Which Vendors Should You Renegotiate? A Vendor Scorecard with Copilot in Excel
Contract renewals usually go one of two ways. Either the vendor sends a price increase and you sign it, or someone spends a week pulling invoices and work orders together to argue back. Most property teams have the data to negotiate well. It's just spread across the AP system, the work order system and a few inboxes.
In this guide we use Copilot in Excel to turn nine months of invoices and service calls into a vendor scorecard, then ask it the question that matters: which vendors should we renegotiate with, and what do we say? Three prompts, about five minutes, and every number checked.
Here's the whole thing, sped up:

What you'll need
A Microsoft 365 Copilot licence and a workbook saved in OneDrive or SharePoint.
Your AP invoices as an Excel table, with a column that says whether each invoice matched a PO. Ours comes from the invoice approval workflow.
Your service calls or work orders as a second table: vendor, date, emergency or planned, response time, the SLA and whether it was a repeat visit.
Step 1: Put both tables in one workbook
Our workbook, Vendor Scorecard 2026, has three tables: 162 AP invoices from January to September, 132 service calls (27 of them emergencies) and the open purchase orders. All three are proper Excel tables with clear column names. That one habit does more for Copilot's accuracy than any prompt trick.
Step 2: Give Copilot the scoring rules
Don't ask Copilot to "rate the vendors". Tell it exactly how, so the score means something you can defend in a negotiation:

Build a vendor scorecard on a new sheet called Vendor Scorecard. One row per vendor with: total spend (Subtotal) Jan to Sep 2026 from the AP Invoices table, number of invoices, invoices outside a PO (PO Match is Over PO or No PO), emergency call-outs from the Service Calls table, emergency calls that missed the SLA (Within SLA = No), and repeat visits. Add a Score column: 100 minus 5 per invoice outside a PO, minus 2 per emergency call-out, minus 8 per missed SLA, minus 4 per repeat visit. Sort by score, lowest first, and colour scores under 50 red, 50 to 80 amber, above 80 green.In under two minutes Copilot joined the two tables, built the scorecard and coloured it. A second prompt added a chart:
Make the Score column wider so the numbers show, and add a bar chart of Score by vendor next to the table, lowest score at the top.
We checked every row against our own calculation. All nine scores matched, from ClearAir at 33 to the office supply vendor at 100.
Step 3: Ask the real question
Which three vendors should we renegotiate with before renewing their contracts, and why? For each, use the AP Invoices and Service Calls data to give the evidence and three talking points for the conversation. Don't change the workbook, just answer here.Copilot came back with ClearAir Mechanical, Metro Lift and Peak Roofing, and for each one a short evidence line and three talking points:

ClearAir Mechanical, score 33: the biggest spend ($70,657.92), three invoices outside a PO, and four of six emergencies past the 4-hour SLA. Talking point: no PO, no payment for non-routine work.
Metro Lift, score 36: ten emergency call-outs, nine of them at Lakeview, and three repeat visits for the same controller fault. Talking point: tie part of the renewal price to uptime.
Peak Roofing, score 71: small spend, but both emergencies missed the SLA, one by 22 hours, and it produced a $4,800 invoice with no PO.

The part that impressed us
Peak Roofing scores 71, which is amber, not red. A simple "lowest three scores" list would still have picked it, but Copilot gave the right reason: it's not the money, it's that the one time you really needed them, a roof was leaking for 26 hours. That's the line that wins the negotiation.
It also connected the over-PO invoice to the PO itself: "the $3,980 compressor replacement exceeded its PO." That came from two different tables, and nobody told it to look.
A quick game: which vendor would you fire?
Three made-up vendors. Pick one to replace before reading our answer.
Vendor A: $90,000 spend, every invoice on a PO, two emergencies, both on time.
Vendor B: $12,000 spend, one emergency, 30 hours late, during a flood.
Vendor C: $40,000 spend, five invoices without a PO, no emergencies.
Ours: nobody, yet. A is your best vendor. B needs an SLA conversation, not a replacement. C needs a process fix: no PO, no payment. A scorecard starts the conversation. It shouldn't end the relationship on its own.
Lessons from the build
Write the scoring rules yourself. Copilot will happily invent a score. Yours will hold up in a meeting with the vendor.
Keep emergencies separate from planned work. Response time on a planned visit tells you nothing.
Ask for talking points, not just a ranking. The ranking takes five seconds to read. The talking points save an hour of preparation.
Check one number per vendor. Ours all matched, which is exactly why you check.
Ideas to take it further
Refresh it every quarter: new invoices and service calls, same prompts, new scorecard.
Add price trends: ask Copilot to compare each vendor's monthly rate this year with last year.
Bring it to the renewal: paste the talking points into the renewal email in Outlook and let Copilot draft it.
Share it with the site managers: they know which vendor actually shows up, and the scorecard gives them the numbers.
Why this matters
Vendor performance lives in three places: AP knows what you paid, the work order system knows how fast they came, and the site manager knows whether the problem came back. Bring your data from different systems into one place and put AI in front of it, and the renegotiation list writes itself, with the evidence attached.
Need help?
Smart Solutions builds vendor reporting, Power Platform workflows and Microsoft 365 Copilot solutions for Canadian property and facilities teams. If you'd rather have purchase orders, invoices and vendor performance in one product, take a look at ProcuraCloud, our procurement platform for small and mid-sized businesses. Contact us to talk about your vendor contracts.




Comments