03 / 07 · Paid media
Spend managed against revenue, not the platform's number.
The paid agent manages budgets, bids and campaigns against what your own data says converted, not against what the ad platform reports. It corroborates every platform conversion against analytics and the CRM before it moves money, writes the ledger entry before it touches the account, and makes one material change per ad set per run so the effect of each can be read.
What it reads
What it analyses
- Cost per confirmed lead or purchase, using your records rather than the platform's attribution.
- Which audiences and placements produce leads that close, and which produce leads that do not.
- Where spend is leaking: frequency, overlap between ad sets, and campaigns still running on a stale offer.
- Whether the account is measuring enough to move budget safely at all, and saying so when it is not.
How it decides
Budget moves only where the corroborated cost per outcome supports it, and only one material change per ad set per run, so the next read is clean. Every change is logged with the expected effect and a review date before the account is touched. On low-volume accounts the agent prefers fewer, larger comparisons over many small tweaks, because a hundred small changes at small n is how accounts end up optimised for noise.
Example experiments
The shape of a test, written before it ships.
Illustrative, not a promise. Every experiment on your account gets this record before anything changes: the hypothesis, what the verdict is read from, what stays fixed, how long until a verdict, and where a person decides.
Hypothesis 01
Reallocating budget from the ad sets with the lowest corroborated cost per lead towards the ones with the highest CRM close rate lowers cost per customer, even if cost per lead rises.
- Measured on
- Cost per closed customer, from the CRM, against the prior period at matched spend.
- Held constant
- Creative, landing pages and total budget.
- Verdict after
- One sales cycle plus the settling period.
- Human gate
- Budget ceilings set by you; moves above a threshold need approval.
Hypothesis 02
A narrower audience built from the profile of customers who actually closed outperforms the platform's broad targeting on cost per qualified lead.
- Measured on
- Qualified leads per unit spend, both arms running against the same landing page.
- Held constant
- Creative, offer, bidding strategy and daily budget per arm.
- Verdict after
- Until both arms have enough qualified leads to compare on medians.
- Human gate
- Audience definition reviewed before launch.
Hypothesis 03
Frequency caps on the retargeting layer cut wasted impressions without lowering conversions.
- Measured on
- Conversions and cost per conversion on the capped campaign against the uncapped period.
- Held constant
- Audience, creative and budget.
- Verdict after
- Three to four weeks.
- Human gate
- The agent may ship this on its own once it has earned autonomy on the account.
Where people decide
You set budget ceilings and the margin or payback target the agent optimises to.
Changes above a spend threshold, and every new campaign, need approval before they go live.
No spend moves on a metric the analytics agent has flagged as unreliable.
What you get
- 01A reconciliation of platform conversions against your own records.
- 02One logged change per ad set per run, with the expected effect and its review date.
- 03Budget moved towards what closes, with the reasoning written down.
- 04A monthly read on cost per outcome, in your numbers, not the platform's.