Tom Blomfield was named one of the Top 5 Entrepreneurs Under 30 in Europe; in addition to developing for the Ruby programming language and co-founding GoCardless, he recently wrote a column for TechCrunch about “x-as-a-service” startups. He notes that tech startups are taking advantage of “more specialized and niche services” to get themselves off the ground faster and easier than ever before; he then compares the trend to the brick-and-mortar world, where it is significantly harder to access outsourced services without size to back you up. Here is his take on these differences:
With most online “X-as-a-Services” you can get up and running straight away. There’s no procurement or long purchasing process; you can simply sign up and go… Implementation consultants and long integration processes are usually a thing of the past. Plugging services together is a matter of minutes or hours, not weeks or months.
…The minimum required volume to use these services is often “one” with a price-point of “free.” These zero-cost trial periods make sense because the infrastructure to power the API has already been built…
In contrast, solutions offered by brick-and-mortar outsourcers are effectively custom-built each time, making the marginal cost of an additional customer prohibitively high. This, in turn, explains the long procurement processes and staggering minimum-order volumes.
Here in the Predixion offices, this reminded us a great deal of analytics expert James Taylor’s approach: think like an industrialist. At our recent thought-leadership events in Boston and Chicago, Taylor explained that one of the major challenges with predictive analytics solutions is that they are often so specialized that costs are driven up and opportunities for reusability become nonexistent.
According to Taylor, “the problem with the predictive analytics market is that it grew up when it was hard and expensive and time-consuming to build predictive analytics models… so it is a little bit artisanal.” That’s what Predixion calls ‘blackbox’ solutions, ones that are built for a certain scenario and apply only to answering one question or finding one metric. Taylor calls them “handcrafted” and continues to say, “but that’s not really what you need if you’re going to scale predictive analytics, you need a much more industrial mindset.”
He says that most companies start with extremely specific tools and hire data mining professionals to build extremely specific models… and think that later this “artisanal” base can somehow be scaled efficiently as your business grows. “You need an analytic structure that’s much more focused on lots of people building models, on collaboration and getting people to work together on large-scale projects. It’s about model management, so that once you have a thousand models you can find them, see which ones are working, which ones need to change; and it’s much more focused on workflow.”
Some people only want to include predictive analytics in a small, limited part of their business. For them, these ‘blackbox’ tools will solve their specific problem and they can move on. However, there’s a reason that James Taylor spoke to our Boston and Chicago audiences about “Predictively Enabling Your Organization:” because a predictive platform has the potential to send your business leaps and bounds ahead of the competition, while a predictive tool does not.
Blomfield noted that game-changing solutions for brick-and-mortar companies are “custom-built each time,” dramatically impeding the businesses growth with time, cost, and size requirements. Predictive analytics can be a revolutionary solution for any type of business; but, similarly, specific predictive models can take months to create and implement, and often are only useful with a data scientist to interpret the results – this puts a hefty price tag on answering one question or analyzing one metric. But when a predictive infrastructure is built for business users, at a fraction of the cost it becomes infinitely faster and easier to answer more questions, find more measurements, understand more data…
and Predict Everything.
To hear James Taylor’s presentation on “Predictively Enabling Your Organization,” click here.
To read James Taylor’s new white paper, “Enabling the Predictive Enterprise,” click here.