Every procurement manager I talk to has the same question before they start: "What's this going to cost?" They've been burned by inflated quotes, vendors who only sell enterprise packages, and consultants who propose six-month transformation programmes. So here are the actual numbers — split by company size, broken down by cost type, with real starter projects and what they run per month.
First: Are You Actually Ready?
Before budget, check readiness. Five yes/no questions. Be honest — the score tells you where to start, not whether to start.
What It Actually Costs: Monthly Tool Budget
The tool costs are the easy part. The numbers below are what a typical procurement team pays after the initial implementation — just the ongoing subscriptions.
| Company size | Team (procurement) | Monthly AI tools | Typical setup |
|---|---|---|---|
| Small | 2–5 people | €75–125/mo | Copilot 2× + Claude Pro 1× |
| Mid-size | 5–15 people | €125–225/mo | Copilot 4–6× + Claude Pro 2× + API |
| Large | 15+ people | €300–600/mo | Copilot 10+× + enterprise API + custom tools |
Microsoft Copilot runs €21/user/month (M365 E3 add-on). Claude Pro is €18–25/user/month depending on plan. Both are off-the-shelf. No IT project. No procurement process for the software itself. Just a credit card and a browser.
Year-One Budget: 10-Person Procurement Team
Year one is more expensive because you're paying for implementation on top of tools. Here's the honest breakdown for a 10-person team starting from zero:
| Item | Detail | Annual cost |
|---|---|---|
| Microsoft Copilot | 3 users × €21 × 12 months | €756 |
| Claude Pro | 1 user × €25 × 12 months | €300 |
| AI Implementation Sprint | 2-week focused engagement | €3,500 |
| Follow-up Sprint | After 3 months, second iteration | €1,500 |
| AI Policy | Customised for your company | €500 |
| Total year one | €6,556 | |
The Cost of Waiting
Three reasons why postponing is expensive, even if it feels like the safe option:
1. Your competitors are already doing it. The gap between AI-enabled procurement teams and everyone else is widening. In 12 months it will be wider. Teams that started in 2024–2025 have already built muscle memory, internal tools, and tested processes. You're not pausing — you're falling behind.
2. Your team is already burning time on things AI handles in seconds. Every week without contract analysis automation, quotation comparison tools, or supplier communication templates is a week of avoidable manual work. That time doesn't come back. And it's being spent by people who could be doing higher-value work instead.
3. AI literacy takes time to build — and the clock is ticking. The biggest barrier isn't technology. It's people learning how to use it effectively. That process takes 3–6 months of regular use before it becomes natural. Starting later just means you're 3–6 months further from that point. There's no shortcut.
Who Should Lead It: Choosing Your AI Ambassador
Every team that successfully implements AI has one person who makes it work. Not the most technical person. Not the most senior person. The person who is curious, trusted by colleagues, and not afraid of looking like a beginner for a few weeks.
Three traits that matter more than technical background:
- Process knowledge. They need to understand what the team actually does day-to-day well enough to spot where AI fits. A 5-year procurement specialist beats a 1-year IT generalist here every time.
- Peer trust. When they say "try this", people try it. Seniority helps, but likeability and track record matter more for change adoption.
- Tolerance for imperfection. AI tools produce imperfect outputs, especially at the start. The ambassador needs to be comfortable iterating and explaining that "good enough and fast" beats "perfect and slow" for many procurement tasks.
The Three Starter Projects
These are low-risk, high-visibility projects you can get live in 2–4 weeks with no custom software development. All three have been built and run in real procurement teams.
Contract Assistant — €25–50/month
An AI that reads incoming contracts and flags risk clauses, missing terms, and unusual payment conditions. You upload the PDF, it returns a structured summary with a risk rating. Runs on Claude API or GPT-4. The tool itself takes a few hours to configure. The prompt engineering takes a day. At €25–50/month in API costs, it handles hundreds of contracts. ROI is immediate for any team that reviews more than 5 contracts per month.
Internal Knowledge Bot — a few euros/month
Upload your supplier onboarding documents, procurement policy, approved supplier list, and standard terms. The bot answers questions from your team in natural language: "What's our standard payment term for logistics suppliers?" "Who's our approved packaging supplier in Germany?" Ends the "let me check and get back to you" loop for common internal queries. API costs are minimal — typically under €10/month for a team of 10 using it daily.
Kaizen Portal — €4.70–14/user/month
A simple improvement idea submission tool where team members submit process inefficiencies. AI categorises the ideas, estimates potential impact, and routes them to the right owner. Built on Microsoft Power Apps + AI Builder. The per-user cost is the Power Apps premium licence. Teams that deploy this typically see a 3× increase in submitted improvement ideas because the barrier to submission drops from "fill in a three-page form and send it to the process manager" to "type one sentence."
