Back to Blog

How to Score Suppliers with AI: Step by Step

Henri Jung, Co-founder at Superkind
Henri Jung

Co-founder at Superkind

A dark brass jeweler's loupe standing upright, with the orange Superkind mascot peering through its lens from the back

To score suppliers with AI, start where the data already lives: your invoices, delivery notes and complaint folder. Monday, 09:10. Thomas, head of procurement at a mid-sized industrial supplier, is in a meeting about a shipment that arrived four days late and short two pallets. He knows there is a scorecard somewhere, built in Excel two years ago, last updated in June. The numbers in that file do not match what accounting booked in DATEV, and the three recent complaints never made it into the sheet. This guide shows how to hand exactly that work to an AI employee, step by step, so your scorecard is live, honest and in the channel where your team already works. Below you will see how Superkind’s ready-made AI team for supplier scoring builds the scorecard, catches red suppliers early and runs the weekly review automatically.

Scoring suppliers has two sides: the one-off evaluation when a new supplier applies (quality manual, certificates, references) and the continuous performance score that runs every week on real orders, invoices and complaints. This guide is about the continuous score, because that is where most procurement teams lose time today and where AI makes the biggest difference.

Key takeaways

The AI reads, you decide - Superkind pulls delivery dates, defect rates and invoice variances out of DATEV, SAP Business One and SharePoint, applies your weights and builds a weighted scorecard every week. The thresholds, the award and the escalation stay with you.

Red flags before stockouts - lead-time variability and defect trends often show a problem three weeks before a line stops. Superkind catches the drop and asks for approval before pausing new orders.

Explain the rules once, they keep running - a rule like “score drop over 15 points: alert the head of procurement; awards over EUR 10,000: sign-off from management” turns a one-off scorecard into a weekly routine.

Get a ready-made AI team for supplier scoring - Superkind comes with a procurement team (Score Keeper, Red Flag Scout, Quote Normaliser) that builds the scorecard, flags risks and compares new quotes automatically. Start for free with unlimited users, or build your own.

Keep supplier data in Europe - prices, defect rates and credit notes are commercially sensitive. Superkind is hosted in Europe, GDPR compliant and never trains on your data.

Your AI team for supplier scoring

Supplier scoringProcurement
Score KeeperAI employee
Routine · every Monday at 7:00
Scorecard for week 44 ready: 48 suppliers scored on 5 criteria. 3 dropped into the red band since last week, 2 moved up.
Red (score below 60)4 suppliers
Yellow (60-79)11 suppliers
Green (80 and above)33 suppliers
Runs automatically every Monday at 7:00
Example: Score Keeper at work in Microsoft Teams

What does it mean to score suppliers with AI?

Scoring suppliers with AI means an AI reads your purchase orders, delivery notes, invoices and complaints out of DATEV, SAP Business One, Excel and SharePoint, applies your weights to each criterion (delivery, quality, price, failures, compliance), and keeps a weighted scorecard current every week, while a person approves escalations and awards. The AI handles the reading, normalising and ranking. The weights, the thresholds and the decision to pause or award stay with your team. One example is Superkind: a ready-made AI team that builds the scorecard, catches red suppliers and compares new quotes against the live score, hosted in Europe.

The difference to a generic prompt in ChatGPT: an AI employee does not stop at describing the method. It knows that Hartmann’s invoice sits in DATEV with a credit note that cuts the real price by 4 percent, that three of its deliveries last month arrived outside the Incoterms-defined window, and that its score should drop by eleven points before the next order goes out.

How it works with Superkind: your AI team for supplier scoring

Superkind is the AI employee for your team: you write to Superkind by email (Outlook or Gmail), in Microsoft Teams, in Slack or in the Superkind web app, like you would write to a colleague, and Superkind does the task in your programs. Hosted in Europe, GDPR compliant, with 1,500+ integrations into the systems where your procurement work already lives: SAP Business One, DATEV, Microsoft Excel, SharePoint, HubSpot, Salesforce, Outlook and Teams.

