Subscribe by Email

Your email:

Browse By tag

Blog - Kofax (formerly 170 Systems) Perspectives on AP

Current Articles | RSS Feed RSS Feed

Understanding AP Errors
Six Sigma for Dummies

Share on Twitter Twitter | Share on Facebook Facebook | Submit to Digg digg it |  Add to delicious  delicious |  Submit to StumbleUpon StumbleUpon |  Share on LinkedIn LinkedIn | Submit to Reddit reddit 

Coaching 3rd grade girls basketball is an exercise in patience.  Lots and lots of patience.  8 and 9 year old girls have a very short attention span so you need to be patient.  Teaching skills like dribbling and shooting and passing takes patience.  Did I mention that you need to have patience?!?

Let's focus on passing as an example.  It seems like a simple enough skill - throw and catch the ball with your teammate.  But it's amazing how often a bad pass or a bad catch results in a turnover.  I would estimate that on my daughter's 3rd grade team, only about 50% of all passes are completed successfully and I am probably being generous.

As a coach, I want to emphasize teamwork through good passing - especially at this young age.  However, the more passes you make, the higher the chance that you will turnover the ball.  Here's the math:

1 pass:     1 - (.5) =                          50% turnovers
2 passes:   1 - (.5 x .5) =                     75% turnovers
3 passes:   1 - (.5 x .5 x .5) =                88% turnovers
4 passes:   1 - (.5 x .5 x .5 x .5) =           94% turnovers
5 passes:   1 - (.5 x .5 x .5 x .5 x .5) =      97% turnovers
6 passes:   1 - (.5 x .5 x .5 x .5 x .5 x .5) = 98% turnovers

So, with each pass, if there is a 50% chance of a turnover, with as few as 3 passes, there is almost a 90% chance of a turnover before all the passes are completed.  It's a real coaching dilemma - if the girls try to pass the ball more than twice, we are virtually guaranteed the ball will be turned over!

Now suppose I decided to work on passing every day for a month and our passing completion rate improved to 80%.  What would this do for turnovers?  Again, here is the math:

1 pass:     1 - (.8) =                           20% turnovers
2 passes:   1 - (.8 x .8) =                      36% turnovers
3 passes:   1 - (.8 x .8 x .8) =                 49% turnovers
4 passes:   1 - (.8 x .8 x .8 x .8) =            59% turnovers
5 passes:   1 - (.8 x .8 x .8 x .8 x .8) =       67% turnovers
6 passes:   1 - (.8 x .8 x .8 x .8 x .8 x .8) =  74% turnovers

As you can see, it's still not that great - after 2 passes, the girls will still turnover the ball more than half the time!!!

What if, by some miracle, the girls learned to pass at a 95% completion rate? 

1 pass:     1 - (.95) =                                 5% turnovers
2 passes:   1 - (.95 x .95) =                           10% turnovers
3 passes:   1 - (.95 x .95 x .95) =                     14% turnovers
4 passes:   1 - (.95 x .95 x .95 x .95) =               19% turnovers
5 passes:   1 - (.95 x .95 x .95 x .95 x .95) =         23% turnovers
6 passes:   1 - (.95 x .95 x .95 x .95 x .95 x . 95) =  27% turnovers

Now, that's a lot better! 

Now, let's switch gears.  Instead of basketball and passing, what if we were talking about payment errors in an AP process?

... well, the implications for AP are pretty profound ...

If the entire AP process consisted of just 6 sub-process steps where each step had a 95% accuracy rate, the entire 6-step process would have a payment error rate of 27% ... YIKES!

A 27% error rate is clearly unacceptable.  The implications here are 2-fold.  To get the error rate down to acceptable levels, each sub-process step must have an accuracy rate that is almost perfect AND the number of sub-process steps must be kept to a minimum.  This argues for a simplified AP process where each sub-process step is automated for higher accuracy.

In general, If you have a manual AP process that looks like this, you will have a higher error rate than if you had a simpler, automated process. 

BTW, what I have just illustrated is the dumbed-down explanation of Six Sigma.  Six Sigma is a measurement of the number of acceptable defects produced by a system and means that the output will be 99.99966% good with no more than 3.4 defects per million.

-Rakesh

All Posts