Weighted Scoring for Vendor Selection: Rank Suppliers Beyond Price
"We picked the cheapest" is not a vendor-selection method — it's a gamble that delivery time, scope gaps and payment terms won't cost more than the saving. Weighted scoring turns those trade-offs into a single, defensible number per vendor. Here's how procurement teams in construction, manufacturing and trade set one up.
Why price alone picks the wrong vendor
Headline price hides four recurring traps:
- Delivery: a 7-month lead time on a 4-month project turns the "cheapest" panel into liquidated damages.
- Scope gaps: a vendor who didn't quote 10% of the items isn't 10% cheaper — you'll buy those items anyway, at spot prices.
- Payment terms: 50% advance to an unfamiliar supplier is financing risk the price doesn't show.
- Quality/brand: unbranded switchgear against a tier-1 OEM is not a like-for-like price comparison.
Weighted scoring doesn't ignore price — it typically keeps price as the largest single factor — but it forces the other factors to be counted instead of argued about in the award meeting.
Before Step 1 — what are you actually buying for?
The same equipment, bought by two different people, needs two different sets of criteria. The distinction is not the item. It is whose problem it becomes after it is installed, and getting it wrong is how a technically correct scoring matrix produces the wrong award.
Buying for your own use
You are going to live with it. Total cost of ownership decides the question: purchase price, installation, energy, consumables, spares, maintenance, downtime and eventual disposal. A cheaper machine that costs more to run loses over the asset's life, and the asset's life is your problem. Weight reliability, the service network, parts availability, warranty and running cost — and accept a higher purchase price to get them.
Buying as a contractor
Your horizon ends at handover. You are not optimising lifetime cost — you are buying compliance with a specification someone else wrote, delivered by a date you are liable for. That changes what the criteria have to capture:
- Specification compliance. Will the client's engineer approve the submittal? An "equivalent" that gets rejected costs far more than the price difference that made it attractive.
- Programme fit. Delivery measured against your installation window. Late is not an inconvenience, it is liquidated damages.
- Warranty against the defects liability period. A twelve-month warranty under a twenty-four-month DLP means you personally carry the second year.
- Documentation. Test and compliance certificates, O&M manuals, as-builts. A supplier who cannot produce them holds up your handover, and your retention with it.
- Payment terms against your certification cycle. You are financing the gap between paying the supplier and being certified for the work. A 50% advance on a package you will be certified for in ninety days is your working capital, not a commercial detail.
Running cost after handover belongs to the client — unless you have taken an O&M or performance obligation, in which case you are back in the first case and should weight it that way.
| What you are buying for | Weighs heavier | Start from |
|---|---|---|
| An asset you will operate | Brand, certification, service network, running cost | Quality-critical profile |
| A contract you must fulfil | Delivery, scope completeness, payment terms, approvable specification | Urgent-project profile |
A weighted score ranks the purchase. It does not rank the decade that follows it. NazmIQ does not compute total cost of ownership and no weighting of these five criteria captures it — energy, spares and downtime are not in the quotations being compared.
So when you are buying to own, do that arithmetic separately and let it override the ranking where it disagrees. Folding a guess at lifetime cost into the "brand" score would give the matrix a confidence it has not earned, which is the one thing a scoring model must never do.
Step 1 — Which criteria should you score vendors on?
Four to six. More than that feels rigorous but dilutes the signal and invites double-counting. A battle-tested set:
| Criterion | What it captures |
|---|---|
| Price competitiveness | Converted, tax-consistent grand total |
| Delivery period | Lead time vs project need |
| Payment terms | Advance %, credit, milestones |
| Scope completeness | Share of the BOQ/BOM actually quoted |
| Brand / certification | OEM tier, type-test certificates, references |
Step 2 — Set the weights before you see the scores
Weights encode your priorities for this purchase. Decide them before scoring — setting weights after seeing the numbers is how a preferred vendor gets reverse-engineered into first place. Three ready-to-use profiles:
| Criterion | Balanced (default) | Urgent project | Quality-critical |
|---|---|---|---|
| Price | 60% | 45% | 45% |
| Delivery | 15% | 30% | 15% |
| Payment terms | 10% | 10% | 10% |
| Scope completeness | 10% | 10% | 15% |
| Brand / certification | 5% | 5% | 15% |
Step 3 — Score each vendor 0–100
- Price (deterministic): lowest comparable grand total scores 100; others score lowest ÷ vendor total × 100. A vendor 25% above the cheapest scores 80.
- Delivery: anchor to the project: meets the required date = 90–100; each month late drops the score materially.
- Payment: lower advance and longer credit score higher; 50% advance to a new supplier should not score like 10% against delivery milestones.
- Scope: percentage of line items quoted is a good objective base, adjusted down for critical missing items.
- Brand: tier-1 certified OEM 90–100; established manufacturer 75–90; unbranded/unverified below that.