You do not start from scratch: Superkind comes with ready-made AI employees for procurement. They run on their own in the background, pull data out of DATEV, SAP and your inbox, and the escalation waits for you where it matters. Use them right away, adapt them, or build your own.

Score Keeper

AI employee for supplier scoring

Your scorecard sits in an Excel file nobody has time to update. Every Monday morning the Score Keeper pulls delivery dates, defect rates and invoice variances from SAP Business One, DATEV and the quality folder on SharePoint, scores each supplier against your weights (delivery, quality, price, compliance, failures) and posts the ranked table in Teams. Runs on its own, no approval needed.

Works inMicrosoft TeamsExcelSharePointDATEV+ 1,700 more
Supplier scoringProcurement
Score KeeperAI employee
Routine · every Monday at 7:00
Scorecard for week 44 ready: 48 suppliers scored on 5 criteria. 3 dropped into the red band since last week, 2 moved up.
Red (score below 60)4 suppliers
Yellow (60-79)11 suppliers
Green (80 and above)33 suppliers
Runs automatically every Monday at 7:00
Example: Score Keeper at work in Microsoft Teams

Red Flag Scout

AI employee for supplier risk

A supplier can slide from green to red over three weeks and nobody notices until a line stops. The Red Flag Scout watches the scorecard and the open purchase orders, catches a drop before it turns into a stockout, and asks you whether to pause new orders and start a replacement search. One place to approve, one audit trail.

Works inOutlookMicrosoft TeamsSharePoint+ 1,700 more
Red flag: Metzger GmbH score dropped from 78 to 54
Red Flag ScoutAI employee
Red flag: score drop over 15 points
Metzger GmbH score fell to 54. Three late deliveries in October, defect rate up to 3.1 percent, two unresolved credit notes in DATEV. 4 open purchase orders worth EUR 42,800.
On-time delivery71 percent (target 95)
Defect rate3.1 percent (target under 1)
Open POs4, worth EUR 42,800
Pause new orders and request a review meetingWaits for your OK before anything leaves.ApproveReject
Example: Red Flag Scout at work in Outlook

Quote Normaliser

AI employee for new quotes and awards

New quotes land as PDFs or email bodies and sit next to a scorecard nobody consults. The Quote Normaliser reads each quote, normalises price, Incoterms, lead time and payment terms, and places the supplier next to its live score before you compare on price alone. Awards over EUR 10,000 wait for sign-off, routine orders run on their own.

Works inOutlookExcelSharePoint+ 1,700 more
Superkind
New chat
My tasks
AI employees
Quote Normaliser
Score Keeper
Red Flag Scout
Steel brackets, request 2026-44-012

For every new quote: normalise price and Incoterms, compare against the live scorecard and recommend.

Reasoning ⌄
Quote Normaliser 11:40
New supplier quote in Outlook

Three quotes read and normalised to delivered price in EUR. Hartmann Stahl has the lowest headline price but is in the red band this week. Recommendation: award to Stein and Co, second cheapest, green band, lead time 2 days shorter.

Hartmann StahlEUR 8.40 per piece, red (54)
Stein and CoEUR 8.70 per piece, green (87)
Ferkamp AGEUR 9.10 per piece, green (82)
Award to Stein and CoOrder value EUR 12,400, over your sign-off threshold.
ApproveReject
Send a message ...
Example: Quote Normaliser at work in Superkind

Every employee is ready to use right away. Or build your own on the platform.

Safe by default

Supplier prices, defect rates and credit notes are commercially sensitive. Superkind keeps them in Europe and never uses them to train models.

From the first scorecard to your own app

The same path, told once on this topic. Each stage builds on the last.

  1. 1

    Start

    Ready in minutes. Sign up for free at app.superkind.ai, forward last month’s delivery notes and complaints to Superkind and watch it build the first scorecard for your top ten suppliers in Excel.

