How Do I Find Accounts That Are Buying Less Than They Used To?

What's actually happening

Accounts rarely quit in one move. They taper. An account that used to order a full pallet every month starts ordering three-quarters of a pallet, then stretches the interval to six weeks, then seven. No single order looks alarming. Stacked over a quarter, the account has cut its volume in half and is most of the way out the door.

There are two separate fades here, and it pays to keep them apart. Declining volume means the order size fell while the timing held: same 28-day cadence, thirty percent less on the invoice. Slowing frequency means the size held while the interval stretched: full orders, but every 42 days instead of every 28. An account ordering on time at a smaller size is a different conversation from one ordering late at full size, and merging them into a single notion of decline loses the part that tells a rep what to say.

What you are looking for is not total volume, it is the change against each account's own past. A small account holding steady is in better shape than a large account quietly sliding. The slide is the signal, and both versions of it are visible in the order records before they reach any annual comparison.

What most distributors do

Most distributors look at top-line revenue and the biggest accounts. A fading mid-tier account gets averaged into a number that still looks fine, so the decline hides inside the aggregate. Nobody is comparing each account to its own past, because doing that by hand across hundreds of accounts is not realistic.

Some teams pull a sales-history export from Epicor P21 or Eclipse and sort by total spend. That surfaces the big names but not the trend. A report of what was bought is not a read on which accounts are decelerating, so the slow faders stay buried until a year-over-year review finally catches them, long after the easy save window passed.

A related trap is comparing this month to last month. Distribution is seasonal enough that a single month-over-month drop means almost nothing, and reps learn to ignore the flag. Rolling comparisons over a longer trailing window, measured against the same account's earlier window, are the ones worth acting on.

A better approach

Compare each account against itself on two axes. First, order size: measure trailing revenue over the last ninety days against the ninety days before it, for that account alone. Second, interval: measure the average gap between recent orders against the account's longer-run baseline. Either one moving materially is worth a call, and both moving at once is worth calling today.

Rank the faders by revenue at stake so a rep starts with the declines that cost the most. A ten percent slide on a six-figure account outranks a forty percent slide on a small one. The point is to catch the taper at the second or third smaller order, not at the year-end summary when the account is already half gone.

Give the rep the numbers, not a verdict. "Revenue down 34% over the last 90 days, $18,400 against $27,900" is something a rep can open a call with. A vague label is not, and reps stop trusting labels they cannot trace back to an invoice.

How Allodial Predict addresses this

Allodial Predict learns each account's normal order size and ordering interval from your existing order history, then detects declining volume and slowing frequency as two separate signals. Accounts showing either surface on the Opportunity List with a plain reason carrying the actual figures, so a rep sees a fading account while the slide is still shallow.

An account tripping both signals appears once, not twice. The strongest reason goes on the row and the rest sit in the evidence panel underneath it, so multiple signals raise an account's place in the list without ever splitting it across two rows. No exports to sort by hand, and no waiting for a year-over-year report to expose the decline.