Step 4 — Compute, rank, and read the result
Weighted total = Σ (score × weight). A worked example with four bidders:
| Criterion (weight) | Vendor A | Vendor B | Vendor C | Vendor D |
|---|---|---|---|---|
| Price (60%) | 70.3 | 87.9 | 100.0 | 92.6 |
| Delivery (15%) | 65.0 | 80.0 | 60.0 | 75.0 |
| Payment (10%) | 80.0 | 70.0 | 75.0 | 65.0 |
| Scope (10%) | 97.0 | 100.0 | 90.0 | 95.0 |
| Brand (5%) | 95.0 | 92.0 | 88.0 | 90.0 |
| Weighted total | 74.4 | 86.3 | 89.9 | 87.3 |
| Rank | 4th | 3rd | 1st | 2nd |
Note what happened: Vendor C wins on price strength despite the slowest-but-one delivery — and the sheet shows exactly how close Vendor D runs (2.6 points). If the project were urgent, switching to the urgent weight profile might flip the ranking — transparently, with the same scores. You can explore this exact scenario in our live sample report.
Step 5 — Test whether the ranking survives the weighting
A weighted total is only a finding about the bids if it holds when the weighting moves. Take each criterion in turn, shift its weight ten points up and ten points down, and recompute. If the same vendor wins every time, the result is robust and you can say so in one line. If a single ten-point move changes the winner, the ranking is resting on your weighting rather than on the bids — and the honest response is to report that rather than publish a confident order.
Look at the table above: Vendor C leads Vendor D by 2.6 points. That margin sits well inside what one weight move can cover, so it is exactly the case the test exists for. A margin of twenty points is not.
"The winner is unchanged under any single ten-point weight move, with a margin of 8.4 points over second place." One sentence, and the conversation stops being about whether you picked the weights to suit the answer.
The other half of that question is when you set them. Step 2 says fix the weights before you see the scores; what an auditor actually asks later is whether they were adjusted once the bids were visible. Record the date you set them, and the question answers itself.
Step 6 — Write the recommendation, not just the number
Close with two sentences: who wins and why ("lowest landed price, full scope, 88/100 weighted"), plus the caveat to clear before award (an expiring offer, an unquoted item, currency exposure). That paragraph is what management signs — and what protects you when the decision is questioned a year later.
How to automate vendor scoring
NazmIQ (nazmiq.com) computes all of this from your uploaded quotes: it extracts the line items, normalises currency and tax, applies your saved weights (editable in settings), and produces the ranked scoring table, recommendation and a two-sheet Excel — deterministically, in Python, not by asking an AI to guess the math.
It also runs Step 5 for you on every comparison. Each weight is moved ten points either way and the ranking recomputed; the report states whether the winner survives and by what margin, and names the criterion if it does not. The weighting is stamped with the date it was last changed and checked against the quotations' own issue dates, so a report can say the weights were fixed before any of these quotes existed — or that they were changed after the quotes arrived. And every priced cell records the supplier line it was copied from, including any unit conversion applied to it.
See a scored comparison, already built
The matrix from Step 4, the sensitivity check from Step 5 and the dated weighting — on a real comparison rather than a worked example. No account needed.
See a sample reportFrequently asked questions
What is weighted scoring in procurement?
A ranking method: score each supplier 0–100 per criterion, multiply by the criterion's weight, sum. Highest total wins, and the matrix documents why.
What weights should I use?
Start from price 65 / delivery 15 / payment 10 / scope 10 and adjust to the purchase — more weight on delivery when urgent. If the bids include brands you have not already approved, add a few points for quality and take them from price.
How do I score price objectively?
Lowest comparable total scores 100; each other vendor scores lowest ÷ vendor total × 100. Deterministic and auditable.
Does the lowest bidder always win?
No — a slightly costlier vendor with faster delivery and full scope routinely outranks the cheapest bid. Surfacing that is the method's whole purpose.
How do I know the ranking reflects the bids and not my weighting?
Test it. Move each criterion's weight ten points up and ten down and recompute the totals. If the winner is the same every time, the ranking is robust and you can state that alongside the result. If one ten-point move changes the winner, say so — a ranking that flips on a small weighting change is a judgement about your priorities, not a finding about the suppliers.
How do I show the weights were not chosen to favour a bidder?
Fix the weights before the scores exist and record the date. The question that gets asked later is not whether the weights were reasonable but whether they moved once the bids were visible. A dated weighting, compared against the dates printed on the quotations themselves, settles it: if the weights were set before the earliest quote was even issued, they cannot have been tuned to its contents.
Should I use the same criteria when buying for my own use as when buying for a client?
No. Buying an asset you will operate, total cost of ownership decides it — energy, spares, maintenance and downtime over the asset's life, so reliability and the service network outweigh a few points of purchase price. Buying as a contractor, your horizon ends at handover: you are buying compliance with a specification someone else wrote, by a date you are liable for. Approvable specification, delivery against programme, warranty covering the defects liability period, and payment terms that survive your certification cycle matter more than running cost, which becomes the client's problem — unless you have taken an O&M obligation, which puts you back in the first case.
How far back should a comparison be traceable?
Far enough that any figure in the summary can be followed to the supplier's own line — the description as they wrote it, the quantity, the rate before any currency or unit conversion, and the conversion applied. Unit conversion is the step worth recording most: it is the one place a supplier's own number is altered before it reaches your sheet.