Talk about a build

Insights

Writing on measurement, evidence
and competitive standing.

Methods and arguments we use in engagements, written out so you can judge the thinking before you talk to anyone.

Podcast intelligence: strategy stated in a competitor's own words

Executive interviews are the least guarded thing a competitor produces. They are also transcribed, dated and public. Where a category has podcasts, this is the highest-signal lane available.

Read

Every company in the market scored at par on conversion

Not one competitor offered a buyer any way to get a specific answer without waiting on a human. A ranked leaderboard would have declared a winner in that column and hidden the largest opportunity on the board.

Read

Our client was winning every category. That was a bug.

The first live competitive audit had our client leading every lens. It looked like a great result. It was uneven research, and the fix dropped them to second. We shipped the honest number and built a gate so it cannot happen again.

Read

Turning the quarterly review into a strategy meeting

A QBR spent defending spend is a QBR your client experiences as a cost. The same hour spent on what the market did and what to do about it is the meeting that renews the retainer.

Read

What to ask your agency for, in one page

Nine questions that will tell you more about the quality of your marketing reporting than any dashboard review. None of them are adversarial and a good agency will welcome all nine.

Read

95% of your market is not buying today. Measure accordingly

If most of your market is out of market at any moment, then a measurement system that only counts people raising their hand is describing a small and unrepresentative slice of the thing you care about.

Read

The llms.txt question, answered with the evidence

We publish an llms.txt file and we will tell you plainly that the evidence says it does almost nothing today. Here is the data, and here is why we ship it anyway.

Read

Are you cited in AI answers? How to check, honestly

AI answers are non-deterministic, which makes "we do not appear in ChatGPT" a much weaker claim than it looks. A testing protocol that produces a defensible finding rather than an anecdote.

Read

Why we let a human overrule the adversarial checker

We measured what would happen if our skeptical second pass ran automatically. It would have demoted the strongest competitor most and handed our client back first place. A checker tuned to argue, applied mechanically, recreates the bias it was built to remove.

Read

We audited our own gate and it caught 2 of 7 fabrication classes

Quote verification is necessary and nowhere near sufficient. Here are the five fabrication classes it cannot see, why every one of them passes a byte-for-byte check, and what we built as a result.

Read

Negative controls for AI research

A verification step nobody tests is a verification step you are trusting on faith. Plant a deliberately false claim on every run, and fail loudly if it ever gets accepted.

Read

An AI analyst that cannot contradict the dashboard

Give a language model its own database access and it will eventually quote a number that disagrees with the chart above it. The fix is architectural: make it read through the same code that renders the tile.

Read

The grounding gate: how to stop an AI research pipeline inventing quotes

The dangerous failure is not invention from nothing. It is a quote that is ninety percent right and ten percent improved. Byte-for-byte verification against a cached source, with rejection rather than flagging.

Read

The sales solicitation problem, and why the excluded count belongs on the page

Every business with a public phone number receives vendor pitches through the same channels as real enquiries. Excluding them is obvious. Showing how many you excluded is the part almost nobody does.

Read

Blended ad spend hides the exact thing you are comparing

A monthly average spread evenly across days is fine until you compare two date ranges, which is the only reason anybody looks at spend. Range-aware spend costs more to compute and is the only version that survives a question.

Read

Reading a competitor's ad library without fooling yourself

Ad transparency records are the best free competitive intelligence available and the easiest to over-read. What they actually prove, what they cannot show, and the findings worth extracting.

Read

Does not rank is not the same finding as does not exist

Ranking data can tell you a competitor is invisible for a term. Only a direct sweep of their own site can tell you they do not offer the service. Confusing the two sends you after entirely the wrong opportunity.

Read

Capacity or expansion: two hiring signals you must never merge

Ten field technicians and three senior leadership roles are completely different pieces of news. Averaging them into an open-role count destroys the only finding worth having.

Read

What a competitor's job postings tell you months before their advertising does

Headcount is the earliest reliable signal of intent. A company hiring for a capability is committing budget to it long before the campaign that announces it, and job boards are public.

Read

How to build a competitor set you will actually learn from

Most competitive analysis fails at the roster. If you audit the companies you admire rather than the ones you lose to, every finding afterwards is about the wrong market.

Read

Why there is no pricing page

Every engagement is quoted after a conversation, and there are no prices anywhere inside the product either. That is a deliberate constraint with a reason, not a sales tactic to make you call.

Read

Absolute scoring beats a ranked leaderboard

A ranking always produces a winner, because sorting always terminates. An absolute standard is willing to report that nobody in a market is doing well, which is exactly when the opportunity is largest.

Read

Why we publish the scoring rubric where competitors can read it

A standard that only works while it is secret was never a standard. Publishing the rubric costs us a defensible-sounding moat and buys something considerably more useful: the ability for a client to hold us to it.

Read

First touch or last touch: which channel actually brought them in

Last-touch attribution is the default nearly everywhere, and it systematically rewards the channel that was already going to close while starving the one that created the demand. Here is how that plays out over a budget cycle.

Read

A lead is a person, not a touch

Deduplication removes roughly 30% of raw contacts in live marketing data. Here is exactly where that third comes from, why it is not a tracking bug, and what a defensible lead definition looks like.

Read

The three numbers in your marketing report that are usually wrong

Lead count, cost per lead and channel credit go wrong in almost every marketing report, without anybody lying. All three fail the same way, and all three fail in the direction that flatters whoever produced the report.

Read

Find out what your marketing
is actually achieving.

One conversation, no obligation: what you sell, who you lose deals to, and what you cannot currently see. If a build is a fit, it is scoped and quoted from there. There is no signup and no self-serve tier.

No prices on this site. Every engagement is scoped and quoted after the first meeting.