From the team
Field notes
Working theory and hard numbers on visitor identification, B2B pipeline, SaaS pricing, and building with AI. Written by the people doing the work, not a content team.
No posts match that yet. Try another word, or ask us directly.
You Can Make Forty Ads. You Can Test Three.
AI made creative production nearly free. Measuring it did not get cheaper. How many ads your traffic can actually settle, and what to do with the rest of them.
The Report Nobody Reads on Monday
Most ad account problems start midweek and get found on Monday. How to set up a daily check that only speaks up when something changed, where AI actually helps, and the one alert most setups miss.
The Model That Can't Look Wrong
A new kind of AI model shipped last week that does not write anything. It returns a decision in about 100 milliseconds for a fraction of a cent, which makes the 39,800 rows nobody reads affordable to judge. It also takes away the thing that used to tell you a model was wrong.
The Window Was the Difference, Not the Ad
Two campaigns, one clearly won, so you moved the budget. The only thing that changed was the attribution window. Meta's own documentation says a 7-day window counts sales a 1-day window drops, and that you cannot fairly compare across the two. You ranked two rulers, not two ads.
The CPM Went Up and Nobody Changed the Ad
Same creatives, same audience, same page, and CPA up 12%. Meta's filings put average price per ad up 9% in 2025 and 12% year over year last quarter, for reasons that have nothing to do with your account. What a price increase does to CPA, and why holding it flat is a request for a 12% better funnel.
The Lead That Had No Website
Three of this morning's five leads typed a domain that does not resolve, and they cost exactly what the real ones cost. At 60% junk you are paying $125 a lead while the dashboard still says $50. The arithmetic, the ten-second checks, and where the bad ones come from.
The Answer Didn't Come From Your Website
Ask an assistant to recommend software in your category and it answers with confidence, from a handful of roundups and threads you do not own. Why your homepage is close to useless as a citation, and where the work actually is.
The Sale Needed Twice the Orders
You ran 20% off and orders came in 30% above a normal week, so the promo worked. On a 60% margin product with ad spend in the math, it needed 133% more orders to break even. The arithmetic, and how to test yours inside one promo.
The Winner Was a Coin Flip
One creative came back at 3.1%, the other at 2.4%, so you scaled the winner. At that volume the gap was inside the noise. What a test actually needs, and why checking it every morning manufactures winners out of two identical ads.
The Research That Used to Be a Hire
A vendor puts a number on a slide and you write it down, because checking it used to be an analyst's job or nobody's. It is now an afternoon. Three claims you can verify yourself this week.
The Lead You Buy Twice
Your cost per lead looks good, so you ask for more budget. The number leaves out the most expensive thing you paid for. Here is the napkin math on your own spend, and what to do with the half the report hides.
Your Tags Fire After They Leave
Your ad platform counted 10,000 clicks. Your analytics counted 7,200 sessions. The gap is not fraud, it is a consent banner nobody clicked. Here is how to measure it on your own site in two minutes.
The Channel You Just Cut Was Working
It posted a bad CPA, so you paused it. Three weeks later blended performance is worse and the report has no idea why. Here is what it could not see, and how to test a channel before you cut it.
The Price on the Pricing Page Is a Floor
In this category the sticker is a starting point, not a total. The real number gets built from annual minimums, per-use overage, and add-on modules you never see until the invoice.
The Lead Was Warm for an Hour
Speed to lead is a religion in this business. Nobody mentions that the fastest follow-up on earth is still too slow for the buyer who never became a lead.
The Quotes That Started and Vanished
Someone wanted a quote badly enough to begin one. The form never got their name. That lead existed, and nobody got to sell it.
The Abandoned Carts You Can't Email
Your recovery flow works fine. It just never meets most of your abandoners, because they never told you who they are.
The Leads You Already Paid For
You are paying twice for the same buyer. The visit already happened. The only thing missing is the name.
The Shoppers on Your Lot You Never Meet
Most of the people shopping your inventory leave your website without a word. Here is a quiet way to turn more of them into sold cars, using the traffic you already have.
The 3% You Kept
You pay to bring everyone to your site, then build your entire business around the tiny fraction who raise their hand.