  2. 2

    Relief

    Tasks become routines. Every Monday morning the scorecard refreshes on its own with the latest data from DATEV and SAP Business One, and only exceptions (red suppliers, score drops) wait for your approval.

  3. 3

    Knowledge

    Learns from day one. Your Company Brain remembers your weights, your thresholds, the context behind each supplier (sole source for the brass fittings, backup for the gaskets) and your sign-off rules. When a buyer leaves, the knowledge stays.

  4. 4

    Growth

    When it gets more complex, you do not start over. Build a custom supplier-rating app with its own interface on the Superkind platform, on top of what Superkind already knows about your spend, your failures and your contracts. Built yourself or with the Superkind team, and it all belongs to you.

Send Superkind your first scorecard

Start for free in Outlook, Teams or Slack.

A vintage brass merchant's abacus with dark sliding beads and a single coral-orange bead on the middle rod

What you need

Check these points before you start. Most of them are already in place in any procurement team that books invoices and tracks orders today.

Checklist before the first scorecard

  • Access to your ERP for purchase orders and goods receipts: SAP Business One, SAP ERP, Microsoft Dynamics or an equivalent
  • Your accounting system for invoices, credit notes and payment history: DATEV, lexoffice or SAP FI
  • A central place for complaints, defect reports and quality notes, usually SharePoint, Google Drive or a QMS
  • Your supplier master data with contract numbers, Incoterms and payment terms, in the ERP or in Excel
  • An agreed set of criteria and weights: typically delivery, quality, price, failures, compliance; weights sum to 100
  • Agreed thresholds for red, yellow and green (for example red under 60, yellow 60 to 79, green 80 and above)
  • An agreed approver per threshold (score drop over 15 points, awards over EUR 10,000, pausing a sole source) and the channel they work in
  • An admin who can connect SAP Business One, DATEV, SharePoint, Outlook and Teams to Superkind and confirm the permissions
  • A free Superkind account at app.superkind.ai
  • Thanks to the 1,500+ integrations in Superkind, these systems can be connected without new tools or custom code

Step by step with Superkind

Six steps from the first forwarded delivery note to a weekly scorecard your team acts on. Each step is something you can do today.

Step 1: Collect the data

Sign up at app.superkind.ai and connect the systems your procurement data flows through: your ERP (SAP Business One or similar) for purchase orders and goods receipts, DATEV for invoices and credit notes, SharePoint or Google Drive for the quality folder with complaints and defect notes, HubSpot or Salesforce if a supplier sits in CRM, and Outlook, Teams or Slack where you write to Superkind. An admin confirms which areas Superkind may read. No API keys to copy, no webhooks.

  • Check - Superkind only sees the folders, modules and mailboxes you allowed, and the audit log records every file it opened.
  • Tip - start with one supplier category (for example mechanical parts or packaging) and the last three months of data.
TeamsThomas to Superkind

Pull the last three months of goods receipts for the mechanical parts suppliers from SAP, their invoices from DATEV and the complaints folder on SharePoint.

Step 2: Agree the criteria

Decide which five criteria matter for this category. For most procurement teams the shortlist is on-time delivery, quality or defect rate, price (including hidden costs like freight and tooling), failures or complaints, and compliance (ISO 9001, environmental, DPA where personal data is processed). Add a sixth if it is already in your buying rules, for example payment terms or sustainability. Fewer is better, four to six is enough. ISO 9001 section 8.4 asks you to evaluate suppliers, but does not prescribe the criteria or weights.

  • On-time delivery (OTIF) - compare the goods receipt date in SAP against the confirmed delivery date, counting a quantity shortfall as a miss.
  • Quality or defect rate - defect-free pieces as a percentage of all received, from the complaint and inspection notes.
  • Price - landed price per unit versus the agreed catalog price, including freight, tooling and credit notes.
  • Failures or complaints - number of complaints in the quality folder in the period, weighted by severity.
  • Compliance - up-to-date certificates (ISO 9001, IATF 16949 for automotive) and the DPA where supplier staff access your systems.

