Seeing Is Believing
The Customer is Always … Right?
Reading minds is outside our capability.
Running a business is hard. There are many moving parts to contend with, both from the customer’s side and that of the enterprise itself. A knife’s edge of difference enables those parts to work symphonically rather than as a cacophony. The cacophony often prevails. Not for nothing is the practice of good management often characterized as more art than science, especially when “good” is a matter of perspective and bias.

We’re dealing with humans. Most simply want to make a living and provide for those closest to them. For that reason, when studying economics in college long ago, I always found incongruous the assertions of those theorists who tried to reduce human behavior and all its attendant unorthodoxies and irrationality to a series of simultaneous equations. Despite the mathematical elegance, something didn’t fit into such a neat solution. People aren’t abstractions, but I was too young and inexperienced to adequately express my misgivings about the incongruity. Plus, I wanted an A.

Time has added depth, and depth comes from time-tested experience. Experience, and hitting many walls, reveals a range of motivations.

A full career offers countless somethings that don’t fit. Customers are strange critters. Here is a sampling of three:

Scenario 1: How can we help you?

“We want to test our board.”

Thanks for clearing up that confusion. How?

“You’re the expert. Tell us.”

It’s your board. You’re the designer. What do you want to test?

“We worked with your company for some testing needs and x-ray inspection in the past. Now we are working on a new product that will require functional testing, and I would like to invite you to one of our virtual meetings. We would like to hear more on the functional testing that can give us close to 100% confidence the board is fully functioning and any other services that your company may provide. We are reviewing the NDAs, and hopefully you or your representative can join us on one of our meetings and go from there.”

Two months later, the NDA is signed, countersigned a month later. Discussions about schedules about meetings about ground rules about SOWs about testing immediately follow, one more month later.

What are you trying to find out about your board?

“That it works.”

Go to the head of the class. Four months’ waiting for one profound insight?

Define “works.”

“The green light goes on, and the red one doesn’t. The board is in design. Once we have preliminary design data, we’ll send it to you.”

Two months later:

In addition to the usual board files and documents (CAD, Gerber files, schematics, bill of materials, assembly and fab drawings, etc.), we need to know in writing what it is you want to test and how you want it tested. This usually takes the form of a detailed written statement of work (SOW) describing what you will supply; what we will supply; what tests and measurements you expect will be performed; and what outputs and data they will produce, and in what format. Expectations and deliverables – inputs as well as outputs from both parties should be made unambiguously clear by this document. The document should also note what, if any, diagnostic capabilities will be expected to be performed using the contemplated test setup. That should be the initial basis for discussion. Until you can produce that document, there is little for us to talk about.

Two months later:

We have received a preliminary netlist, bill of materials, and a set of mechanical drawings. We’ve also received a spreadsheet that appears to be either a compilation of testing ideas – a “wish list,” if you will – or a snapshot of the minutes of an engineering brainstorming session, taken directly from a white board unedited. The notation resembles C++ rather than standard written English. Unfortunately, after six months we are no closer to discerning what you want. You are the customer, and you need to help us, so we can help you. There is no shortcut to this.

Nevertheless, they persist:

“Can you now quote this?”


“Why not?”

You still haven’t told us what you want. We are not mind readers.

“We’ll get back to you. It’ll probably take a few months.”

The things one learns after college. Economic theory doesn’t speak to the problem of building the plane while simultaneously attempting to fly it.

Eight months later, we’re still waiting. A human child has a faster gestation period.

Scenario 2: “We have a board we are designing for a system intended to expose vulnerabilities in secure networks. We need a quote for flying probe testing and a thorough description of your capabilities. Oh, and can you explain how flying probe works?”

You mean stuff we’re not supposed to talk about, by agencies that don’t exist, that need capabilities that never happened, are not happening, and have no chance of ever happening, and that you don’t understand anyway? Oh, and this meeting never took place, correct?


Reading you loud and clear.

“We’ll post files for you to download on our Crypt.share account. They are very large, like the board.”

Large, like your ignorance. Crypt.share. Irony noted.

How big?

“24″ by 28″ (for you Russian spies, that’s 609.6mm x 711.2mm).”

How many layers?


How many nets?

“Approximately 14,000.”

How many installed components?

