How do you compare vendors whose pricing models are structurally different?

ToolCrate is a marketplace where neighbors rent tools like drills and generators to each other. Tomasz Bellweather is the product manager deciding between two AI vendors for rental damage disputes. Odalys Fenwick handles those disputes by hand today. Wynn Castellano runs finance and asked to see the vendor comparison before it went to budget.

How to answer this, stage by stage

Nobody is scoring whether you can read a price sheet. They're scoring whether you can convert two different price sheets into one number you can actually compare.

Let's learn

For three years, Odalys Fenwick was the only person at ToolCrate who could tell real damage from bad lighting in a photo. She compared before-and-after pictures of rented tools by hand, about twenty five minutes per flagged case, and closed roughly nine hundred confirmed damage claims a month across the whole marketplace.

This is the job either vendor is meant to speed up. Neither one changes what actually gets decided.

Two vendors pitched AI photo review to replace the manual comparison. Scuffguard charges thirty five cents for every photo its model analyzes. Ledgerlens charges a flat twenty four hundred dollars a month, plus eighteen dollars for every claim it confirms as valid.

At its worst, comparing sticker prices instead of converted costs means picking the pricier vendor while believing you picked the cheaper one. Building this comparison honestly costs about a day of Tomasz's time. Skipping it risks locking in a vendor that costs five times more than the alternative, every month, for as long as the contract runs.

What I would leave alone: when two vendors price the exact same way, say both charge a flat monthly fee for the same feature set, reading the smaller number really is the whole comparison. The conversion step only matters once the pricing shapes actually differ.

The lesson: a cheap-looking price and a cheap price are not the same thing. The only way to tell them apart is to run your own volume through both equations before you sign anything.

Now here is the same thing as a story

The short version above is what you'd say defending this pick to ToolCrate's budget review. Read this one for how the comparison actually got built.

Tomasz Bellweather had run four vendor searches before this one, and every previous comparison had been easy: line up the flat monthly fees, pick the lower number, move on. He expected this one to go the same way.

Two shapes, not two numbers. That's what made the old habit stop working here.

Scuffguard's quote read thirty five cents per photo. Ledgerlens's quote read twenty four hundred dollars a month plus eighteen dollars per resolved claim. Read as sticker prices, Scuffguard looked obviously cheaper, and Tomasz nearly wrote it into the recommendation memo that way.

Wynn Castellano, ToolCrate's finance lead, pulled the comparison sheet ahead of the budget meeting and asked one question: "What does nine cents even mean here, and where did that number come from?" Tomasz realized he'd never actually run ToolCrate's own volume through either vendor's formula.

Neither vendor's quote mentions a single one of these four numbers. They're the whole comparison.

Tomasz had considered simply asking each vendor for "an average customer's monthly bill" instead of building the comparison himself. He dropped that idea. A vendor's reference customer has its own flagged-to-claim ratio, not ToolCrate's, and their quoted average would quietly reflect someone else's marketplace, not this one.

He ran the real numbers: forty thousand rentals a month, twenty six hundred flagged for review at four photos each, nine hundred confirmed claims. Scuffguard's cost came out to three thousand six hundred forty dollars a month, all from photo fees. Ledgerlens's cost came out to eighteen thousand six hundred dollars, the flat fee plus nine hundred claims at eighteen dollars each.

The answer isn't fixed. It depends on which of ToolCrate's own numbers moves next.

The gap held against a sanity check too: ToolCrate earns about six dollars twenty cents in commission per rental. Scuffguard's cost is about one and a half percent of that. Ledgerlens's is about seven and a half percent, higher, but not an absurd number for a fraud and dispute tool to cost.

BOUND, the estimate that shows its own arithmeticNot a gut call on which price sheet looks friendlier. BOUND is what makes a pricing comparison checkable instead of just confident.

Each vendor's real risk is which of ToolCrate's own numbers it's most exposed to.

The recap, one line per letter: break it down is stating cost as a flat fee plus a per-unit price times real volume, own the numbers is naming ToolCrate's actual rental, review, and claim counts, use a range is admitting the claim rate moves month to month, nail the sanity check is comparing both costs to commission per rental, and direction is naming the claim rate as the lever that actually decides the winner.

And if you want to be sure it really works, try it somewhere elseSame five letters, a fishing cooperative instead of a tool rental marketplace. A different old decision breaks the second estimate.

Herringcove Fishing Cooperative runs dockside cameras that estimate catch weight from photos instead of a hand-tally at the scale. Halden Voss, who runs cooperative operations, was comparing TallyNet, priced per image processed, against Wharfmetric, priced per ton processed plus a flat fee that rises with how many docks run the cameras. Mapped onto BOUND: break it down is cost equals TallyNet's per-image rate times images captured, versus Wharfmetric's flat dock fee plus its per-ton rate times tons weighed. Own the numbers is assuming five docks, about eleven thousand images a month across all of them, and about four hundred tons of catch weighed monthly. Use a range is admitting the tons-per-dock figure varies by season, not a fixed number year round. Nail the sanity check is comparing either cost against the cooperative's own per-ton dock fee revenue, to see if either estimate is plausible. Direction is naming the number of docks running cameras at all as the single biggest swing, since Wharfmetric's flat fee scales with dock count and TallyNet's does not.

Herringcove faced the same two shapes ToolCrate did, just wearing different names.

The old decision here isn't a side-by-side sticker comparison, it's a different reversal: Wharfmetric's dock-tiered flat fee pushed Herringcove to run cameras at only two of its five docks, to keep the flat fee down, which meant the other three docks kept using hand tallies with no photo record at all. That made sense as a way to control cost early. It stopped making sense once those three unmonitored docks became the ones with the least reliable catch data in the whole cooperative.

Swap the trigger and it still runs.
Speed: an interviewer caps you at sixty seconds. Say "convert both to cost per unit of your own real volume, then name the swing factor," and stop.
Cost: there's no time to gather a full year of volume data before the decision is due. Say so honestly, and use the best three months you have, flagged as a range, rather than presenting a guess as a fixed number.
The model gets better, for real: if Scuffguard's model gets sharper and flags fewer photos for review, that's still worth re-running the comparison for, since a lower review rate lowers Scuffguard's own cost without changing Ledgerlens's at all.

Where people run it wrong.
They compare sticker prices instead of converting both to the same unit first.
They borrow a vendor's reference-customer numbers instead of running their own real volume through the equation.
They never name which single usage number would flip the comparison, so the pick never gets re-checked as volume changes.

How to use it live. The moment someone hands you two vendors with different pricing shapes, ask: what's the one unit both of these can be converted into, using our own numbers, not theirs? Let that answer decide, not whichever price sheet reads smaller.

Flashcards (tap any card to flip it)

Check yourself Score: 0 / 0