Empirical AI for media companies
AI decisions your board can defend.
We work with publishers and online media businesses on the questions that carry real consequence: what to license, what to build, what traffic to plan for, and what to leave alone. Every recommendation is tested against your own data before it reaches a slide.
01
Licensing & content deals
02
Audience & search exposure
03
Newsroom & product build
What we measure
ILLUSTRATIVE · CLIENT DATA WITHHELD
Search referral index
Sessions from organic search, indexed to 100. Dotted tail is modelled.
Revenue exposure
Share of revenue dependent on search-sourced audience, by product.
Licence valuation range
Three methods, one archive. The spread is the negotiation.
Approach
Evidence first, then advice.
01
Measure
Four weeks inside your analytics, contracts, and cost lines. We establish what is true before anyone argues about what to do.
02
Test
Narrow pilots with a defined stopping rule. Each one answers a single question and is cheap enough to abandon.
03
Commit
A written position your team can take to the board, with the reasoning and the numbers left visible.
How an engagement runs
10—14 weeks
WEEKS 1—4
Measure
Analytics, contracts, cost lines. One baseline everyone agrees on.
WEEKS 5—10
Test
Narrow pilots, each with a stopping rule set in advance.
ADOPT
Commit
A written position, with the numbers left visible.
STOP
Abandon
Recorded and priced, so it is not retried next year.
Engagements
Selected work
Clients are named on request. Sectors and scope below.
Greater Baton Rouge Business Report
Valued a three-year archive licence and set the walk-away number.
LICENSING
425 Business
Rebuilt the traffic forecast around search-referral decline; repriced two products.
AUDIENCE
Biz New Orleans
Audited eleven proposed AI tools; approved three, retired the rest.
NEWSROOM
Journal of Business
Drafted editorial disclosure standards for synthetic audio and imagery.
GOVERNANCE
Writing
All notes
JULY 2026
What an archive is worth when the buyer already has a copy
Three valuation methods, and the one that survives due diligence.
JUNE 2026
Referral decline is not a marketing problem
Why the forecast, not the funnel, is where this shows up first.
MAY 2026
A stopping rule for every pilot
Pilots without an exit condition become budget lines.
Start with an assessment.
Four weeks, fixed fee, one written position at the end. If we do not think we can help, we will say so in the first call.