A Tiny Mistake

A Tiny Mistake

At four decimals even simple math becomes dangerously slippery.

By slippery, I mean that at 4 and sometimes even 3 decimals, mistakes are greatly amplified.

This is a simple example.

So I was working as a cost estimator in a plastics plant many many years ago.  We were a small shop... like $15MM/annum.  There were two of us on the estimating team and we did the same job.  We'd grab projects out of the inbox in the order we received them and do our thing.  One day my buddy, purely by chance, grabbed *the wrong project*.

It was a cute little plastic bottle maybe an inch in diameter and 2 inches tall, with a snap fit lid.  Request said that it was to be made of FDA grade material and was to house a powdered candy product.  Fine.  We both studied the specs and the RFQ.  It's always better to go over the thing with 2 sets of eyes before diving in alone.  It all seemed fine.. except we both had some thoughts on doing food packaging, which was not in our little company's sorta... wheelhouse.  We actually went to our boss about that, but he thought it was a good looking project, so let's put together a bid.  Dude goes back to his cubicle with the package.

So dude has prints.  No CAD... this wasn't pre-CAD days, but it was long enough ago that we still often did initial cost projections from drawings.  Of course the 2 main cost components in manufacturing are material and time.  Sounds simple, right?

So dude calculates the volume of this shitty little cylinder by hand.  With CAD of course volume calcs are dead nuts right off the model, but he's working from prints and so has to do the math.  Simple.  Inch diameter x the wall thickness x the height... you know pi x r^2 and that shit... and he gets his volume and proceeds onward to estimate cycle time and packaging and all that.  The quote is eventually approved and submitted. Miraculously, 5 weeks later, our plant is awarded the contract.

Here's the thing.  The project was 10MM (10,000,000) units per year.  Expensive multi-cavity tooling and *very* high volume production.  It was a cheap commodity and the pricing was out to 4 decimals.  Something like $.0121 per unit or whatever.

So everything's groovy... we tooled up (took 14 weeks), did pre-production sampling and DOE processing... dialed that shit in over 3 weeks and went into production.  Right on time.  All good.

So now we're like 25 weeks (half a year) from that fateful day my buddy grabbed the RFQ from the basket and did his math.  And our little shop has something like $425K invested in this thing between engineering, tooling, automation, packaging, material and staffing.

Well... at some point accounting calls and we're summoned to a meeting. I'm almost sure it must've been a Monday... BOHICA, y'all.  Seems candy bottle project is hemorrhaging cash.  Engineering files are pulled, quote files are gathered... it's a post-mortem clusterfuck CSI.

Going back through the file, I find dude's old notes on his volume calculations for the weight of the part.  Together we retraced his steps.  You remember I said it had a snap fit lid?  Well that means that the top of the cylinder had a groove around it into which the lid snapped.  When dude calculated the thickness of the wall of the cylinder, he went from the OD *of the groove* instead of the OD of the *cylinder*.

Picture the trench scene from *Star Wars*... dude calculated the diameter of the Death Star based on the *bottom of the trench* instead of the outside shell where the laser turrets were mounted.  In a nutshell, he muffed the weight of the part by some 35%.

So let's do some math.

The material... FDA grade gamma-sterilizable HDPE or whatever the fuck... was $2.36/#.  My dude's volume calcs said the thing weighed 22.2#/M (that's lbs. per thousand units).  Simple calc:  10MM units at 22.2#/M is $523,920 in material per year.

But the part *actually* weighed 29.97#/M.  And that's an ugly fucking dog.

29.97#/M x 10MM units is $707,292 material cost per year.  Fuck ugly dog Delta is $183,372.

Dude muffed his quote by like $.0183/unit, and his ugly dog was out there running full tilt on the floor even as we were talking about it, costing the plant 4x his annual salary.  Plant was paying ~$40/hour 20 hours per day to make these things.  By the end of the meeting we were $130 in the hole.  Which is like, employee pizza lunch is cancelled this month.

Not good.

So now it's a true clusterfuck.  The shop manager is pissed.  He storms out swearing and shuts the presses down, refusing to make any more widgets.  His ass is on the line too, monthly reports and whatnot.  Customer service chick bolts to call the customer and tell them that we cannot produce parts for the quoted price.  Last thirds of tooling and automation payments are most likely now off the table (something like $185K.)  The customer is out there somewhere *launching a brand new product* in like 3 weeks and "needs fucking parts... now."  And my estimator buddy's ass is grass.  Shit has truly hit the fan.

In the end, we built a week's bank of parts on our dime for the sake of decency and the customer sent a truck. They took all of the tooling and automation to another plant.  Dude and I stood on the loading dock and watched 25 weeks work and $300K+ annual revenue drive away down the road.

Mercifully, our boss was an old engineer and former cost estimator who understood exactly how such a thing can occur, and dude's ass was, via some intense political machinations, spared.

But here's the lesson.  Dude made a tiny mistake at a critical point in the manufacturing process (a process which starts long before parts are ever produced.)  But the *customer was equally at fault.*  Candy folks made an error in judgment by choosing the lowest bidder for their very important project simply based on saving money.  This is never the right thing to do.

Ever.

My company's low bid should have been a clear warning sign.  Alarms should have been ringing in Candyland's purchasing department.  This is the reason you multiple source.  You have to keep suppliers honest, and you have to root out fuck-ups way way up front.

It's axiomatic.  Never go with the low guy. He doesn't know what he's doing, or his business is in trouble and he's low-balling.  Never go with the high guy. He doesn't know what he's doing or he's fat and happy and doesn't need the work (you'll be a 3rd class citizen). You go with the guy in the middle.

The bullseye.

Oh... and most importantly, always quote at least 3 sources.  If you single source anything, then it's you who doesn't know what you're doing.