All Insights
Deep-dives on AI marketing, content systems, personal branding, and startup growth. Real frameworks from real usage.
The Expected Value of a Promo: Why Your $20 Discount Nets $500, Not $13,500
A discount is not a cost of $20. It is a contingent payout you wrote, and most of it subsidises customers who were going to buy anyway. Here is the arithmetic, the breakeven conversion rate you have to clear, and the Python that finds it.
Read the full post →134 posts
8 Reels, 41,423 Views, $0 Spend: The Build Log
Eight reels for a B2B automation services company did 41,423 views and 18,732 reach in five days with no ad spend. Here is the full per-reel distribution, the audience data, and everything the numbers do not prove.
Read article →NotebookLM Invented a Chart From Ad Panel Data
I fed six Meta Ads Manager screenshots from an agency account to Google NotebookLM. It returned a twelve-slide deck whose headline claim is a unit error and whose most persuasive chart was drawn from data I never supplied. Here is the audit, and the checklist that came out of it.
Read article →Hook Rate vs Hold Rate: What Actually Moves
Across three video creatives in one Meta ad account, hook rate spanned 1.22x. Hold rate spanned 2.65x as reported, and 2.18x once the hook stage is divided back out. The funnel arithmetic, the counterexample, and a diagnostic table.
Read article →Meta's Cost Per Result Is Not One Unit
Meta's "Cost per result" box measures a different thing depending on the campaign objective, and it never says which. Six panels from one ad account, reconciled by hand, showing a 1,000x unit trap.
Read article →Content hit rate: how many posts before it means anything
Half of a client's reels found distribution and half died. I turned that into a planning model, with runnable Python, that answers how many posts a month you need before the number stops being noise.
Read article →A Reel Got 20,415 Views. Zero Came From Followers.
Eight reels, 41,423 views, no ad spend — and the top reel's audience read 100.0% non-followers. The arithmetic on what cold reach buys you, and what it does not.
Read article →Perplexity freshness and the publishing cadence maths
What is actually published about freshness in AI citations, where the two biggest datasets disagree, and the arithmetic for when refreshing an old page beats writing a new one.
Read article →Streak mechanics for B2B retention: where the reward goes
Cost per saved account falls every single month, so the cheapest save is always the latest one. Under a fixed budget the month that saves the most accounts sits in the middle: month 4 in the worked model below.
Read article →The Content Pruning Score That Cut 267 Posts to 103
The exact five-input formula and thresholds I used to cut 267 published posts to 103, plus the merge-direction mistake that nearly destroyed 18,120 words.
Read article →Get Real Insights on AI + Marketing Every Week
Actual frameworks, not generic tips.