I agree actually. Sears automatically called us about a service call. And it wasn't a big deal.

That being said, when we bought a stove at Home Depot, we received a personal call from the truck driver delivering the product to see if we were ready and where would be the most convenient place for him to carry it into the house.

So, automation probably does work sometimes, but if it's not all the time, then it's just an administration problem (as noted in the thread above)

Ya.. I wonder if you could crowdsource something like this? Pay folks a smallish commission to place the calls for you through a Twilio interface. I know of some answering services that work more or less that way, and they seem to do well. I'd imagine you can get a $2-$4 (somewhere in that range) per call, so pay a $1 commission per call. Use Twilio to record the calls to audit quality.

"Crowdsourcing" works well when workers can vote on a best solution. But you only get one crack at a customer contact.

How many customers are we talking?

If you get 10 customers a day, you treat each one like they're a duke; if 100 a day, each one is made to feel like a person. 1000 customers per day are a vague blur, an avalanche of humanity of whom only occasional glimpses are clear, 10,000 customers a day is statistics, and 10 Million per day is an abstraction.

Different businesses deal with different customers and different customer volumes change the shape and functionality of the organizations that serve them.

The question for you is, at what scale does a business built on somewhat random customer service experiences work?