Forecasting

One of the least favorite parts of running old-school circulation functions 20-25 years ago was forecasting sales by hour the week a catalog was in-home.

What an awful, awful job.

Fool's gold.

A catalog would have an in-home window courtesy of the USPS (Monday - Wednesday). Your printer would deliver your catalogs deep into the USPS system, and then the USPS would deliver the catalog to the mailbox.

I had one analyst 100% dedicated to forecasting, and half of what this person did was create a forecast by hour for delivery measurement purposes (which to be fair fed our call center forecast, even though the call center had been gutted by the transition of customers to online purchases).

• 8:00am = $100,000 sales.
• 9:00am = $140,000 sales.
• 10:00am = $200,000 sales.
• 11:00am = $180,000 sales.

The forecast was created based on history. We knew what slow/average/fast delivery looked like, and we had a forecast for sales for the catalog.

So on Monday morning, there would be a phone call ... usually around 11:00am.

Ring ring.

Ring ring.

Ring ring.

Kevin:  "Hello".

Chief Merchandising Officer:  "Kevin, the catalog is off by 15%. Is this because delivery is slow, or is the forecast wrong?"

Kevin:  "Maybe customers don't like the merchandise?"

Chief Merchandising Officer:  "You make d&*$ sure your forecast is accurate."

It was never be possible that the catalog was awful. It was always the forecast.

For the next four days my analyst and I would get screamed at ... by merchants ... by inventory leaders ... by the CFO ... by the CEO ... by randos.

By the following week, one week later, it was clear that the catalog was awful. Sure, delivery might have been a bit slow, but the catalog was a clunker and the merchants were accountable and attention would shift from blaming the forecast to blaming delivery to blaming my team for "mailing the wrong customers". Just non-stop petulance from those accountable.

The secret to forecasting is to not micro-manage hourly forecasts. None of us are capable of accurately forecasting anything hourly.

All of us are capable of forecasting stuff at a high level. And honestly, this is all that matters (unless you are running a call center, of course). Forecasting high-level issues allows your brand to pre-emptively prevent bad outcomes. Micro-forecasts create myriad opportunities for somebody to criticize.