“We gathered that night at the hotel, but we were changed. Humility had
crumbled the walls between marketing and engineering. We had heard the
voice of the customer as a team and they were unhappy! Thank God the
whole team was there so we could decide on the spot.
appetizers, we slashed requirements. For soup we compressed schedule.
For the main course we sliced up the configuration into bite-size
pieces. For dessert we integrated customers into our testing. For
after-dinner drinks we polished the new presentation and said a short
“The next day was sunny. Breakfast was actually jovial.
We looked back at yesterday with humor, not to cover our humiliation but
because we were changed.”
The next meetings went far better. We
took wounds, but none like those at Fishkill. We continued with three
more meetings. We returned to the factory and presented our new path.
“Are you nuts?” our product board asked. “You can’t change direction midstream!”
missing a beat, an engineer quoted the customer. The product manager
explained how we’d manage technical risks. We were now convinced by a
shared experience that we knew what to do. Better yet, because we had
the data, we switched the burden of proof from ourselves to our
In December and January we met with ten
customers in South Korea, Japan and China. Changes were minor. Now,
instead of pivoting, we were building a virtual backlog.
February, we returned to Fishkill. “Congratulations, gentlemen,” IBM
said, “We apologize for November. With the train heading our way, we’ll
get on now. We’ll take two more of your current units now (millions of
dollars each) and replace all seven with the new one when it ships.”
90 days, the team pivoted the product, saved a major account, captured a
large order, cut schedule from 18 months to nine and developed a
virtual backlog. Within two years after release, product market share
was up to 70 percent. Twelve of 17 competitors were gone. The division
was the dominant global player.
Here’s why it worked:
Meet customers as a core, cross-functional team authorized to act within a defined envelop.
Report to a “board,” to explain and defend your strategy and business case.
Meet customers’ decision making teams, not just users.
on site in the customers’ habitat: Home, office, lab, factory or
playground. See what they do. Meet their colleagues, form relationships,
help fix broken stuff while you’re there and return to the site as
Test sell a validation prototype: Renderings, specs
and props. Don’t sell a vision! It will result in dangerous
Ask questions. Take notes. Ask question of fact, not opinion, then shut up. Take notes. Tag data and quotes.
Disclose limitations to surface objections. It enhances your credibility.
Conduct trial closes.
Meet in waves, two customers per day, Tuesday to Thursday, to get hit over the head by the patterns that emerge.
Find and fix bad news. Bad news early is good news.
Agile is nothing new for anyone who does software. I see it moving into the hardware world more and more as well. It's a great idea, but takes a lot of leaps of faith on the part of the team involved (and an even bigger leap from the upper management). With the story above, the amazing part isn't necessarily that this new methodology worked...it's that it was accepted immediately once the team decided upon it. That kind of sea change can cause a lot of turmoil in a company.
I love the recommended practices listed at the end of the story. Especially the part that says: •Ask questions. Take notes. Ask question of fact, not opinion, then shut up. Take notes. Tag data and quotes."
A critical part of boosting U.S. innovation is finding ways, as lean startup Steve Blank says, to help startups NOT fail. Blank and Frank Robinson have figured out that startups need to get out of the office and meet with as many potential customers as they can find. In a startup course at Stanford, Blank makes student entrepreneurs meet with at least 100 customers. This is one way we can get our edge back in the global tech competition.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.