Step 3: Set the weights and the thresholds

Weighting is where the criteria become your scorecard. Agree with the team and write it down. There is no normative weighting: ISO 9001 asks for a method, not for numbers. The weighting below is a workable starting point for a mid-sized industrial supplier; automotive teams usually weight quality higher (quality 30 to 40, delivery 25 to 30), trade teams weight price and delivery higher.

TeamsThomas to Superkind

Weights for mechanical parts: delivery 30, quality 25, price 20, failures 15, compliance 10. Red under 60, yellow 60 to 79, green 80 and above.

Superkind confirms with “Noted.” From now on every scorecard in this category uses these weights, and any change goes back into the Company Brain for the next period.

Step 4: Build the scoring model

Superkind now applies the weights to the data. For each supplier, each criterion gets a score from 0 to 100 (on-time delivery of 94 percent against a target of 95 is 94, defect rate of 0.6 percent against a target of under 1 percent is 100), the weighted sum is the supplier score. The scorecard lands in Excel on SharePoint or as a Teams message, grouped red, yellow, green, with the drivers next to each supplier.

SlackThomas to Superkind

@Superkind build the scorecard for all mechanical parts suppliers for October with the weights above and post it in the procurement channel.

Answer: “Scorecard built for 48 suppliers. 4 red, 11 yellow, 33 green. Top driver of red: Metzger GmbH, three late deliveries and two unresolved credit notes. File on SharePoint, summary in the channel.”

Step 5: Set up the weekly routine with approval

Now the scorecard stops being a one-off. Tell Superkind once what should happen every Monday, where exceptions go and who signs off. Routine steps run on their own, exceptions wait for a person. One rule a team lead can read on a slide.

TeamsThomas to Superkind

Every Monday at 7:00: refresh the scorecard, post it in the procurement channel, alert me when a supplier drops into red or loses more than 15 points. Pause orders on a sole source only after my OK.

Answer: “Noted. The scorecard refreshes every Monday at 7:00. Red suppliers and score drops over 15 points go to you first; sole-source pauses wait for your OK before anything leaves the team.”

Step 6: Report in Teams or Outlook and escalate the reds

The weekly scorecard goes into the procurement channel every Monday. For a red supplier, Superkind drafts the review request with the three drivers (OTIF 71 percent, defect rate 3.1 percent, two unresolved credit notes in DATEV) and a short recommendation (pause new orders, start a replacement search for the two most critical parts). You approve, Superkind sends the message to the account owner in your name through Outlook or Teams.

OutlookNadine to Superkind

Who dropped into the red this week, and are any sole source?

Answer: “Four red this week. Metzger GmbH is sole source for the brass fittings. Two open purchase orders worth EUR 42,800. Draft to pause orders and request a review meeting is in your Teams inbox, waiting for approval.” If a replacement search is on, the compare-quotes guide covers how Superkind normalises new quotes against the live score.

Set it up as a routine

Once two or three weekly scorecards have gone through cleanly, extend the routine to the full procurement workflow. Superkind can push the scorecard into Teams, send a short management report by email, and feed the live scores into every new quote comparison.

  • Weekly - fresh scorecard every Monday at 7:00 in the procurement channel, red suppliers and score drops over 15 points flagged.
  • Monthly - a one-page report to the managing director with the trend per category, open escalations and the next three replacement searches.
  • Per quote - when a new quote arrives, Superkind reads the price, Incoterms and lead time and places the supplier next to its live score before you compare on price alone. The full flow is in the compare-quotes guide.
  • Per contract - at renewal, Superkind drafts the review and the clauses that need updating. The AI for supplier contracts guide covers the sign-off.

When the process grows, for example several sites, a shared supplier panel or a category-specific scoring model (automotive weighted on IATF 16949, trade weighted on price), you do not start over. On the Superkind platform you build a custom supplier-rating app with its own interface, on top of everything Superkind already knows. Built by your team or together with the Superkind team, and it all belongs to you. The 1,500+ integrations keep the app connected to SAP Business One, DATEV and SharePoint without new middleware.

