Finding Analytic Success — part 2

It is time for a third opinion…

Decision-First AI
Data Based
Published in
5 min readMar 16, 2017

--

Around the globe, organizations struggle to find and keep analytic success. Data repositories are structured and restructured in a never ending attempt to enable a “Center of Excellence” and a consistent “Return on Investment”. These are buzzwords that I discussed in Part 1. However, Part 2 will start with a different one — “Single Source Of Truth”.

In the boardrooms of the corporate universe, this phrase is both overused and poorly understood. Here is a secret — unless you are in a business that likely has little to gain from analytics, there is NO SINGLE source of truth. Sadder still, most data management divisions seems clueless and too many analytic teams feel this is beneath them.

Systems of Record

Let’s face it you likely have more than one channel, more than one platform, and/or more than one product. Your business is likely served by more than one vendor, if not you need to read up on outsourcing and focus on the aspects of your business that are true core strengths. Vendors, channels, platforms, and products all have Systems of Record.

Identifying a System of Record can be tricky. Auditing it is time consuming, but often quite rewarding. In my career, I have twice uncovered multi-million dollar discrepancies for world-leading banks. You want to get as close to the base data as possible. Log files, daily readout, anything frequent, raw, and unmolested can work. If all else fails, turn to finance. The general ledger can work in a bind.

Data needs to conform to the System of Record

Your IT department assures you that their data is accurate. Your analytic group assures you quite the opposite. An audit vs systems of record is the only way to figure out who is actually right. Many executives right this level of audit off as a waste of time, but if you have the P&L to support an analytic group than you are moving enough money and generating enough traffic that data integrity issues could be losing you a small fortune.

Have someone identify ALL of your Systems of Record. Have them audit ALL of your data tables. And then consider having someone audit the Systems of Record, this can be time consuming but again it could be very worth it. Auditing systems of record requires customer listening, sampling, and a myriad of other techniques that will need to wait for a later article.

Your business should be fully Instrumented

Anything that moves, costs money, makes money, or touches a customer (figuratively) needs to be instrumented. Anything instrumented will produce a System of Record and needs to be in your data warehouses. So chalk that up as one more audit.

So who is going to do all this auditing? DM/IT will claim everything is good. To admit otherwise is the equivalent of saying “boss I suck”. Worse still, your analytic team is more inclined to bitch & moan, point fingers, and sigh heavily at the prospect of carrying out an audit like this. Many just don’t have the skill set needed, which is why they often behave to poorly.

So what do you do when the home owner claims the house they are selling you is ‘near perfect’ and you don’t have the skills (or time) to inspect it yourself? You hire a third party inspector. And that is the right answer here as well.

Efficiency

A third party should be able to do work like this at a significant cost reduction to what you will pay if you staff it internally. They should have senior salaries to manage and junior salaries to crunch. Better still, knowing your business is a liability here. That’s right, the better I know the ins and outs of your operation, the more tempted I am to buy into the we’ve always done it this way or there is nothing we can do about that nonsense that pervades all internal cultures.

Outside eyes will have outside perspectives. They will not fall in with IT or analytics unless it is right that they should do so. With any luck, they may also have relationships with your vendors that are not biased by a client-provider relationship. Find a third party with some knowledge of the compliance space — a skill set in low supply among many analysts — and you really couldn’t ask for a better deal.

Just don’t imply that you might hire the third party to fix any issues they find. In fact, be very clear that you won’t. You don’t want to introduce incentive to create issues that don’t exist in order to drive incremental work. Also, don’t ask for referrals and decline them if they are offered before the audit is complete. In the end, if the consultant is really awesome, you may chose to change your mind… but play the game.

Great data fuels great analysis

If you want to find analytic success, you need to be sure that you have instrumented your business fully. You need to know that data is being collected and stored. And you need to have a System of Record for each area of your business.

The data needs to audited. The numbers should also box to the general ledger. The debate on data integrity needs to be put to bed. Once and for all…

A third-party auditor may be just the independent set of eyes you need to cleanly and efficiently put your house in order. You might also want to consider a recurring engagement or perhaps you will opt for a new set of eyes the next time. You may also want to considers auditing your Systems of Record. It may prove rewarding. The choice is yours — but hopefully now, the task is clear.

About the author — George Earl is an analytic executive with more than two decades of experience in analytics and business intelligence. Learn more here.

--

--

Decision-First AI
Data Based

Decision-First AI is an investment company focused on the future of data. We maintain this medium publication to further analytic debate and discussion.