Paid Ads: Google & Meta
Spend that answers to a number you agreed in advance.
Paid media is the fastest way to find out whether your offer works. It is also the fastest way to spend a year's budget proving that it does not, which is why we start by agreeing the number that decides whether a campaign lives or dies.
You own the ad accounts. Always. If we stop working together you keep the account, the conversion history, the creative and the audiences — none of which we will hold hostage.
What it covers
The work itself. Not a list of adjectives — the things that actually get done.
Account and tracking audit
Before spending anything we check that conversions are actually being recorded correctly. A surprising share of accounts optimise against a broken signal, which makes every decision after it wrong.
Measurement setup
Server-side tagging where it is warranted, consent-mode handling, offline conversion import for longer sales cycles, and a clear definition of what counts as a conversion and what does not.
Google Ads
Search, Shopping, Performance Max, Demand Gen and YouTube. Including the unfashionable parts — negative keyword hygiene, search term review, and stopping Performance Max from eating your brand traffic and claiming credit for it.
Meta Ads
Prospecting and retargeting structured so you can actually read the result, with creative testing as the primary lever rather than endless audience permutations.
Creative testing
Enough variants to produce a signal, built against the messaging hierarchy so the ads say what the brand says. Where volume is the constraint, our AI UGC service feeds this directly.
Landing pages
The ad is half the job. We test page-level changes — offer framing, form length, proof placement — because the cheapest conversion-rate gain is usually after the click, not before it.
Budget and bid strategy
Target CPA or ROAS set against your actual margin, not a platform-suggested number. Scaling rules agreed in advance so nobody is improvising at month end.
Incrementality checks
Geo holdouts or brand-campaign pauses where the account is big enough to support them. Platform-reported ROAS flatters itself; occasionally testing what happens when you switch a campaign off is the honest correction.
Weekly change log
Every change, dated, with the reason. So performance shifts can be attributed instead of guessed at.
How we do it
Method, in the order it happens.
You can judge an agency on its process before you have any results to judge it on. That is the point of publishing this.
-
Agree the number first
Target CPA or ROAS derived from your margin and sales cycle, agreed before launch. Without it, every reporting conversation becomes an argument about which metric to look at.
-
Fix measurement before spending
We validate conversion tracking end to end. If the signal is wrong, the algorithm optimises confidently toward the wrong outcome and the spend is wasted regardless of how good the ads are.
-
Start narrow
Highest-intent, lowest-risk campaigns first — usually brand defence and bottom-of-funnel search. Establish a working baseline before spending money on the harder inventory.
-
Test creative, not just settings
On modern platforms the algorithm handles most targeting. Creative and offer are the levers you still control, so that is where the testing budget goes.
-
Scale in steps you can reverse
Budget increases in defined increments with a checkpoint at each one. Doubling spend overnight resets learning and tells you nothing about why performance moved.
-
Check the platform is telling the truth
Attribution is generous to whoever reports it. Where account size allows, we run holdouts to see what would have happened anyway.
-
Report with losses included
Weekly change log, monthly review against the agreed number, and the tests that failed shown next to the ones that worked.
What you keep
Deliverables that remain yours whether or not we continue.
- Full ownership of every ad account, audience and conversion history
- A documented measurement setup you can hand to anyone
- All ad creative and copy produced during the engagement
- The agreed target CPA or ROAS and the margin maths behind it
- A dated change log for the whole engagement
- Landing page variants and test results
- Any incrementality test design and findings
When this is the wrong service
We would rather say this now than three months in.
- The offer has not converted for anyone yet. Paid ads will find that out expensively rather than fix it.
- Budget is below the level where the platforms can exit learning. We will tell you the threshold honestly rather than take the retainer.
- You want the agency to own the ad account. We do not work that way and neither should anyone else.
- You need attribution to be simple and certain. It is neither, and we will not pretend otherwise to make a report look tidier.
- Margins cannot support customer acquisition at any realistic cost per sale. That is a pricing problem, not a media problem.
Before you commit to anything
We publish how we work, what we will not claim, and how an engagement runs, so you can judge us on method rather than on adjectives.
Questions
About Paid Ads: Google & Meta.
Tell us what is actually stuck.
No proposal until we understand the problem. If this is not the right service for it, we will say so.