Data protection and EU hosting

Supplier scoring touches commercially sensitive data: prices, failure rates, credit notes and sometimes personal data of named supplier contacts. These numbers belong in Europe, and increasingly a supplier audit asks you where your AI runs. Superkind is built for that.

What this means for a scorecard: the invoices Superkind reads in DATEV, the delivery notes in SAP Business One and the complaints on SharePoint never leave the EU and are not used to train a model. For many customers in automotive and public-sector supply chains, this is already a tender criterion. Superkind uses AI models from OpenAI, Anthropic, Mistral and Google, processed in Europe. For deeper procurement workflows, see the AI for procurement guide.

Other ways to score suppliers

Superkind is not the only way. Here is an honest comparison of the common options.

WayEffortError riskBest for
SuperkindLow: connect systems, write weights and rules in plain languageLow: live data from DATEV and SAP, approval where you want it, EU hostingProcurement teams scoring 20 to 500 suppliers across SAP, DATEV and SharePoint, with 1,500+ integrations
Excel scorecard, built by handVery high: copy data from SAP and DATEV every week, hunt for complaints by handMedium: data drift, outdated weights, one maintainerSmall teams with fewer than ten suppliers and little change month to month
ChatGPT aloneMedium: paste tables, copy answers backMedium to high: no link to your ERP or DATEV, no audit trail, data leaves the EU on free plansA one-off exercise when nothing is at stake
Dedicated SRM software (Jaggaer, Ivalua, Tacto)Medium to high: new tool, data model, onboardingLow in the tool, higher at the integration pointsLarge procurement organisations with a dedicated SRM project and several buyers per category
Manual, no scorecardLow in the short run, high in the long runHigh: red suppliers found only after a line stopsCompanies with one or two suppliers and no formal procurement process

The gap with Excel is maintenance. The gap with generic AI tools is traceability. The AI for procurement guide covers the fit for the full purchasing workflow, and the compare-quotes guide explains how new quotes feed back into the live scorecard.

Common mistakes

These five mistakes cost the most time when teams start to score suppliers with AI.

  • Scoring without complaints and credit notes - a scorecard built only on delivery dates misses half the story. Connect the quality folder on SharePoint and credit notes in DATEV before the first run.
  • Too many criteria - a scorecard with twelve criteria nobody weights honestly collapses into noise. Stay with four to six, tune the weights after a quarter.
  • No threshold for the red band - a score of 54 means nothing on its own. Agree the thresholds (for example red under 60) before the first scorecard goes out, so a red is a trigger, not a conversation.
  • Letting the AI pause a sole source - sole sources sit outside the normal rule because a pause costs you more than a bad delivery. Always route sole-source pauses to a person for approval.
  • No audit trail of the weights - the EU AI Act and ISO 9001 both expect you to show how a supplier was evaluated. Keep the weights, the thresholds and every change in the Superkind log.

6 tips for scoring suppliers with AI

1. Start with your top ten suppliers

Pick the ten suppliers that account for most of your spend or your risk. A good scorecard on ten suppliers beats a weak one on all of them, and the first ten pay for the setup.

2. Write the weights as a one-liner

“Mechanical parts: delivery 30, quality 25, price 20, failures 15, compliance 10. Automotive: shift ten points from price to quality.” Short, concrete sentences keep the weights honest and make a review easy.

3. Track lead-time variability, not just the average

A supplier with a steady 12-day lead time is often better than one with an average of 8 but a range of 3 to 16. Ask Superkind to score the standard deviation of the delivery window, not only the mean.

4. Put approvals where the approver works

If your managing director lives in Outlook, red-flag approvals go there. If procurement works in Teams, approve in Teams. Superkind uses the same rules and approvals in every channel across the 1,500+ integrations.

5. Make data protection a scoring criterion for suppliers that touch personal data

Suppliers who process personal data (logistics, maintenance, IT contractors) need a DPA and an answer on EU hosting. Add compliance as a scored criterion, not a yes/no footnote. For deeper DSGVO work, see the AI for supplier contracts guide.