5 Myths That Keep Teams Stuck
Copilot is one tool. Claude, ChatGPT, Mistral, Gemini, and half a dozen others are available today. Many are better suited for specific procurement tasks than Copilot — contract analysis, structured data extraction, multi-step reasoning. Having Copilot on your M365 licence is a starting point, not a ceiling. Use whatever tool fits the job.
This depends entirely on which tool you use and how. ChatGPT Free may use your inputs for training by default (check settings). Microsoft Copilot for M365 runs inside your Microsoft 365 tenant with enterprise data protection. Claude's API and Pro plans do not train on your data. The risk is real but manageable — it requires a one-page AI usage policy, not a multi-month security review. Most teams can address this in a day.
Not in the next five years, and probably not the way people imagine. What AI replaces is the low-value work: formatting reports, copying data between systems, chasing approvals, writing first drafts of routine supplier emails. What it doesn't replace is judgement — supplier relationship management, negotiation strategy, risk assessment on complex situations. The procurement people who get displaced first are the ones who refuse to use AI, not the ones who use it well.
Modern AI models work with messy inputs. PDFs, scanned documents, unstructured email threads, inconsistently formatted spreadsheets — current models handle all of these. You don't need a data warehouse or a clean ERP before you start. You need a specific use case and the right prompt. The "we need clean data first" position is usually a delay tactic, not a genuine blocker. Start with what you have.
The smallest teams I've seen get value from AI are 2–3 person procurement departments at companies with under 100 employees. The tools are SaaS. You don't need IT to provision them. You don't need infrastructure. You need a browser and 30 minutes to learn the basics. If anything, smaller teams move faster because there's less coordination overhead and fewer approval layers to navigate.
4 Mistakes That Kill AI Pilots
Picking the wrong first project
The first project needs to be visible, fast, and low-stakes. If the first project is six months long, has dependencies on three departments, and requires budget approval from the CFO — it will not survive. Pick something one person can build in a week. Get a win. Then expand.
Waiting for permission from IT or legal
Involving IT and legal is necessary — eventually. But waiting six months for a security review before anyone touches a ChatGPT account guarantees the pilot dies. Start with publicly available tools, don't put confidential data in, use free tiers to prove the concept, and involve IT once you have something worth reviewing. Show, then formalise.
Treating AI as a one-time implementation
AI tools improve as you learn to use them. The prompt you write in week one will be significantly worse than the one you write in month three. Teams that treat implementation as "done" after the initial setup stop improving early. Build a habit of reviewing what's working, what's not, and what to try next. A monthly 30-minute retrospective is enough.
Not measuring anything
If you can't show what changed, you can't justify continued investment, get budget for phase two, or convince sceptical colleagues. Track two numbers from day one: time spent on the target process before AI, and time spent after. Even rough estimates work. A Google Sheet with weekly entries is enough. You don't need a full ROI framework — you need a number you can point to in the conversation with your manager.
What External Help Costs
If you're working with a consultant or implementation partner, here's the rate card to benchmark against. These are Sourcing Sense rates — use them to sanity-check any other quote you receive.
| Engagement type | What you get | Price |
|---|---|---|
| Hourly advisory | Specific question, decision support, troubleshooting | €150/h |
| AI Readiness Audit | Process mapping, readiness scoring, written recommendations, 2-hour workshop | €500 |
| The Sprint | 2-week focused engagement: tool selection, first pilot built and live, team training, documentation | €2,500 |
| Team Training | Half-day to full-day workshop, practical exercises, prompt library, custom to your team's context | €1,500 |
| Custom AI project | Full design and implementation of a specific procurement AI tool | €1,500–7,500 |
For context: a full-day management consultant at a Big Four firm in EMEA costs €2,500–5,000. The Sprint at €2,500 includes two weeks of work, a live deliverable, and a trained team. The comparison holds up.
What to Do Next Monday
Enough numbers. Here's the concrete sequence to get moving without a six-month planning cycle:
How to Work Together
Three options depending on where you are and how much support you need:
The Audit
Half-day process mapping, readiness scoring, written action plan, 2-hour working session. Useful if you're not sure where to start or need to build an internal business case.
The Sprint
Two weeks. First pilot live, team trained, documentation done. You end with something working, not just a plan. Best for teams that want results in the first month.
Team Training
Half-day or full-day workshop. Practical exercises, prompt library, real procurement use cases. Custom to your industry and processes. Delivered on-site or remote.
"The companies that are winning with AI right now are not the ones that spent the most on it. They're the ones that started fastest, stayed practical, and kept iterating. The budget is not the constraint — the decision to start is."