Procurement decisions with AI assistance beyond the unit price

5 days ago
3 min read

Updated: 2 days ago

Introduction

Buying decisions in small businesses are made on the number in front of the buyer, which is almost always the unit price or the purchase price. Everything else — installation, consumables, downtime, training, disposal, the cost of the next machine being incompatible — arrives later and is attributed to something else.

This is not a knowledge problem. Owners know that the cheap machine has expensive consumables. It is a calculation problem: modelling five years of running cost across three options, with different assumptions about volume and reliability, is an hour of work that never gets done under time pressure. Getting that model built is where the assistance is worth having.

1. Procurement decisions with AI assistance should model total cost of ownership

Not the invoice.

Purchase price, installation, training, consumables, energy, maintenance, spares, expected life, downtime and residual value. Over the realistic life of the asset, at your actual volumes. This reorders options frequently.

2. Get the volume assumption right first

It drives everything.

The cheaper machine with dearer consumables wins at low volume and loses badly at high volume. Since the answer depends entirely on this number, spend the time on it rather than on the comparison.

3. Include the cost of the switch itself

Frequently decisive.

Retraining, changed procedures, new tooling, parallel running, the temporary drop in output while people learn. These are real costs of changing supplier or equipment and they are habitually excluded from the comparison that justified the change.

4. Ask what happens when it breaks

Support is part of the product.

Response time, spares availability, whether anyone local can service it, lead time on parts. A cheaper machine with a three-week parts lead time is not cheaper in a business that cannot stop for three weeks.

5. Check what you are locked into

The cost that appears later.

Proprietary consumables, service agreements, software subscriptions, incompatible fittings. Lock-in is where an initially competitive purchase becomes expensive, and it is visible before you sign if you look for it.

6. Get a written specification before you get quotes

The discipline that prevents regret.

What it must do, at what volume, to what tolerance, in what space, with what utilities. Without this you are comparing three suppliers' interpretations of what you need rather than three answers to the same question.

7. Talk to somebody who already owns one

The most reliable information available.

A user two years in will tell you about the failure mode, the consumable cost and the support reality in five minutes. This outperforms every specification sheet and costs a phone call.

8. Decide the criteria and weights before you look

Otherwise the decision follows the presentation.

Written criteria with weights, agreed in advance. Buyers routinely rationalise towards the option with the best salesperson, and a scoring sheet completed before the demonstrations is the defence against it.

9. Review the decision after twelve months

The only way the process improves.

Actual running cost against the model, and whether the assumptions held. Most organisations never do this, which is why the same optimistic assumptions reappear in every subsequent purchase.

Be careful about applying total cost of ownership to trivial purchases. The analysis has a cost, and for low-value items a simple price comparison and a reliable supplier is the correct answer.

Conclusion

Model the whole life of the purchase, because unit price is a small part of what you will pay.

Establish your real volume assumption first since it determines the answer, include the cost of switching and the temporary loss of output, ask about parts lead times and local service before you buy, look for proprietary consumables and service lock-in, write a specification before requesting quotes so you are comparing answers to the same question, speak to an existing owner, agree the scoring criteria and weights before any demonstration, and compare actual running cost against your model after a year.