How do you spot customers who are quietly buying less?

You spot customers who are quietly buying less by comparing each customer's purchases over the last three to six months with their own pattern over the year before, not with the company average. A customer who drops consistently below their own normal, in volume, frequency or range of products, is scaling back. In your total revenue that decline is offset by growth elsewhere, so you have to look for it customer by customer.

Why you do not see it

A quiet decline is hard to spot because nothing happens. No cancellation arrives, no complaint, no lost deal in the CRM. The customer remains a customer. They just order less often, less each time or only the standard products. Sometimes they have found a second supplier, sometimes something has changed internally, sometimes there is a new buyer.

Three things make it invisible:

• Totals hide it. If your revenue grows 6 percent this year, nobody notices that twenty existing customers are together buying 15 percent less. New customers make up for it.
• The CRM looks forward. The CRM tracks deals and pipeline. A customer with no open deal does not show up as a risk, even if their orders halve.
• Monthly figures fluctuate. A customer who orders EUR 20,000 one month and EUR 8,000 the next may simply be having a quiet month. A trend only becomes visible over several months, and by then nobody is looking.

The consequence: the first time anyone notices is when the customer stops entirely or does not renew a contract. By then it is too late to ask what was going on.

What forms of decline should you look for?

A decline takes more forms than falling revenue. Look at all four:

1. Lower volume. The same products, smaller quantities.
2. Lower frequency. The same orders, but every six weeks instead of every four.
3. Narrower range. The customer used to buy five product groups, now two. Often the first sign that a competitor has taken over part of the business.
4. Less usage for the same invoice. With subscriptions and licences the invoice stays the same, but usage drops. Revenue only falls at renewal, but the signal is there months earlier.

The third pattern in particular is missed, because the customer's total revenue sometimes barely falls if the remaining product groups are the largest.

Method: each customer against themselves

1. Export invoice lines or orders for the last 24 months. Customer, date, product group, amount.
2. Calculate revenue, number of orders and number of product groups per customer per month.
3. Set a baseline for each customer. The average monthly revenue over months 4 to 15 ago. This excludes the most recent period and partly smooths out seasonal fluctuations.
4. Calculate recent purchases. The average of the last three months.
5. Flag customers whose recent purchases are 25 percent or more below the baseline. Adjust the threshold to your sector. For strongly seasonal purchasing, compare with the same months last year.
6. Do the same for frequency and product groups. Flag a customer who went from five product groups to two, even if their revenue has only fallen slightly.
7. Sort by euros. The difference between baseline and recent, times twelve. That is what is at stake per year.
8. Have the account owner review the top of the list. Is there an explanation, such as a completed project or a known reorganisation? Record it. Without an explanation: pick up the phone.

This can be done in Excel with a pivot table, and in Power BI or a similar tool with a few calculated columns. It does not require AI. It does require the discipline to repeat it every month.

Worked example

Worked example: suppose you are a wholesaler with 400 regular customers and EUR 15 million in revenue. The analysis flags 35 customers who have been at least 25 percent below their own baseline for the past three months. Their combined baseline was EUR 180,000 a month, their recent purchases EUR 120,000 a month.

• Difference: EUR 60,000 a month, or EUR 720,000 a year if it stays that way.
• If some of that slides into customers leaving altogether, it becomes more.

Not all of that revenue can be won back. But a customer who has moved part of their purchasing elsewhere can often still be persuaded if you see it within a few months. After a year, the new supplier is established.

What to do with the list

The list is a conversation agenda, not an accusation. A few rules of thumb:

• Call, do not email. The question is open: "we see you are ordering less in product group X, is that right and what has changed?" You learn more in a five-minute call than from a month of data.
• Record the reason. In the CRM, on the customer. After a few months you will know which reasons come up most: price, lead time, a competitor, an internal change.
• Distinguish between lost and paused. A customer who has finished a project may come back. A customer who has gone to a competitor has to be won back.
• Link it to your retention process. A customer on this list should get extra attention at the next renewal.

How this differs from churn and from upsell

A quiet decline is the early stage of churn. The customer is still there, but leaving. If you only measure churn, you measure too late. It is also the mirror image of upsell: a customer growing in usage is an opportunity, a customer shrinking is a risk. Same data, same method, opposite direction. See how to find missed upsell.

With subscriptions there is one more thing. A customer who uses less but pays the same is not a leak today, but will be tomorrow. And a customer who pays less while using the same is an invoicing error. You find the latter with checks for errors in recurring revenue.

Can software see it earlier?

A rule such as "25 percent below the baseline" is crude. It misses customers with an irregular pattern and flags customers who are simply having a quiet month. Statistical methods and machine learning can calculate an expectation per customer, taking season and trend into account, and flag deviations earlier. Whether that works reliably depends mainly on how much history you have. Read more in can AI predict revenue leakage.

For most companies the first step is not smarter calculation, but looking customer by customer at all. Declining purchases also feature in the list of signs your business is leaving revenue on the table, and they are one of the areas in the broader approach to finding revenue leakage. The Revenue Audit covers retention as one of the eight areas it works through, with a euro amount for every finding.

Checklist

• Can you pull each customer's monthly revenue for the past two years without manual work?
• Can you see how many product groups each customer buys, and how that is changing?
• Does someone receive a monthly list of customers dropping below their own normal?
• Is the reason for a decline recorded in the CRM?
• Do customers who are scaling back get extra attention before renewal?

Frequently asked questions

How far back should I look?

At least twelve months for a baseline, preferably 24 to recognise seasons. For customers who order only a few times a year, you need more history before a decline means anything.

What threshold is right?

It differs per company. Start with 25 percent and see how many customers it flags. If the list is too long to call, raise the threshold or look only at customers above a minimum revenue.

Is this also relevant for fixed-price subscriptions?

Yes, but then you look at usage rather than invoice amount. A customer who hardly uses their licences any more will often cancel or ask for a smaller package at the next renewal.

Who should receive this list?

The account owner for each customer, with a copy to whoever is responsible for retention. Without an owner it becomes a report nobody reads.