Use your store data to power a feedback loop

The last few weeks have brought the season's fresh rain to our area. Rainstorms like we get at this time mean a lot of rain in a short period of time.

I learned last year that while tomatoes need water, too much will cause them to split. These sorts of storms will cause most of the fruit to split and once they split, they suffer.

In between the storms it's still hot enough to grow and ripen the fruit though.

That means I've had to be picking tomatoes multiple times per week when there's rain in the forecast. By harvesting the non-split fruit, we've had an abundance of tomatoes. Then after the rain stops I'll pick any split fruit and compost it. That lets the plant focus on the other good fruit.

This process, while a bit last-minute at time, has let us harvest platefuls of tomatoes that we would have lost otherwise.

The key is monitoring the rain forecast. Once I have that data, a feedback loop can be created with two behaviors:

• If rain is forecast in a day or two, pick everything that is starting to ripen.
• If rain has finished, pick anything that has split.

Keying behavior to a metric is a great way to build these sorts of feedback processes.

They can be built on aggregate behavior: e.g. if revenue is below 80% of the norm by the 20th, then schedule a flash sale.

Or they can be built on a trend: e.g. if AOV dips below $100 then offer Free Shipping for $120 orders. If AOV goes above $150, stop free shipping.

This is part of what being data-driven means. Using data to determine what behavior you do.

Sadly, after this next rain storm we'll probably lose the rest of the heat for the season so I'll be picking the last of the tomatoes this weekend.

Eric Davis