“Approximately 15,000.”

Do you have files ready to review?

“Only partial design data.”

Then I guess I can only give you a partial quote. I’ll leave a page or two out of our usual boilerplate.

“Management wants a complete quote and a time estimate for turnaround.”

When will you have a complete, and final, set of files ready for testing?

“In about 60 days.”

How many boards are you building?


Do you want us to evaluate the board for JTAG/boundary scan testing, in addition to the standard flying probe program?

“No, because that will only add delays. We’re on a tight build schedule.”

That’s a contradiction in terms with a board of this size and complexity. When do you expect to have the two assembled boards ready for test?

“In about 90 days.”

And you want a quote from us now, based on an incomplete design?

“Yes, and please base it on your fastest possible turnaround.”

That’s a relative statement. Programming and debugging a flying probe program will probably take two weeks.

“How many shifts do you operate? How many engineers can you dedicate to this task?”

Same as the number of micromanagers needed to screw a program designing LED bulbs.



“Management is hoping for a two-day turnaround.”

Management can hope for many things, 90 days out. Hope is a good virtue to have. They’ll need it (in abundance). It all looks so pretty color-coded on a white board.

“Why is that?”

Sorry, to answer that question I’ll have to charge you for consulting time.

“Can we still get a quote?”

Yes, based on the data we have, and subject to change once we have the actual production data. Call it budgetary, as in B.I.G.

Man yelling on a web call
The quote is submitted. Memorable in dollars as well as time. Predictably, there is silence. Three weeks later, I have to reinitiate contact.


“We decided to go in another direction.”

How many directions are there? What’s the real reason?

“Our EMS partner was uncomfortable about rework in the event of a board failure.”

Easy enough to solve. Just build a perfect board. Two of them, in fact. Should be a slam dunk.

“They were concerned about the logistics of handling repair and rework, and letting that process take place outside of their purview was unacceptable to them.”

So, you let your EMS company call the shots? We never discussed this subject in our conference calls.

Somebody woke up at midnight in a cold sweat, fearing a loss of control.

“We won’t be needing your services this time around. We’ll keep you in mind for future projects.”

How comforting. I’m sure you or your algorithm will do just that.

Cue voodoo doll pins. Nothing says “we’ll keep you in mind” like an eternal curse.

Scenario 3: “We have just sent you a 32-page statement of work.”

I’m impressed.

But skeptical. Multisyllable words fill up pages. Shouldn’t confuse form with substance.

“In accordance with our SOW, we’d like a quote for x-ray inspection, microscopy, dye and pry, and cross-sectioning of a sample set of boards, both before and after thermal cycling. As this is an automotive application, we’ll need a full report of findings, with high-resolution images to support those findings. In the event of a documented failure, we also reserve the right to request supplemental SEM, EDS/EDX, XRF, FTIR, and C-SAM analysis, if further investigation to an elemental level is deemed necessary by management.”

That pesky management again.

This is the third version of this SOW. The first was issued 18 months ago; the second, 12 months ago.

I can see the cutting-and-pasting. It reads like it was assigned to an intern, condemned for a summer to scribble away at requirements dictated from a textbook. Penance for the sin of being young and eager.

Why the changes?

“Other vendors’ comments were incorporated in the updated SOWs.”

Are we competing against them for this business?

“No. One vendor dropped out after we awarded them a contract. Another was disqualified after delivering substandard work.”

I see. Substandard how?

“Failure to follow every step in the SOW to the letter.”

Nice of you to keep us in mind after they flamed out. What if some of your new ex-vendor-inspired instructions are at cross-purposes with common sense?

“Follow them just the same. We are the customer.”

Well, dear customer, your cross-section requirement for four boards, at 16 locations per board, during two cycles, involves 128 cross-sections. You’re looking at a five-figure sum for that aspect of the project alone. Never mind your SOW also requires dye and pry in the same locations as cross-sectioning. That will be a bit of a problem. And we haven’t even begun to discuss the scope and cost of x-ray inspection.

“Your number far exceeds our budget.”

It also far exceeds your judgment.

We’ll keep you in mind as a future customer.

I’ve always wanted to say that.

Robert Boguski headshot
Robert Boguski
is president of Datest Corp. (; His column runs bimonthly.