6. Review the weights every quarter

Weights that fit your old supply plan rarely fit the next one. A short review every quarter, with the trend data in the Superkind scorecard, keeps the system honest and the team on the same page.

Conclusion

Scoring suppliers with AI does not mean software picks the next supplier. It means the reading, normalising, scoring and reporting is done for you every week, your weights and rules apply every time, and, where it matters, a person approves before a sole source is paused or an award goes out. The fastest way to see whether it fits: forward last month’s delivery notes and complaints for your top ten suppliers to Superkind with “Build the scorecard” and compare the first run with the Excel file you keep today. If you want to go wider first, read our guide to AI for procurement, our step-by-step on comparing quotes with AI and our AI for supplier contracts piece.

Frequently Asked Questions

A supplier scorecard is a weighted scoring table that rates each supplier against the criteria that matter to you, typically on-time delivery, quality or defect rate, price variance, failures and compliance. Each criterion gets a weight (for example delivery 30, quality 25, price 20, failures 15, compliance 10), each supplier gets a score from 0 to 100 per criterion, and the weighted total is the supplier score. ISO 9001 section 8.4 requires you to evaluate and re-evaluate suppliers, but not with a specific method. Weightings are a judgement call that reflects your category and risk.

Yes, for the data part. An AI employee like Superkind reads invoices in DATEV, delivery notes and purchase orders in SAP Business One or Excel, defect reports on SharePoint and quotes from your inbox, builds the weighted scorecard every week and flags suppliers that drop. The weights, the thresholds for red and yellow, and the decision to pause an order stay with you. The AI does the reading, the normalising and the scoring, not the award.

Five criteria cover most procurement categories: on-time delivery (OTIF), quality or defect rate, price (including hidden costs like freight and tooling), failures or complaints, and compliance (ISO 9001, environmental, DPA for data processors). Automotive leans harder on quality (IATF 16949), trade leans harder on price and delivery. Start with 4 to 6 criteria and tune the weights after a quarter of real data. Lead-time variability often matters more than the average.

Not on its own. Superkind drafts the review request and the pause-order note and asks for approval (Approve or Reject) before anything leaves the team. You decide the rule, for example a score drop of more than 15 points goes to the head of procurement, awards over EUR 10,000 go to the managing director. Only after approval does Superkind send the message in your name through Outlook or Teams.

Superkind works in the systems your procurement already runs on, with 1,500+ integrations: SAP Business One for purchase orders and goods receipts, DATEV for invoices and credit notes, Microsoft Excel or SharePoint for the quality folder, HubSpot or Salesforce if the supplier is in CRM, and Outlook, Teams or Slack for the message you write to Superkind. It moves the data between them so nobody retypes a delivery note into a spreadsheet by hand.

With Superkind every supplier number, defect rate and invoice is hosted and processed in Europe, GDPR compliant and aligned with the EU AI Act (Regulation (EU) 2024/1689), and the data is never used to train models. A data processing agreement is available through the Trust Center. Supplier prices and credit notes are commercially sensitive, so where the data is processed matters as much as the scorecard itself.

You can start for free: Superkind gives a one-off of up to EUR 150 in credits, no credit card, unlimited users. The Business plan starts at EUR 50 per month for the whole company, again with unlimited users rather than a price per seat. Enterprise is a custom price. As of October 2026. A full weekly scorecard typically costs a few hundred credits; EUR 100 covers up to 1,000 tasks or 5,000 questions.

Related Articles

Henri Jung, Co-founder at Superkind
Henri Jung

Co-founder of Superkind, the AI employee teams write to in Outlook, Teams and Slack. Henri helps companies, from small teams to the Mittelstand, hand their routine work to AI - without new tools, consultants or an in-house AI team.

Send Superkind your first scorecard.

Start for free in Outlook, Teams or Slack. Or let us show you in 30 minutes how it runs in your systems.