Polestar Solutions

Field Notes

Scoring That Bends to the Preferred Answer

When scoring criteria arrive after the proposals, evaluation stops being objective and starts justifying a decision already made. Here is how to fix the weights at intake.

Key points

  • But once the proposals are open and one vendor's console is visibly slicker while its cost is 15 percent higher, the pull is to nudge the console weight up and the cost weight down, just enough.
  • A response that looks strong relative to two weak competitors can still sit at the 40th percentile of what the market delivers.

Why deferred criteria always bend

A scoring model is a set of weights. Weights are opinions about what matters. The instant those opinions are formed while the prices and features are visible, they are contaminated, because the human brain is very good at finding reasons for a preference it already holds. This is not a character flaw in your team. It is the default behaviour of any evaluation where the standard is set after the evidence.

Consider a simple example. Say a committee is choosing between two data platforms. Before quotes arrive, most people would agree that reliability and total cost matter more than, for instance, the polish of the admin console. But once the proposals are open and one vendor's console is visibly slicker while its cost is 15 percent higher, the pull is to nudge the console weight up and the cost weight down, just enough. Each nudge feels justified in isolation. Together they flip the ranking. The score now says what the room wanted it to say.

"A score set after the evidence is not an evaluation. It is a rationalisation with a total at the bottom."

PART TWO

Why the problem survives every process you put around it

Most organisations already have a procurement policy that says criteria should be agreed up front. The problem persists anyway, for three structural reasons.

First, timing. Intake usually arrives as a request to buy a named thing, not a problem to solve, so the criteria conversation feels premature. We wrote about that pattern in the intake ticket that already names the vendor. When the request is a product name, nobody thinks to argue about weights until the RFP responses force the question. By then it is too late to be neutral.

Second, effort. Building a genuinely weighted, defensible scoring model from scratch is real work, and it is work with no visible output until much later. Under deadline, teams skip it and promise to build the grid when the proposals arrive. That promise is where objectivity dies. It is the same failure mode as setting the budget before anyone checked the market price: the decision that should anchor everything gets made last, under pressure, with the wrong information.

Third, and most quietly, deferred criteria are useful to whoever already knows what they want. A blank grid is a gift to the strongest opinion in the room. If your criteria are fixed and public before the quotes land, that person has to argue about weights honestly, in the open, before they know which vendor a given weight favours. That is exactly the argument they would rather avoid, which is why the grid stays blank.

app.isvcosell.com/benchmarks

The benchmark library supplies the standard criteria and weights before any proposal is opened.

THE SAME JOB, TWICE

TODAY, BY HAND

Wait for the three proposals to arrive, then open a blank spreadsheet to build a scoring grid

Argue the weights in a room where everyone can already see the prices and features

Score each vendor by hand, adjusting weights whenever the ranking looks wrong to someone senior

Draft a justification memo that explains the final total after the decision is effectively made

Roughly 14 hours, spread across two weeks and three meetings

WITH ISVCOSELL

At intake, pull the standard weighted criteria for the category from the benchmark library

Confirm or adjust the weights with the committee before any quote is visible, and lock them

Load proposals against the fixed grid so each is scored on the same standard

Review the ranked output with percentile context, prices revealed only after scoring is set

About 40 minutes of your attention

What changes: 14 hours of grid-building and defending becomes about 40 minutes of confirming a standard. Across an evaluation team that runs, for example, two sourcing events a month, that is roughly 28 hours reclaimed monthly, and more importantly the score now measures the vendors instead of the room.

PART THREE

The platform motion: fix the standard at intake

ISVCOSELL treats the scoring model as an intake artifact, not an end-of-process one. When a category is opened, she proposes a weighted criteria set drawn from the benchmark library rather than a blank sheet. Those weights are not invented for your event. They reflect what actually differentiates outcomes for that category across a wide base of transactions, so the starting point is a market standard, not one person's preference.

The committee still owns the weights. The difference is when they own them. You confirm and adjust the model before a single proposal is opened, and the model is locked once quotes are in flight. Prices are held back from the scoring view until the weighted evaluation on non-price criteria is complete. That sequence, standard first, evidence second, is the entire mechanism. It costs you almost nothing and it removes the room in which scoring bends.

Because the criteria are benchmarkable, each vendor's answer is scored not just against the others in your shortlist but against the wider distribution. A response that looks strong relative to two weak competitors can still sit at the 40th percentile of what the market delivers. That context is what stops a narrow field from flattering a mediocre winner. If your requirements themselves are the problem, that is a different failure worth reading about in the requirements document that collapses your shortlist to one.

app.isvcosell.com/benchmarks/data-platform

Each criterion carries percentile bars, so a proposal is scored against the market, not just the shortlist.

"Lock the weights before the prices, and the strongest opinion in the room has to make its case in the open."

