Blog
What we have learned watching companies change. Written by us, about patterns, not about any particular company.
Start here
The long ones. Each stands on its own and the rest of the site links into them.
- AI agents in go-to-market: what gets cancelled, and what survivesGartner expects over 40% of agentic AI projects to be cancelled by 2027. The dividing line is how much authority the agent has to act.Sources read 2026-09-09
- What decides whether ChatGPT names your companyThe research is public and predates the question being asked. Statistics, quotations and named sources, in that order.Sources read 2026-09-09
- What X’s open-source algorithm actually rewardsThe weights are published. A reply is worth 27 likes, a report costs 738, and there is a slop score in the code.Sources read 2026-09-08
- LinkedIn, AI slop, and what changedFour fifths of long-form posts now read as AI. LinkedIn shipped a button to report them, and a model that reads.Sources read 2026-09-08
- The quiet weeksThe weeks you shipped and nothing went out. Why they cluster, and what they cost.Start here
- How to choose a go-to-market tool when you are the whole teamFour questions that decide it, and the one most comparison tables never ask.Start here
Buying window patterns
What a company does in the weeks before it buys, and how the sequence reads.
- Sales-led to product-ledThe pricing page loses its demo gate, then someone gets hired to own activation. That order matters.Buying window patterns
- Moving upmarketSecurity pages, SSO, and an account executive job posting in the same quarter.Buying window patterns
- Replacing infrastructureA tag disappears. That is a decision someone had to defend in a meeting.Buying window patterns
- Building a lifecycle functionA head of lifecycle gets hired, and buys their tooling in the first quarter.Buying window patterns
- Expanding sales capacityFunding, then three sales hires, then a CRM. The first seller has no playbook.Buying window patterns
Method
How any of this is measured, and where the measurement misleads.
- Why stacked signals lieThree unrelated facts about a company can be arranged to look like a story. That is the main failure mode of signal scoring.Method
- What a buying window is notIt is not a score, not a fit, and not a person visiting your pricing page.Method
- How we measure whether any of this worksA slice of every account's budget runs three dumber versions of the same idea.Method
Hypotheses
Shapes worth testing, written so a result means something either way.