Approach
Fix measurement first. Then let the machine spend.
The model does the volume work every week. A person with twenty-three years in search reads every proposal and decides what ships.
[NEED: name for the method]
The model reviews search terms, drafts negatives, proposes bid and budget changes, and writes copy variations. Every week, in every account.
A person reads every proposal, ships the ones that hold up, and writes down why the others didn’t.
Both are needed because the model optimizes to whatever it’s told. Most accounts tell it the wrong thing, usually to maximize form fills.
A week in the account
What the model proposed, and what shipped.
[NEED: a real anonymized week. The rows below are illustrative, from the homepage.]
Week of [date] · [Client, anonymized]
[N] proposals reviewed
The model proposed
What ships
Add 48 search terms as phrase-match keywords
Added 6. The other 42 are researchers and job seekers.
Raise target CPA 15% to capture more volume
Held. SQL rate fell last week; volume isn’t the constraint.
Negatives: 37 terms from last week’s waste
Accepted. This is the work the machine does better than we do.
Pause the LinkedIn retargeting campaign (low CTR)
Kept it. It sources a third of opportunities, measured in the CRM.
Rotate 12 new generated headlines
Two, rewritten against this month’s sales call notes.
5 proposals · 1 accepted · 4 reshaped
every decision written down
The four phases
How an engagement runs.
Weeks 1–2
Audit and tracking
Conversion actions, GA4 key events, tag health and CRM stages get checked and fixed. You get a written read on the account before anything changes. Nothing gets bid on until the numbers are trustworthy.
Weeks 3–4
Rebuild
Campaign structure, search-term tiers, negatives, landing-page alignment and ad copy grounded in what already converts. The new structure goes live against the fixed tracking.
Months 2–3
Bid to pipeline
CRM stages flow back to the platforms as offline conversions, with value rules, so bidding targets SQLs and opportunities instead of whoever fills a form.
Ongoing
Scale with guardrails
Budget follows lead quality by campaign. Increases wait until the pipeline data supports them. Monthly review with sales, quarterly with leadership.
What we won’t do
- Bid on your brand to pad the numbers.
- Click “apply all” on platform recommendations.
- Report clicks and impressions as results.
- Grow your budget to grow our fee. The fee is fixed.
Where the work happens
Google Ads, Microsoft Advertising, LinkedIn Campaign Manager and OpenAI ads. GA4 and Google Tag Manager for measurement. HubSpot and Salesforce for offline conversion imports. [NEED: other tools to name, and whether to name the AI tooling]
Who does the work
The person on the strategy call is the person in the account. [NEED: contractors, if any, and what they do.] More about the firm.
See the method on your account.
A 30-minute strategy call ends with a written read on your account and a recommendation.