The high cost of poor automation? Recently, an automated call from my utility department served as a great example of automation gone wrong.

Recent call from GRU regarding a boil water notice

In theory, an automated system that immediately notifies customers about boil water notices seems like a no brainer. In fact, this scenario sounds like the best use of automation.

One caveat… making sure the customer actually resides within the impacted area.

Take the above call from GRU to the bakery on January 22nd. The bakery received notice from the automated call advising that water should not be used for drinking or cooking.

The bakery had just started a bread batch and… well, that was immediately scrapped. We also ensured that anything using water (washing, ingredients, etc.) was stopped until we weren’t under the notice.

Only one problem… we were never under a notice… the call was an error.

The boil water notice was for an area of Gainesville on the other side of town; miles away from us. GRU doesn’t know (at this time) whether we were called for a transposed number or if our phone number was once used by another account in the past.

Whatever the reason, this call highlights an important aspect of automation…

Bad Automation is Worse than No Automation

Automation isn’t some buzz word that magically makes everything smell like unicorns and roses. It takes work to get things right (many, many tests) and more importantly… needs ongoing testing to verify value/accuracy.

We estimate the actions taken at the bakery cost us about $1,000.00 in lost product/labor.

The bigger impact? We simply can not trust automated messages from GRU.

Taking $1k from a small business works that way… we generally try not to have that impact again.

How can you Improve Quality?

Simple testing can provide great results. Even a yearly test will help reduce inaccuracies… depending on the cost to the customer, this can be monthly, quarterly, etc.

Only you know the annoyance/value matrix that best serves your customer base.

Bottom line: If you’re not willing to test for quality, you shouldn’t automate. If you’re not willing to put in the effort to ensure quality automation, you shouldn’t automate.