Claude Penland

By Claude Penland - marketing and business strategy for companies that are good at what they do and hard to find.

The Catalogue of Ships: Why the numbers in your drive are worth more given away than kept

Book Two of the Iliad stops the war cold so Homer can list the ships. Who sailed, from where, how many hulls, under whose command. It is the passage every reader skips, and it is the passage scholars have been citing for three thousand years, because it is the one place the poem quits singing and starts counting. The war reads as real because somebody did inventory.

That is the whole argument. The rest of this is just me making it inconvenient for you.

1. The research keeps finding the same two levers

The academic work on generative engines did something refreshingly dull. Researchers took source pages, rewrote them nine different ways, pushed them through the systems that write answers, and measured which rewrites got pulled into those answers more often. Keyword tricks did essentially nothing. Two edits did nearly all of the work: real statistics with sources, and quotes from named humans with credentials. Visibility moved by double digits, and it moved hardest in the crowded categories where every vendor already sounds like every other vendor.

The reason is not mysterious. A model composing an answer needs somewhere to put specificity. It has four hundred interchangeable ways to say your industry is changing fast. It has exactly one way to say that dwell time at grocery distribution centers in the Southeast runs 2.7 hours against a 45-minute appointment window – and that sentence has your name welded to it.

2. Most โ€œproprietaryโ€ data is just unopened

Every operator I meet has a folder. Warranty returns by SKU. Denial codes by payer. Dwell times by receiver. Reorder intervals going back to the Obama administration. The reflex is to call it proprietary. Usually it isnโ€™t proprietary, itโ€™s just unpublished. Proprietary means a rival could beat you with it. Unpublished means nobody can use it at all, including you, while it quietly rots in a drive nobody has opened since the last CRM migration.

There is a parable about a servant who buries the money he was given. He is not rebuked for losing it. He is rebuked for making it inert. Sitting on data is not a neutral act.

3. Go dig. Here is where to put the shovel.

Stop hunting for somebody elseโ€™s study to cite and go count something only you can count. A few I would pay money to read:

  • A freight brokerage already knows detention hours by receiver type and region. Nobody has ever published what an average dock actually costs a driver. That figure would be quoted in logistics answers for the next five years.
  • A dental group has fifteen years of charts. What share of patients who skip two consecutive cleanings need a crown within five years? The whole profession is guessing. You have the answer sitting in your practice management system.
  • A commercial insurance broker can say which kitchen layouts generate fryer burn claims, and at what rate per thousand covers served.
  • An equipment dealer in the Corn Belt can put bushels on each day of planting delay, banded by soil temperature. Growers argue about this at the co-op every spring with nothing but folklore.
  • A medical billing shop knows denial rates by payer and code. Better yet, it knows what share of denials come back paid on first refile – a number that is worth a keynote.
  • A regional staffing firm knows time-to-fill by metro and role. Every CFO in the country is currently estimating that number badly.
  • A residential mover knows breakage by packing method. Publish it and you have written the definitive citation on whether dish barrels are worth the upcharge.

None of that needs a data science function. It needs one person with database access and a free afternoon.

4. Then put a human beingโ€™s name on it

The second lever is a person, not a brand voice. A named human with credentials – a title, a tenure, twenty-two years on loading docks, licensed in four states, board-certified in something. These systems learned from a corpus in which credentials predict reliability, so they reach for the sentence somebody is accountable for. An unattributed claim from a company is marketing. The same claim from your named VP of operations is testimony. Identical words, completely different gravity.

5. The giveaway is the distribution

The Rosetta Stone is a tax decree. Bureaucratic housekeeping about temple revenues, of no importance to anyone. It survived because it was carved in public, in three scripts, in a place where strangers could read it. The confidential decrees of that era are gone. The one they gave away unlocked a language.

You will not get distribution by parking the good stuff behind a form. A machine cannot fill out your form, which means every gated PDF is now invisible to the thing your buyer is asking. And this race has a very short lead: the first credible number on a topic becomes the anchor, gets restated, gets cited by people citing the people who cited you, and hardens into the default answer. Second place gets summarized as โ€œsome estimates suggest.โ€

6. Legal will say no. Here is the yes.

The answer is almost always the same three moves: aggregate, lag, anonymize. Nothing below two hundred observations. Ninety days delayed. No client names, no geography tighter than a metro. State your method plainly – sample size, window, what you excluded and why. That is a defensible dataset, and frankly it is more rigor than most of what your industry currently passes around as a benchmark.

7. Publish the dog that didnโ€™t bark

Holmes cracks one case on the fact that the dog stayed quiet. Negative findings are findings. The retention tactic that moved nothing. The material that failed in salt air. The segment that never converted no matter what you spent. Almost nobody publishes these, which is precisely why they get quoted – a system trying to give a balanced answer has almost nothing to reach for on the โ€œit didnโ€™t workโ€ side of the question.

Last thing

Borges imagined a library containing every possible book and made the quiet point that total information is indistinguishable from noise. That is the corpus these machines are reading now. Infinite shelves, mostly interchangeable. The way out is not to write more. It is to be the one shelf with a number on the spine, a name underneath it, and a date.

Go find your ships. Count them. Publish the list.


Discover more from 1000 Startups

Subscribe to get the latest posts sent to your email.

Claude Penland

Claude Penland builds the marketing and business strategy for companies that are good at what they do and hard to find. Thirty years operating, one exit, eight of them as a practicing casualty actuary.

The free two-page read is genuinely free. Email claude@1000startups.com and I'll send back what I can see from the outside. Or see the work samples and how to work with me.

Leave a Reply