Expired SNAP Tests in July, Backordered Preventive in May: The Two Ways a Practice Pays for Guessing the March Order

You tried forecasting once. It was a spreadsheet, and it was wrong.

Most owners who have been in practice more than a decade have already done this. Somebody exported three years of sales out of Cornerstone or AVImark, dropped it into Excel, drew a trend line through the flea and tick numbers, and produced a March order that was confidently wrong in both directions — too much of the six-month heartworm preventive, not enough of the chewable your top three doctors actually recommend. The spreadsheet went in a drawer. The order went back to being what it always was: what we bought last spring, plus a bit, minus whatever the rep talked us out of.

That failure was not laziness and it was not bad math. It was that a general practice has exactly the kind of sales history classical forecasting handles worst — a few years deep, wildly seasonal, full of item codes that changed when you switched distributors, with a gap in 2020 that means nothing and a spike in one week of April that means the local news ran a tick story.

That is the part that changed.

The two ways a preventive order goes wrong, and what each one costs

Over-ordering shows up in July. Short-dated inventory is the veterinary version of spoilage. Vaccines carry real dating. In-clinic test kits carry dating and want refrigeration. Preventive boxes sit on the shelf and quietly cross their date while your practice manager avoids looking at the shelf, and the write-off gets buried in cost of goods where nobody has to say it out loud.

The subtler version of over-ordering is a manufacturer rebate structure you committed to and then missed by a hair, or hit on product you did not need. Every practice has done a fourth-quarter purchase to reach a tier. Some of those were smart. Some of them were a loan to your own shelf.

Under-ordering shows up in May, and it costs more. A client comes in for the annual, the doctor recommends twelve doses, and the front desk says we are out, we can order it, it will be here Thursday. That client goes home and buys it online that evening. And here is the part that makes this the expensive error: they do not come back to your shelf next year. One stockout does not cost you one sale. It moves a household's entire preventive spend off your shelf, permanently, and it takes the compliance conversation with it.

Ask your practice manager which of those two errors your clinic makes. Most will say both, in the same season, on different products.

What is different about 2026 forecasting

Demand forecasting is now one of the highest-adoption uses of AI in industry — roughly 48 percent in manufacturing — and pricing optimisation sits around 72 percent in retail and e-commerce. Those two numbers are not there because somebody discovered forecasting. They are there because the current generation of models finally deals with the kind of history real businesses have: sparse, messy, seasonal, with holes in it and categories that were renamed halfway through.

For a veterinary practice, demand forecasting means predicting how many doses, test kits and surgery slots you will actually sell in a given week, using your own dispensing history, your own reminder list and your own county's season — instead of reordering what you ordered last spring.

The practical difference from the 2019 spreadsheet is that you no longer have to clean the data first. You can hand over the messy sales export with the renamed item codes, the reminder list showing who is due between March and June, your wellness plan enrolment count, and last season's actual stockout dates, and get back a week-by-week picture rather than an annual total. Annual totals were never the problem. The problem was always the six weeks in the middle of spring when everything lands at once.

flowchart TD
  A["Three seasons of preventive and vaccine sales"] --> E["Week-by-week forecast, March to June"]
  B["Reminder list: patients due Mar-Jun"] --> E
  C["Wellness plan enrolments and renewals"] --> E
  D["Last spring's stockout dates and backorders"] --> E
  E --> F["Order plan by week, checked against rebate tiers"]
  E --> G["Pricing and 12-dose bundle decision"]
  F --> H["Practice manager approves before the March order"]
  G --> H

The pricing half, which owners avoid

Most practices set product prices once a year with a markup formula and then hold them until something embarrassing happens. Meanwhile the online pharmacies reprice constantly and run promotions timed to exactly the weeks your clients are getting reminders from you.

Pricing work in a practice is not about undercutting anyone — you will not win that, and you should not try. It is about three decisions you are currently making on instinct: how a twelve-dose bundle is priced against six plus six, which manufacturer rebate you pass to the client as an instant discount versus keeping as margin, and whether your exam fee or your product margin is carrying the practice. Owners routinely discover, when they finally look, that the preventive shelf is running on thin margin while the exam that generated the recommendation is underpriced. That is a fixable mistake, and it is invisible until someone lays the two next to each other.

Note the boundary carefully: forecasting and pricing analysis belongs to product, tests, boarding and elective procedures. Setting a price on a euthanasia, an emergency intake or a hospitalised case by algorithm is a line you should not go near, for reasons that are about your practice's soul more than your compliance.

Working the numbers on one product line

An illustration for a three-doctor practice, using round numbers. Replace them with your own.

Thirty-five households at $260 is $9,100 of annual sales, or about $2,548 of margin — and that repeats every year those households stay online. Add the $3,200 write-off from over-ordering the other half of the shelf and this practice is paying roughly $5,750 a year for the March guess. If better weekly forecasting removes even half of both errors, that is about $2,875 recovered annually, against a forecasting cost that at current model prices is closer to a monthly phone bill than a software purchase.

How to prove it without taking anyone's word for it: this spring, log two columns. Every time a client leaves without product because you were out, write the date and the product. Every time something is written off for dating, write the date and the value. Do it for one season. Whatever total those two columns reach is your actual annual cost of guessing, and it is usually larger than the owner expected and smaller than the vendor claimed.

Where the forecast will be wrong, and what to keep in human hands

A forecast does not know that your senior associate is leaving in April, and her clients are 40 percent of the twelve-dose recommendations in the practice. It does not know that the new manufacturer rep has a spring promotion that will shift which brand your doctors recommend. It does not know that the mixed-animal practice fifteen minutes north just closed and you are about to inherit their book. Those are the three things most likely to break a spring number, and all three live in the owner's head, not in the sales history.

So keep the approval human, and keep it specific: the practice manager and the medical director review the weekly numbers together before the March order, and the medical director gets a veto on anything that would change what doctors recommend. Recommendation is a medical decision, not an inventory one. The day your shelf starts driving your protocol, you have made an expensive mistake that will not show up in any report.

One more limit worth saying plainly: a forecast will not save a practice whose reminder compliance is poor. If half your due patients never come in, the problem is the reminder and the phone, not the order.

Frequently asked questions

We only have three years of clean sales history. Is that enough?

Yes, and that is precisely what changed. Short, seasonal, patchy history is the case the older statistical methods handled badly and current models handle acceptably. Three seasons plus your reminder list is a workable starting point; four is better.

Should I be repricing my preventives more often?

Not constantly — clients notice, and it reads as unfair. Review twice a year, before spring and in the autumn, and treat the twelve-dose bundle and the instant-rebate decision as the levers rather than the shelf price itself. What matters more than frequency is knowing what margin each line is actually carrying, which most practices cannot state from memory.

Does this help with surgery scheduling, or only inventory?

Both, and the surgery side is often the bigger win. The same weekly view tells you when spay and neuter demand from spring litters will hit your surgery board and whether you need a relief doctor for two specific weeks in June — a decision usually made three weeks too late.

What is the smallest useful first step?

Pull one export: preventive and vaccine sales by week for the last three years. Not by month. Weekly is where the season lives, and monthly totals are the reason your old spreadsheet lied to you.

Closing the loop with the client on the phone

Forecasting only pays if the patients who are due actually come in — the order plan and the appointment book are the same problem seen twice. CallSphere builds AI voice and chat agents that answer the practice line and web chat around the clock, book the appointment when the client calls back about a reminder, and capture the details when the front desk is on another line during the busiest weeks of spring. It does not forecast your inventory — it helps make sure the demand you predicted actually walks through the door.