PART FOUR

What changes when the standard comes first

1 The weights argument happens blind. You debate what matters before you know which vendor each weight helps. That is the only condition under which the debate is honest, and it is the whole point.

2 Prices are revealed last, not first. Non-price scoring is completed against locked criteria before the commercial numbers enter the view, so cost cannot silently reshape the other weights.

3 Every score carries market context. Percentile bars mean a winner has to be good in absolute terms, not merely the least weak of three. A narrow field can no longer manufacture a strong result.

4 The justification writes itself. Because the model was fixed and logged at intake, the audit trail already exists. There is no after-the-fact memo reverse-engineering a total, because the total was produced by a standard nobody could see the answers to.

5 The incumbent gets no free pass. When criteria are set from market benchmarks rather than the last contract, you avoid the trap in the spec that already picked the winner. Read more in our note on the requirement written to fit the incumbent.

PART FIVE

What this does not fix

Be honest about the limits. Fixing criteria at intake removes the mechanical way scoring bends. It does not remove human intent. A committee determined to reach a predetermined answer can still weight a criterion aggressively at intake, before the quotes, and defend it. The gain is that they now have to do so in the open, with market benchmarks visible as a counterweight, rather than quietly nudging numbers once the answer is in view. That is a large improvement in accountability, not a guarantee of neutrality.

Nor does a good scoring model tell you whether the requirement was right in the first place. If the underlying need is wrong, a rigorous, well-weighted evaluation will simply select the best answer to the wrong question. Benchmarks discipline the scoring, not the strategy behind it.

And it does not stop a vendor's terms from shifting after you score. Price lists move, and the winning proposal you scored in March may not be the offer on the table in June. That is a separate watch problem, one we cover in catching a price list change the week it happens. What fixing criteria at intake does give you is a clean, defensible, market-anchored score, produced before anyone saw a price, which is the one thing a room with a favourite can never talk its way around.

FF

About the author

Fredrik Filipsson, Cofounder, ISVCOSELL

Fredrik has spent more than twenty years in enterprise software, with time at Oracle, IBM, SAP, and Salesforce before moving to the buy side. He structured and priced the kind of large agreements most buyers only see once or twice in a career, which taught him where the leverage sits and how far a vendor will actually move. He started ISVCOSELL to hand that knowledge to every sourcing team.

More posts by Fredrik Connect on LinkedIn →

See it in the product

How benchmarking works → Browse the use cases → Every feature → Calculate your time saved →

FREE TRIAL · FULL PLATFORM · NO CARD REQUIRED

Fix your scoring before the prices arrive

The free trial opens the benchmarking database, 1,483 vendors deep, plus the negotiation guides, playbooks, and talking points for your own renewals. No card needed, a corporate email is all it takes.

Start your free trial → Or decode a contract free, no account

Free for 30 days, no card needed. Your data stays isolated at the database, and you can export or delete it any time.

Watch it in action

ISVCOSELL: the three minute demo What discount should we expect? One question, every agreement

Browse the full demo library →

V ISVCOSELL

A ISVCOSELL product · © 2026

PLATFORM Benchmarking Negotiations Contract management AI workflows Document search Renewal calendar

PRODUCT Use cases Features Security Pricing Deployment options About

RESOURCES Getting started ROI calculator Agent protocol FAQ Blog Request access

More in Field Notes

1,483 vendors, one method: how the benchmark library is built

A benchmark is only as good as the deals behind it and the honesty of how it is compared. How the library is built from modelled deal cohorts, normalized, placed in the right peer cohort, and graded by confidence.

Read

300 vendors, 52 weeks, one team: the renewal calendar problem

The average enterprise runs 300+ software vendors and every one of them renews. Why notice windows are where budgets quietly die, and how a renewal desk with AI agents turns the calendar from a threat into leverage.

Read

A calmer desk, and Main Apps where the work starts

The platform now wears the desktop look: warm paper, one interactive colour, and Main Apps folded into Home so your instruments live where you start.

Read

A live analyst in your ear: inside the call copilot

The vendor call is where prepared positions meet improvisation, and the rep does this every day. The live call copilot runs a whisper rail beside the conversation: live transcript, grounded prompts, and the exact fact you need at the moment the claim is made.

Read

Adobe ETLA vs VIP: seat reclaim, right-profiling, and the walk away

An Adobe ETLA renewal is decided before you discuss price, by how many seats sit idle and how many are over-profiled. How to reclaim the waste, right-profile the rest, and build the VIP walk away Adobe respects.

Read

Agent to agent: how the Agent Negotiation Protocol works

When a buyer's AI agent negotiates with a vendor's AI agent, someone has to keep the record straight. How the open Agent Negotiation Protocol handles identity, mandate, and a ledger neither side can rewrite.

Read

Want help putting this into practice?

Contact us to discuss your project.

Get in Touch