Main content

Don't be just a little bit digital

Peter Coffee Profile picture for user Peter Coffee March 18, 2015
Hint: doing the same old things with digital technology is probably doing the wrong things, and may not even do them better, reflects Salesforce's Peter Coffee.

At this month’s InterZone conference in Banff, Scott Larson (CEO of UrtheCast) made a statement about what it means to be a “digital company”: that it “can serve one person as a market, but is infinitely scalable.”

My tweet of that comment has been among the most-viewed things that I’ve ever socially shared, and it emphasizes—by contrast—the narrow and incremental character of many recipes for “digital transformation.”

Hint: doing the same old things with digital technology is probably doing the wrong things, and may not even do them better.

What makes Larson’s comment interesting is that it’s utterly focused on outcome, not on mechanism. It’s not about being computerized, packetized or schematized. It defies the path-of-least-resistance approach that says, “Put it in a computer somehow, and then we can say we got it done.”

Instead, Larson’s definition demands the impossible, in terms of conventional thinking about how a business will interact with its customers – and then uses previously unthinkable techmagic to get us from here to there, with focus on the “there.”

As is so often the case, people saw at least part of this coming. It’s been forty-five years since Alvin Toffler, in the still-worth-reading Future Shock, projected the progress of automated manufacturing toward “the day when diversity will cost no more than uniformity”.

What probably exceeded even Toffler’s imagination is today’s progress on the side of customer power, as well as producer capability. Almost anyone today can search out a provider, and specify a customized product without specialized knowledge in that domain, as easily as the mass-customization producer can turn out the result.

Without that power to discover and direct, we would not have a critical mass of Larson’s “one-person markets,” because it would require immense personal wealth to behave like one. We can’t all be Medici, but today producers don’t require that kind of patronage to survive while awaiting their next commission.

Instead, whether we’re speaking of wristbands, T-shirts, sports equipment, or bespoke M&M candies, someone has figured out how to parameterize the production process so that the economies of mass production are preserved even while offering enormous flexibility of configuration. For that matter, this is the crux of the multi-tenant application platform, if anyone wanted to build one of those.

If you try to do this by speeding up the conventional setup cycles of traditional manufacturing or operating models, you will fail. Instead, one must look for every fixed element in those models and make it simply and reproducibly adjustable – then provide a language, and some form of interpretation engine, by which any conceivable configuration can be summoned up at negligible cost of transition from one to another.

No more a secret

If this sounds like the secret of the Jacquard Loom…well, that’s why it’s no longer a secret. It was just too good an idea to reserve for weaving cloth.

Digital is not just the means of encoding the nouns: it also encodes the verbs, in exactly the same medium, so that in principle a verb can compose and then compel the performance of entirely new verbs. Lisp programmers love this “lambda” capability, which the cloud now performs at enormous scale.

Being a digital company therefore demands the courage to make your own life a lot harder, up front, by building systems with adjustments in places where it would be cheaper to put fixed connections – along with an apparatus to make those changes on the fly. Unless you work backward from that goal, you’ll probably build something whose digits might as well be carved in stone.

Jacquard Loom
Jacquard Loom

One doesn’t need to imagine the worst-case scenario, because real-life examples exist. In one case, a Web site offered access to a local government’s public documents – but each page of each document file contained a single image, representing a scan of one page of an original hard-copy document. The result was nearly immune to indexing or search, let alone to repurposing of pieces of the content (such as boilerplate text, or descriptions of assets or tasks).

How much work, at how much cost, had gone to create so little value? And yet, someone could certainly have checked a box and said “We now have a digital government archive.”

Before you laugh this off as extreme, ask if your own creation and storage and use of key data are living up to the digital opportunity. If it’s digital, it’s capable of perfect reproduction without error or degradation – but are your data backup capabilities designed to avoid common failure modes, and regularly tested to be sure the backups are valid? If your primary data were corrupted, whether by accident or malice, would you wind up with backups of nothing but garbage before you realized that something was wrong? Are version-control disciplines in place, so that paths of revision and correction are both auditable and reliable? Are retention requirements, privacy rules, and data-destruction disciplines codified and regularly verified?

Without these capabilities, “digital” storage preserves all the faults of traditional storage: single points of failure, inability to figure out what happened, and pockets of putrefying data that serve no business purpose but could someday create a regulatory or litigatory stink. At the same time, it creates new failure modes that older media such as parchment or microfilm have impressively resisted over the years in many kinds of natural disaster.

Converting an organization from ink-on-dead-tree media to bits, is only making that organization a little bit digital. Scott Larson would ask, “Can you serve a single unique individual as a market? Can you do so at unlimited scale?” Those are criteria that will drive an organization forward, from merely changing its tools of representation to rethinking every bit of its behavior.


Value-add means never leaving ‘well’ enough alone

A mandate for trust-first design

‘Ugly’ + ‘confusing’ doesn’t equal ‘smart’

History, horoscope, hands-on: making analytics matter

A grey colored placeholder image