Friday, July 8, 2011

Estimating and Cheating

A perennial problem in issuing proposals for automation lies in keeping everyone honest.

Here’s a proposal to automate a large portion of a task currently outsourced to a clerk.

The clerk is happy with the way things are – getting paid for work-from-home on an hourly basis.

I want to convince the client to install an automated procedure to do the boring and repetitive work, freeing up the clerk for intellectual tasks.

Nobody likes change, and outsourced clerks feel threatened by change, especially if there is even a faint whiff of “the computer doing my job”.

Preliminary Discussions

Preliminary discussions see the clerk and client (who has a long-term business relationship with the clerk) downplaying the time-cost of processing a document.

I think it takes 30 minutes per document, realistically. The Client says 15 minutes. The client wants to reduce the impact of savings in advance, to protect the clerk.

I have seen too many un-measured estimates to trust that kind of figure. I factor in mistakes, tea-breaks, and all the other delays that add to the cost of a project.

Nonetheless we will go ahead and use 15 minutes. Per document.

I maintain I can trim 80% of any manual task. That’s 12 minutes. That leaves 3 minutes, you think, for the clerk?


That leaves 15 minutes for the clerk; after all, the automation is taking place before the clerk gets the automatically-processed documents.

Here’s where we keep everybody honest.

Essentially I have to process a document with one click, leaving less than 3 minutes tidy-up for the clerk to do. That’s well within my capabilities, and I’m not worried about that.

But now I can use the “15 minutes” to my advantage.

Advantage 1:

“Let’s see what value the clerk can add to these documents in 15 minutes”.

Now the 15 minutes seems like an awfully short time.

Advantage 2:

“Here’s a batch of 20 documents; get them back to us in five hours, starting from NOW”, and let’s see how much they have improved intellectually. How much value do you get for your outsourced buck now?

Advantage 3:

Here is a fresh batch of 20 documents seen by neither of us.

Send the raw 20 documents off to the clerk and ask them to do the basic cut-and-paste and formatting manually, which they claim they can do. “Here’s a batch of 20 documents; get them back to us in five hours, starting from NOW”

At the same time process the documents with my application. The time taken will be close to zero.

Wait five hours, then compare the output of my process with the documents returned by the clerk.

If mine are better but you’ve paid $200 to your clerk, what’s your decision?

In short, when the client/clerk set an unrealistically low estimate of time taken to perform a task, I can turn that against them, if need be, and put them under pressure as part of the evaluation.

Talk to Me !

No comments: