Harbor — the spine

HARBOR · THE WALKTHROUGH · STEDMON HARPER · 17 SEP 2026

00 WHAT I FOUND

They tell you everything. Just not until after you have bought.

Nothing is hidden. It is just late. When you decide, you know very little. Five minutes after you pay, you know almost all of it.

01 HOW I LEARNED IT

One day of field work. Four people, two live accounts, three websites.

4 conversations 2 real insurance accounts 3 insurance websites 1 paper bill 6 guesses written down first

A claims adjuster of forty years. Someone who switches often. Someone who has stayed with one insurer for years. Someone who buys everything through an agent. Then I read two live insurance accounts from one household, every page, and drove three insurance websites end to end.

Every claim on these pages is labeled. Observed means I saw or heard it. Inferred means I worked it out, so you can argue with it. Uncertain means I do not know.

02 WHAT I EXPECTED TO FIND

Six guesses, each with the thing that would prove it wrong.

I wrote them before I spoke to anyone, so I could not talk myself into them later. Here they are unedited, including the one that turned out wrong.

People buy insurance because they have to, not because they want to — so the normal rules of shopping may not apply	No verdict. I never got a clean read on this one, and I am not going to pretend I did.
People who start a quote and vanish have already bought — somewhere else	Right. Three people in my record never even said no. They just stopped answering.
The fear is not price. It is losing coverage you cannot check	Right. And nobody can check it on their own.
Below a certain saving, people stay out of habit	Right. One switcher named the number: $100 a month.
Staying loyal pays, and your history does not transfer	Wrong. I based it on a household I thought had tried driving-tracker insurance and left anyway. They never had it. I found that out by checking my own facts.
People shop when something happens, never on a calendar	Half right. One man checks every six months and has never moved.

The one I got wrong is still on the list. Anyone can show you the guesses that turned out right. Leaving the wrong one in is the only way you can tell I wrote these before I started, and not afterwards.

03 BEFORE AND AFTER, IN ONE ACCOUNT observed

The same insurer, on either side of the moment you pay.

Before you decide

“Covered up to your policy’s limits.” A short list of obvious exclusions. And a nine-question tool that would tell you what coverage you actually need — parked on a marketing page, outside the buying flow.

The day after you pay

Every coverage listed, with its limit and its own price. Every fee, every letter, every price change, logged and dated. It is genuinely good.

And they can explain things very well. They just point it elsewhere. One unpaid balance of a few hundred dollars produced seven chase messages and a referral to debt collectors. Over the same months, at either insurer, messages explaining my coverage: none.

04 THE BEST ARGUMENT AGAINST ME observed

An agent already does this job, and you cannot out-trust a person.

“You’re missing the interpersonal.”

A man who runs six policies across three households through one agent — “because we just trust them”

He is right twice. The gap is real, and somebody already sells the answer to it: an agent. He cannot compare two policies without his. He switched insurers when that person changed jobs.

So this has to beat a phone call to someone you trust, or be built for people who do not have one. Three of my four buy direct, off a screen, with nobody in the room.

05 WHY THIS AND NOT THE OBVIOUS FIXES inferred

Three easier answers, and why I set each one aside.

Shopping is not too hard. Nobody described it that way. One man switched insurers in a single day. Another has looked every six months for years and never moved.
It is not secrecy either. That is the finding above — they publish nearly all of it. Which means saying more is not the fix. Saying it earlier is.
And it is not the money. Price is what makes people look. It is not what makes them move. Every switch in my record had a different trigger: a denied claim, new drivers in the house, an agent changing jobs.

That is the gap this explains. A man who checks the price every six months has never once switched. He can compare the number. He cannot compare what the number buys.

06 WHAT I WOULD DO ABOUT IT

Show people the difference before they buy.

Put the policy someone already has next to the quote they are being offered. Line by line, in their own numbers. And say plainly which parts it could not compare.

The part that matters

It is allowed to come back and say: nothing here needs to change. A comparison is only worth anything if it can come out against the company showing it to you.

How it would work. You sign in to your current insurer, the same way you connect a bank account to a finance app. That plumbing already exists and reaches almost every insurer in the country. Prefer not to sign in? A photo of your policy page works too.

When: after you see a price, before you pay Cost: two people, a working version in days One person owns it — including the times it says no

07 HOW WE WOULD KNOW INSIDE 30 DAYS

Two questions, in order, and how we would run them.

First question: will anyone actually connect their policy? The whole idea depends on someone signing in to their current insurer. Nobody I spoke to said they would, and I did not measure it. So that gets measured first. If people will not connect, nothing else matters.
Second question: does seeing the difference change what people do? Find people who are shopping. Show half of them the comparison. Show the other half nothing. See whether the two groups behave differently.
How we run it: we recruit people rather than wait for them. Harbor is new and has no traffic. Waiting for enough shoppers could take months. Recruiting means we choose how many and how fast — and we build the working version while the test runs, because that part now takes days rather than months.
And if too few people take part, it says so. Under about sixty, the answer is we do not know yet. Not a pass, not a failure. A test that can only ever agree with me is not a test.

08 WHAT WOULD MAKE ME WRONG uncertain

Four other explanations I could not rule out.

Maybe it is just expensive and risky, and nobody chose anything	Chasing money you are owed is a product you buy off a shelf. Explaining coverage is done case by case. It changes state to state. And it sits right next to advice you need a license to give. This is the one I would bet against myself on.
Maybe the law keeps the helpful tool out of the buying flow	Recommending a coverage level to a named person needs a license. But both insurers publish that tool openly. Only its placement differs. A legal problem would ban it, not move it.
Maybe explaining it changes nothing	Everything I found shows that nobody explains it. Nothing I found shows that explaining it would move anyone. That is exactly what the first test is for.
Maybe people could compare fine — it is just not worth the afternoon	Then I have added work to the moment I am trying to unblock. My answer: signing in once is not an afternoon. Whether people accept even that is what the first question measures.

09 WHAT I AM CHOOSING NOT TO DO

Three things I could have gone after and did not.

Pricing, regulation and advertising are not on this list — you ruled those out yourselves. These are mine.

Making the quote itself better	The part Harbor already owns, and the part every competitor is already polishing
Home insurance	Usually bundled into a mortgage payment, so it behaves differently. I have one data point and it is inertia, not a test
People who buy through an agent	Every screen I looked at is direct-to-customer. One of my four sees none of them

10 WHAT THIS MEANS HARBOR SHOULD BE open

Something that helps you leave cannot be owned by a company that needs you to stay.

That rules out every established insurer here. And it would rule out Harbor too, once Harbor had customers of its own.

And what builds up is not the software. The code is a weekend. A year of real policies gives you a record of what each insurer writes and what it means — which only exists if you do the volume. That is what would stop someone copying it.

So my answer is not a feature for an insurance company

I do not think Harbor should try to be another insurer. It should be the thing that holds your record of what you are covered for, and lets you take it with you. Everyone else in this market needs you to stay. That job is open. And the plumbing to build it now exists.

Two things I cannot answer. If the insurers receiving switches pay for that record, whose side is it on? I have an argument I would like to be true, not an answer. And I have no numbers — nothing I did tells me what an insurer would pay. One hour with someone who runs an insurance operation would settle both. I did not spend it.