Paid Social — Meta, LinkedIn, TikTok
Meta Ads that survivecontact with a spreadsheet.
Paid social gets judged on platform-reported numbers, and platforms are generous to themselves. Set the measurement up so you can judge it on real numbers, and the rest is craft: structure, audiences and creative volume.
Who is claiming this conversion?
Cost per acquisition
Down
Continually, in competitive industries
Why paid social gets cut first.
Usually not because it does not work. Because nobody can prove that it does.
Meta takes credit for everything
Default attribution windows plus view-through means the platform claims conversions it merely witnessed. Add it to Google’s number and you have more reported conversions than actual customers.
Signal loss gutted the reporting
iOS restrictions, consent banners and browser changes cut the data the pixel receives. Without the Conversions API and proper deduplication, you are optimising on a fraction of the events, counted twice.
The creative treadmill never stops
Paid social burns creative faster than any other channel. Without a testing structure and a production pipeline, performance decays and everyone blames the targeting.
Nobody agreed what success looks like
Prospecting judged on last-click. Retargeting judged as if it created the demand. Two campaigns doing different jobs held to the same number, so one of them always looks like it is failing.
What the work actually involves.
Get the signal back first
Before campaign structure, before creative, the events need to arrive and be counted once. Conversions API alongside the pixel, deduplication on event ID, consent handled so you keep what you are entitled to keep. This is the same discipline as the tracking work, applied to Meta.
- Meta Conversions API implemented alongside the pixel
- Event deduplication on event ID, verified rather than assumed
- Consent Mode and CMP integration so events are legal and complete
- Event match quality raised with the parameters you are allowed to send
Meta event diagnostics
previewDuplicate purchases removed
1 in 3
Typical before dedupe is fixed
Structure that lets creative do its job
Modern Meta buying rewards consolidation and creative volume, not fifty ad sets slicing an audience into pieces too small to exit learning. The account gets structured so each campaign has one job, exclusions are clean, and testing produces a readable answer.
- Consolidated campaign structure built for the current auction, not 2019
- Clean exclusions so prospecting and retargeting stop bidding against each other
- A creative testing framework with defined winners and a rotation cadence
- Catalogue and Advantage+ setups where the product range justifies them
Creative test read-out
Platforms managed
4
Meta, LinkedIn, TikTok, Bing
Judged against numbers you can defend
The end point is a view of paid social your finance team will accept. That means comparing platform claims against back-office reality, sense-checking with holdouts or geo tests where budget allows, and reporting blended acquisition cost rather than three contradictory dashboards.
- Platform-reported figures reconciled against CRM or back-office revenue
- Holdout and geo testing where budget makes it viable
- Blended cost per acquisition across channels as the headline number
- Reporting built in Looker Studio or Power BI, not screenshots of Ads Manager
Blended acquisition cost
-18%
After reconciliation and restructure
Reporting stack
previewEverything that is on the table.
Meta primarily, plus LinkedIn and TikTok where the audience is genuinely there.
- Full paid social account audit
- Campaign structure and account rebuild
- Meta pixel and Conversions API implementation
- Event deduplication and match quality improvement
- Catalogue setup and Advantage+ shopping campaigns
- Audience strategy, exclusions and suppression lists
- Creative testing framework and rotation cadence
- Ad copy and creative direction briefs
- LinkedIn campaigns for B2B lead generation
- TikTok campaigns where the audience justifies it
- Landing page and offer recommendations
- Lead form versus landing page testing
- Holdout and geo incrementality testing
- Cross-channel blended reporting
- CRM integration and lead quality feedback loops
- Monthly reporting and stakeholder presentations
I do not produce video or design assets in volume. I brief them, direct them and can build simple statics in Figma, Photoshop or Canva — but for real creative output you want a creative, and I will tell you that.
Paid social across nine years.
B2B and B2C, agency and in-house, across Meta, LinkedIn, TikTok and Bing.
Lower
Continually reducing cost per acquisition for businesses in competitive industries, sustained rather than a single good month
B2B
And B2C accounts managed with budgets from £1,500 to £10,000 per month across multiple channels
8+
Accounts managed simultaneously agency side while reporting to account managers and clients
CAPI
Conversions API, deduplication and consent handled by the same person building the campaigns, not thrown over a wall
Client names withheld under agency and contract terms. Happy to talk specifics on a call.
How a paid social engagement runs.
Same sequence as everything else: fix the signal, then spend the money.
Audit
Account structure, event setup, deduplication, audience overlap and creative performance, written up with what is broken and what it is costing.
Signal repair
Conversions API, deduplication and consent fixed before restructuring. There is no point optimising towards events that are half missing and double counted.
Restructure and test
Rebuild campaigns, clean up exclusions, launch a structured creative test with a defined read-out period rather than endless tinkering.
Scale and reconcile
Scale what works, and report it against back-office revenue so paid social is defended with real numbers when budgets get reviewed.
Whether paid social fits your business.
It suits some businesses far better than others, and the difference is fairly predictable.
This is a good fit if
- You have a visually demonstrable product or a clear, specific problem you solve
- You can produce or commission new creative regularly — this is the real constraint
- Your buying cycle allows for demand generation rather than only capturing existing intent
- You have a reasonable average order value or customer lifetime value
- You are willing to judge it on blended numbers rather than platform-reported ones
Call someone else if
- You have no creative capacity and no budget to produce any
- You sell something highly technical to a tiny, specific buyer set — try LinkedIn or search
- You need results this fortnight with no learning period
- You expect paid social to perform like search does — different intent, different job
- You will judge it purely on last-click attribution, in which case it will always look bad
Paid socialquestions.
Something not covered? Ask me directly — you will get a straight answer, not a proposal.
Why do Meta and GA4 never agree?
Different attribution models, windows and identity resolution. Meta credits view-throughs and uses a longer click window; GA4 is last-click by default. Neither is lying, they answer different questions. My job is to make the gap explainable and get both closer to the back office.
Do we need the Conversions API?
Yes, in almost all cases now. Browser-only pixel tracking loses a meaningful share of events to restrictions and consent. But it must be deduplicated properly or you swap missing data for double-counted data, which is arguably worse.
How much creative do we need?
More than you think and fewer than agencies sell. Three to five genuinely distinct concepts to start, then refresh the winners monthly. The variable that matters is distinct angles, not the number of aspect ratio exports.
Can you run LinkedIn as well?
Yes. LinkedIn is expensive per click and worth it for a narrow set of B2B cases where the targeting justifies the cost. I will tell you honestly if your audience is reachable more cheaply elsewhere.
What budget does paid social need?
Around $2,000 to $3,000 a month is a realistic floor for a fair test on Meta, because you need enough conversion volume to exit the learning phase. Below that, spend it on search where intent already exists.
Will you produce the ads?
I brief, direct and can build simple statics. For video and volume creative you want a specialist, and I would rather say that than deliver mediocre assets and blame the algorithm.