Subscribe to DSC Newsletter

Managed Analytics – A Shortcut to Success


Summary:
 If you’re having trouble getting started with data science and predictive analytics, a managed analytics program may be just what the doctor ordered.


If you’re a fortune 1000 company, one of the ecommerce giants, or one of the Big Web User companies, read no further.  You are the 20 in the 80/20 rule.  You already have invested in predictive analytics and have fully embraced both the strategic and tactical value it offers.

However, if you’re part of the 80% in this rule, you are likely somewhere between putting your toe in the water, and outright non-adoption.  The 80% isn’t just a made up number.  Gartner reports that at most 20% of companies are engaged in real data science and predictive analytics.  And that’s what’s reported by Gartner ‘responders’ who are more analytic-savvy and larger than most. 

Yes there’s a practical lower end to this scale, companies too small to be able to absorb the investment of a small but full time advanced analytics group.  In our earlier article ‘A Step-by-Step Guide for Getting Your Company Started with Predicti...’ we estimated that minimum size in the range of $50 Million to $150 Million annual revenue.  But there is a huge gap in-between these smaller companies and the Fortune 1000 with thousands and thousands of companies who could benefit but can’t seem to get started.

For this group the idea of ‘Managed Analytics’ can have some appeal and we’re going to explore some of the pros and cons here.

What is Managed Analytics?

Managed Analytics is outsourcing plain and simple.  Find a talented predictive analytics organization that provides staff, tools, expertise, and experience and make a medium to long term services deal.

Companies offering Managed Analytics aren’t tough to find.  Gartner and Forrester both have listings.  In their case however, they list only the largest (and most expensive) providers such as Deloitte, IBM GBS, and Infosys. 

But the fact is that there are tons of mid-size data science consultancies that will offer this service.  This includes many who are also platform developers who also have strong service and implementation groups who would be happy to strike a deal to bring their expertise (and their platform) into a longer term relationship.

The Benefits

Any mid-size company can start a predictive analytics center on their own but they face two challenges, both real and perceived.  1.) The real and perceived problem of hiring the right talent in the face of today’s DS talent shortage, and 2.) Not knowing exactly how to get started.

A Managed Analytics arrangement of two or three years duration can be a quick fix for both these problems.  These data science workers will be contractors in your organization and probably a third to a half more expensive than a direct hire.  But they can hit the ground running and they ought to be able to deliver something valuable within the first 90 to 120 days.

If you’re managing this correctly and have made this agreement in advance, you can slowly phase out the contract data scientists with your own folks and have appropriately paced knowledge transfer.  At the end of the agreement your people are fully trained and in charge.

Some Managed Analytics providers will structure the deal to allow you some flexibility in offering full time employment to their temporary staff.  However you find them, there needs to be a parallel path of permanent hires that begins as soon as the ink is dry so that experience and skills can be transferred.  A few mid-level folks may be analysts you already have on staff who are enthusiastic and capable of being trained up.  However, the top half should be fairly senior and experienced data scientists and big data practitioners depending on your circumstance.

The Pitfalls

There’s a saying in consulting ‘no one ever got fired for hiring (name of big name firm here)’.  And while that’s true, the Forrester and Gartner reviews point out that even the largest providers are subject to resource turnover.  The reputation and experiences of a large company provider belong to that company and not to their individual consultants.  You want to look carefully at the experience of the specific individuals who will be working for you.

The team you agree to accept, and it’s yours to specify if you wrote your contract correctly, needs to be folks with all three consulting dimensions: experience in data science, experience in your industry, and experience with the processes you want to focus on (supply chain, customer experience, price optimization, etc.).

You will also want to pay attention to the data science tool set your Managed Analytics provider will bring with him.  There’s a continuum from data scientists who code purely in R or Python, to the opposite extreme of ease-of-use analytic platforms like SAS or SPSS.  This shouldn’t be a make-or-break consideration in your planning.  Just be aware that once you start down the road of a particular tool set you are probably going to want to standardize on that toolset and hire folks who bring that particular tool-use knowledge with them.

To some extent access to specific talents can be a regional issue.  If you’re near one of the high tech hubs you can probably hire for any skill set you want.  If you are far from one of these hubs it will be worthwhile analyzing the employment ads for your region to see what skills others are hiring for as these may be the easiest to find without paying for relocation.

If you’re really far from data science resources, some portion of your team may be able to work remotely, even off shore.  If you’re not already comfortable with this, this is your opportunity to have it demonstrated by your contract team and to learn how to adapt.

How a Managed Analytics Relationship Should Begin

Your Managed Analytics partner has something to prove and you should insist that they do.  The relationship should start with an intensive full-court-press to get the new data science team up to speed and working as quickly as possible.

This probably starts with a two to five day information sharing intensive.  Some of this is to provide education to you about what is possible, but mostly this is about clarifying and agreeing what the initial business targets should be and in what priority.

There can be variety here.  Some may be focused on supply chain forecasting and inventory turns, and others on new customer acquisition or cross sell.  It’s up to you to set the agenda.  Whatever your agreement you should expect to see the working benefits of the first projects within 8 to 10 weeks.  Project identification and prioritization is an on-going process.  Likewise there should be a specific plan for operationalizing the result, training up your operating people to integrate the new tools and discoveries, and for starting the process of knowledge transfer almost immediately.

It may very well be worthwhile to have two or even three firms compete by structuring the initial week or 10 day intensive as a short term deliverable.  This means your people will have to go through the discovery process more than once, but given the stakes, it may well be worth the investment to discover which provider gives you greatest confidence.

Also, if the initial plan your Managed Analytics partner gives you is long on reports, dashboards, and alerts, but short on the use of true predictive analytics and fails to leverage big data opportunities you may have found a BI consulting firm and not a predictive analytics firm.  BI skills are easy to find, predictive analytics much less so.  There is a role for data visualization in predictive analytics to communicate results or provide actionable alerts but make sure you’re getting what you paid for, a solid program of predictive analytics and big data enablement.

For the great majority of mid-size companies and even the smaller ones that haven’t really gotten off the ground with predictive analytics, this sort of guided introduction may be just the ticket for lift-off.

 

Bill Vorhies

Editorial Director

DSC

 

 

 

Views: 2183

Tags: analytics, managed

Comment

You need to be a member of Data Science Central to add comments!

Join Data Science Central

Comment by Boris Shmagin on September 25, 2016 at 11:11am

This a god essay about Big Data in our time

Videos

  • Add Videos
  • View All

© 2019   Data Science Central ®   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service