Forecasting Parts Demand From Your Own Job History

Why this matters

A reorder point tells you when to refill a part you already stock. It says nothing about the part you are about to need far more of next month, or the one your call mix is quietly retiring. Forecasting is the upstream move: reading your own job history to see demand coming before it lands, so you stage the right parts for the season and for the work already sold instead of reacting one stockout at a time. The data is sitting in your job records. Most shops never turn it into a plan.

Forecasting is a different altitude than reorder points

Keep the two jobs separate in your head. A reorder point is reactive and works one part at a time: on-hand hits a number, you refill. It assumes demand looks like the recent past. Forecasting is proactive and works at the level of the whole operation: it asks what the next season and the booked pipeline will pull, then resets the reorder points and stocking levels before the demand hits.

Reorder points keep you level in steady weather. Forecasting is what keeps you from being caught flat when the weather changes, because it moves the numbers ahead of the curve instead of chasing it.

Three layers of demand hiding in your history

Your usage is not one signal. It is three stacked on top of each other, and pulling them apart is most of the skill.

• Baseline. The steady drip of your normal repair mix, roughly the same week to week. This is what your everyday truck and shop stock already covers.
• Seasonal. The predictable swing tied to weather and the calendar: the parts that spike when the first heat wave or freeze hits, the slow months when a category nearly stops moving. Your history from the same stretch last year is the best read you have on this.
• Event-driven. Demand tied to a specific known cause: a batch of installs you have booked, a maintenance campaign, or an aging install base entering its failure window. This is the layer reorder points miss entirely, because it has no recent-past footprint yet.

Name which layer a spike belongs to before you react to it. A seasonal surge you stock up for; a one-time event you order for that event and then stand down.

The forward signals your own shop already holds

You do not need a market study. You need to read the demand you have already created, ordered by how reliable and how near-term each signal is.

• The booked schedule. Jobs sold and on the calendar but not yet done are the strongest, shortest-horizon signal you have. Those parts are as good as spoken for.
• Sold estimates not yet scheduled. Approved quotes waiting for a date are near-certain demand a little further out. Mine them for the big-ticket parts.
• Maintenance agreements coming due. Contract work is the most predictable parts demand in the building, because you know the equipment and the visit date in advance.
• Install-base age. Units you installed years ago age into their common failure window on a schedule. A wave of same-vintage installs becomes a wave of same-part repairs later.

The pattern: the further out you forecast, the softer the signal, so lean hardest on the booked and contracted work for anything expensive or slow to source.

Forecast the category, not the single part

Individual part demand is noisy. One customer, one odd job, and a single item's month looks nothing like its average. Aggregate demand is far steadier. Forecast at the category or job-type level first ("cooling-season repairs," "water-heater replacements"), get that number roughly right, then let the mix inside the category and your reorder points sort out the individual items.

Chasing a precise forecast on one low-value part is wasted effort. Getting the category trend right, and stocking ahead of it, is where forecasting pays.

Turn the forecast into a move, then close the loop

A forecast that does not change a stocking decision is a diary entry. Make it move something:

• Raise reorder points and safety stock ahead of a known seasonal ramp, and trim them going into the slow stretch so you are not holding peak stock in the off-season.
• Pre-stage parts for booked and contracted work rather than reordering per job at the last minute.
• Flag the aging install base so the parts for a predictable repair wave are on hand before the first call.

Then check yourself. Once a season, lay the forecast next to what actually moved. Where you were badly off, find out why: a seasonal read from a year that did not repeat, a booked job that fell through, an install wave that came early. The forecast gets sharper every cycle you compare it to reality instead of setting it and forgetting it.

References

• See related: Setting Reorder Points So You Never Run Out Mid-Job, The Truck Stock That Earns Its Space
• See related: The Dead Stock That's Quietly Costing You
• U.S. Small Business Administration (SBA), inventory control and demand-planning guidance
• Trade-standard practice for seasonal demand planning and consumption-based forecasting