A Website Cannot Grade a Pallet

Let me show you the difference between a website and software, using one page.

A website tells you about a pallet.

Software takes your spreadsheet, counts the pallet, prices the pallet, tells you how confident it is in the answer, and then tells you not to trust it too much.

That last part is the one I want to talk about.

What the page actually does

Gideon Commerce is a marketplace I am building. One of its pages is for liquidation sellers, the people moving pallets of returns and overstock, and it is built around the manifest, which is the spreadsheet that comes with a lot and says what is supposedly in it.

The published flow is three steps.

Drop in a CSV and the tool auto-maps the title, SKU, UPC, quantity and retail price columns, and you fix anything it got wrong.

Then it grades the lot. It counts the units, totals the retail, estimates resale at a configurable recovery rate, and produces a confidence score that flags missing data.

Then it generates a draft lot listing with as-is disclosures and freight or local-pickup options.

That is not a page about liquidation. That is a job somebody used to do by hand at a kitchen table at eleven at night.

The number it gives you, and the number it refuses to give you

Here is what separates a real tool from a calculator with good marketing.

The worked example published on that page shows a sixty unit lot at $3,800 estimated retail, an estimated resale of $1,960, and a confidence score of 0.72.

Read that last one again.

Not a number. A number, plus how much the software trusts itself.

That is the most honest thing on the page, and it is the thing a brochure site structurally cannot do, because a brochure has no idea what you uploaded.

And directly under the tool, in its own box, the page says the estimates and confidence scores are directional, that you should always review a manifest before buying or listing, that Gideon does not guarantee resale value, and that liquidation lots are typically sold as-is.

Software that tells you how sure it is, and then tells you to check its work anyway, was built by somebody who has been wrong about a pallet before.

Three numbers for one pallet

Now look at what the same page publishes a few inches apart.

The grading example describes a sixty unit lot at $3,800 retail and estimates resale at $1,960. Further down that page, under live lots, sits a mixed electronics returns pallet of about sixty units, stated retail $3,800, asking $1,450.

I am not going to tell you those are definitely the same pallet. The unit count and the retail figure match exactly, and that is my observation, not a claim the page makes.

Take it as the lesson either way, because it is the whole lesson.

Retail is a story. The estimate is a guess with a confidence score attached. The asking price is the only one of the three with a human being standing behind it.

Anybody who has ever bought a lot knows those three numbers are never the same number. The tool that shows you all three is worth more than the tool that shows you one and acts certain.

The labels that do not match the math

Two things on my own site do not line up, and I would rather say it than let you find it.

First, the grading panel labels the resale estimate 45 percent. But $1,960 against $3,800 retail is closer to 52 percent, and 45 percent of $3,800 is $1,710.

The charitable read, and probably the right one, is that the recovery rate is configurable and the example is set to something other than what the label says, or that the estimate runs line by line off the manifest instead of as one flat percentage of the total. Either way, the label and the number disagree on screen. That is a fix, not a defense.

Second, and this is the third time I have written it down: the home page hero panel still shows a $2.00 deduction on a $100 sale and calls it a 1 percent fee. Two dollars is the free tier 2 percent rate wearing the paid tier label. The fee builder further down that same page gets it right and shows $1.00. I flagged this on August 18 and again on September 6. It is still there.

One more on that same page. The comparison section is headed on a $100 sale you keep $100, then shows Gideon at $96 sitting below two of the competitors, then claims about $7.20 more in your pocket than the average of the others. The three competitor bars as displayed average $95.67. That is three different numbers inside one section. Flagged September 6. Still live.

I am publishing my own product's broken labels on my own site.

That is not me being hard on myself for content. That is the standard. If I will not tell you what is wrong with mine, you have no reason to believe anything I tell you about yours.

What this is and what it is not

Gideon is a demo build. The footer says so. The listings are labeled demo and the sellers in them are not real sellers.

I am not claiming a single sale, a single user, or a single dollar of revenue, because there is not one to claim yet.

What I am claiming is the architecture. The manifest grader exists. The confidence score exists. The disclaimer exists. You can go read all three right now.

The lesson you can take without ever touching my marketplace

Most people asking me for a website actually need three pages and a phone number. I tell them that. I have told paying clients that.

But some people are sitting on a job they do by hand, over and over, that a piece of software could finish in nine seconds.

That is the tell. The question is not do I need a website.

The question is what do I retype, recount or re-estimate every single week.

That thing is not a website. That thing is your software.

And when you build it, build the confidence score too. Give people the number and how much you trust the number.

Anybody can publish an answer. Publishing your own uncertainty is what makes the answer worth something.