How to spot customers who are buying less before they leave

Why does declining spend go unnoticed?

Because nothing announces it. A lost tender or a complaint creates an event someone reacts to. A customer who orders every five weeks instead of every four creates nothing at all. The order still arrives, the invoice is still paid, and the account still looks active.

The information exists, but it sits in the ERP, which records what was ordered and invoiced, while the sales team works in the CRM, which records conversations and deals. Repeat orders rarely appear as new deals, so the biggest accounts can look quiet in the CRM while their buying changes underneath. Without a regular comparison against each customer's own history, a gradual decline only becomes visible at the annual review, if then.

Which changes in buying should you watch?

Changes in purchasing frequency

Every regular customer has a rhythm: weekly, monthly, seasonal or tied to their own production cycle. Work out each customer's usual interval between orders from their history, then flag accounts whose current gap has run well past it. A gap of one and a half times the usual interval is a reasonable starting point to tune from, rather than a rule.

Declining account spend

Compare spend over the last three or twelve months with the same period a year earlier, not with the previous quarter, so seasonal patterns do not set off false alarms. A fall matters more when it is sustained across several months than when it rests on one unusually large order last year.

Missed expected reorders

Where a customer buys the same items on a known cycle, such as consumables, call-offs against an agreed volume or a regular print run, the next order is predictable. A predictable order that has not arrived is the clearest single signal there is, and the easiest to act on, because the conversation is simple: are you still expecting to need this?

Reduced product-category purchasing

A customer who used to buy three product families and now buys two has changed something, even if total spend has held. Category changes often show up before spend does, because a customer can move one line elsewhere, change their own product mix or stop a project while keeping the rest.

Does buying less mean the customer has gone to a competitor?

No, and it is worth being careful about this, because the assumption shapes the conversation. A fall in spend can mean the customer's own sales are down, a project has finished, they have changed their product or specification, they are running down stock, prices have moved, they have consolidated purchasing at head office, or they are buying some lines elsewhere. The data tells you something has changed. Only a conversation tells you what.

That is why the right first action is almost always a call from the person who owns the account, framed around the customer's plans rather than around the lost revenue.

How should you prioritise accounts?

A long list of drifting accounts is as useless as no list. Rank them so each account manager has a handful to look at each month:

1. Value at risk. The difference between what the account usually spends and what it is spending now. A small percentage drop on a large account can matter more than a large drop on a small one.
2. Strength of the signal. A missed predictable reorder outranks a gentle fall in spend; several signals together outrank one.
3. Duration. Three months of decline is more telling than one.
4. Strategic weight. Accounts that matter beyond their spend, such as reference customers or those in a sector you are building, can be moved up by judgement.

Keep the scoring simple enough that an account manager can see why an account is on their list. A score nobody understands is a score nobody acts on.

How should follow-up be assigned?

Every flagged account needs one owner and a date, recorded where the rest of the account's history lives, in the CRM.

• The account owner makes the first contact, because they hold the relationship.
• Where there is no clear owner, the sales director assigns one, which is often the most useful outcome of the whole exercise.
• Operations or customer service are brought in where the likely cause is service, delivery or quality rather than demand.
• The outcome is recorded in a few words, such as project finished, stock run-down, price, or now ordering again, so the next review starts from what was learned.

How do ERP and CRM data support account management?

The ERP holds the facts: orders, invoices, products and dates. The CRM holds ownership, conversations and tasks. Connecting the two means the signals above can be worked out from the ERP's history and shown on the company record in HubSpot or Pipedrive, beside the person who owns the account and the history of every conversation with it.

In practice that usually means, on each company record:

• spend for the recent period against the same period last year
• the last order date against the customer's usual cycle
• the product categories bought recently against those bought before
• a flag, a task or a place on a list when an account crosses a threshold

The ERP remains the system of record; the CRM shows what the sales team needs from it. The wholesale and distribution page includes an illustration of such an account view, built from fictional data, and packaging and printing shows how the same idea applies to repeat orders run from a print MIS.

If an example dashboard is used internally to agree what the view should contain, label it as an illustration until it is built from real data. Figures that look real get quoted.

A monthly account-review checklist

Run this once a month, in a fixed slot, with each account manager and their list:

1. Refresh the data and confirm it covers orders up to the end of last month.
2. List accounts whose gap since the last order has run well past their usual cycle.
3. List accounts whose spend is down on the same period last year by more than your agreed threshold.
4. List accounts that have stopped buying a product category they bought regularly.
5. Check any predictable reorder or call-off that is due and has not arrived.
6. Rank the combined list by value at risk and strength of signal, and cut it to what each account manager can act on this month.
7. Confirm one owner and a follow-up date for each account, recorded in the CRM.
8. Review last month's flagged accounts: was the contact made, and what was the reason?
9. Record the reasons in a few words, and note any that point to service, delivery or pricing rather than demand.
10. Adjust thresholds that flagged too many accounts, or too few.

What are the limitations?

• Data quality decides the result. A customer set up twice, or branches recorded as separate customers, will look like two declining accounts rather than one stable one. Check this first.
• Seasonality and projects distort short periods. Compare against the same period last year and look at trends, not single months.
• Small or irregular buyers produce noise. A customer who orders twice a year has no meaningful cycle; leave them out of the frequency signal.
• Sales through merchants are invisible. Where your product reaches end users through distributors, your order history shows the distributor's buying, not the end customer's.
• The signal is not the reason. It tells you where to look, never what you will find.

What has Flowbird done in this area?

Flowbird connects HubSpot and Pipedrive to the systems that hold order history, and has done so with Sage 200, Sage 50, Xero, TidyERP, the Linnworks order system and the Tharstern print MIS. For one UK distribution business, Flowbird brought customer and company records from an on-premises Sage 200 into HubSpot through a secure tunnel, so the account the sales team works from is the one the ERP holds; that project is described in Connecting Sage 200 and HubSpot. Showing buying patterns on those records is the layer built on top of that foundation. Our article on why retention fails in the structure, not in the campaigns covers the account-management side.

Where to start

The CRM & ERP Opportunity Audit is a fixed-price, two-week review that tells you whether your systems could show which customers are buying less or are due to reorder, and what that would take. It checks what your CRM data can carry and maps one revenue-critical process end to end. It does not need access to your ERP and does not analyse your order history; that analysis is the work that follows, if you go ahead.

For the other side of the same data, read how manufacturers and distributors grow revenue from existing customers.