browser pixel | Front-end events from the browser. | Yes — baseline. |
| Conversions API (CAPI) | Server-side events with hashed customer data. | Yes — recovers ~25% lost. |
| Pixel + CAPI dedup | Match key so events aren't double-counted. | Yes — without it, ROAS inflated. |
| First-party customer data | Email/phone hashed and sent to Meta for matching. | Recommended — raises match rate. |
| Shopify reconciliation | Daily compare of Meta's reported revenue vs Shopify actuals. | Recommended — catches drift. |
You can verify in Meta's Events Manager: open any conversion event and look for "Deduplicated" with a healthy match rate (70%+). If you don't see that, the wiring is incomplete.
ROAS reported by Meta is gross revenue ÷ ad spend. Useful but incomplete. The real number is gross profit ÷ ad spend, by cohort, reconciled against actual Shopify revenue and margin.
Example divergence: a campaign reports 4.5× ROAS from Meta's dashboard. After backing out 45% gross margin, return rate, attribution overlap with email, and the long-tail of repeat purchases — the contribution margin per acquired customer is £2.10 after 90 days. Probably profitable; not the £200 average revenue Meta's view suggests.
An account being run reports this number monthly, not "Meta says ROAS is 4.5×."
Properly running Facebook ads for a sub-£10M DTC business requires:
Single-person agency engagements at £2k/month are usually doing signal 1 only. Signals 2–4 require either a multi-person team or an operating layer with the analytics + creative under the same roof.
An honest agency answers all four. A pretender hand-waves at one or two.
The honest answer is "whatever makes the unit economics work." Practical starting range: 10–15% of total revenue, scaling up only if marginal customer contribution margin stays positive. Below £3k/month spend, Meta's algorithm doesn't have enough events to optimise reliably — you're paying for learning without enough volume.
It depends on margin and lifecycle. A 45%-margin business needs roughly 2.5× first-purchase ROAS to break even. A 60%-margin business breaks even at ~1.8×. Cumulative ROAS over 6 months matters more than first-purchase ROAS — repeat revenue lifts the real number considerably.
In-house if you have a senior strategist + a channel operator + creative production already. Agency if any of those three are missing AND your monthly spend justifies the retainer (£2k retainers cost £2k regardless of whether spend is £5k or £50k). Hybrid (in-house operator + outside strategist) works well for businesses in the £20k–£100k/mo spend band.
Every 4–6 weeks for the active rotation. 2 evergreen winners kept indefinitely; new concepts tested on the side. Fatigue triggers (CTR decline, frequency > 2.8, CPM rising) override the calendar — refresh earlier if signals say so.
Yes — but manageable. CAPI + first-party data closes most of the gap. The thing that's gone forever is granular click-attribution to specific creative; the thing you can still measure is incremental revenue lift via geo holdouts or matched-market tests. Meta's own privacy resources have current best practice.
Qwrki runs paid social as part of the operating layer — same team that runs the ads also runs the analytics that measures them and the creative production that feeds them. We don't sell "Facebook ads management" as a standalone product because running the channel honestly requires more than just an ads-team. Book a call — we'll audit your account against the four signals in the first session.
How to structure a Google Ads account so spend stays accountable: brand separation, negatives, search terms hygiene and the weekly operating routine.
What Performance Max actually does, the visibility you trade away, the brand-traffic problem, and how to run PMax so it earns its automation.
A practical method for setting a paid media budget: work back from contribution margin, size spend for learning, and govern with MER instead of platform ROAS.
