One version of the truth

In previous Industry Insights in this series, I defined the business intelligence (BI) market and its drivers. Now I will look at the issue of "one version of the truth" in more detail.

The very idea of one version of the truth will be driven by corporate information logistics. Data should flow through the business in a certain way and should, in theory, only reside in one place in one format for one business use. This means the transactional data resides in the transactional system in one place for that system. The management and reporting information also resides in one place and is accessible to all users and tools. The tools and users may take the data out and perform different actions on it for different needs, but once this need is met, the altered data is checked back in again.

As they say in the movies: "I will say this only once." So one version means just that. Not data replicated in many different tools for different users. Once. (This should be clear by now.)

This could mean a data warehouse, and I won't use this term as it comes with some baggage, as we all do. The in-memory and the virtual cubing worlds are here, but so what really, does the business care? Solve the problem quickly and if new stuff is used, the business will have no idea what is being talked about, but all other businesses will want one.

How to meet the immediate data need will determine the long-term success of a solution. If the build-once-use-many criteria can be met, the solution will flourish and be fruitful. If not, it is heading for the outfitters where they make the nice coats with the long sleeves that tie up at the back.

It also means, ideally, a single point of maintenance for the business logic, although this is not quite as easy, and the different needs of the business often dictate the rules are applied in different places.

This is not an excuse to create many versions of the data. So the budgeting, consolidations, scorecards/dashboards, reporting, analytics and forecasting should all be fed the same set of base numbers. One strategy to execute, one understanding of the shortfalls and one understanding of the business.

So, this is the one version. It is not that straightforward though, as all the data in all the sources do not need to be replicated. If aggregated reporting is all that is needed and the reporting tool at the lower level does the job, then the data only needs to be extracted at the level where it needs to be manipulated.

What about the truth? This is not a philosophical debate, but if companies are trying to analyse a business and are creating goals, the first step taught is how to break the goals down into measurables.

There is an answer to this, but not the one companies want to hear. Surely, the only real truth in the business is having every single possible cost allocated to the lowest possible granularity in the business, down to a customer or product level. This will then allow the company to roll the truth up in unlimited ways and give users the same answer many different ways; full profitability analysis.

Surely, the only real truth in the business is having every single possible cost allocated to the lowest possible granularity in the business, down to a customer or product level.

Keith Jones is MD of Harvey Jones.

This will mean if profit per customer is added up, it will equal the reported profit of the business in the profit and loss. If the same is done with product, the same result will occur. All costs, R&D, marketing and revenue accounted for. All deferred income, amortisations and lifetime value of a customer of product will be calculated. This is the only real truth in a business, so when I say one version of the truth, like life, I have to be careful how much truth I really want. It comes at a price.

It is better to be roughly right than precisely wrong, so businesses need to know what each level of truth it is looking for. It could be one version of a general ledger across multiple lines of business, a single balance sheet, or balancing the ledger to the sub-ledgers so there is a fully balanced drill-through. The costs should be allocated to a level where sense can be made of the business, and to the level where educated decisions can be made about where the business needs to be in the future.

So, the largest factors influencing the size and scale of the solution will be how much data is contained in the one version and how much truth there is in the data. The former needs to fit the budget and the latter needs to be of sufficient value to provide significant business value. Getting the mix right will probably be the largest factor in determining the success of that valuable first plateau in the solution.

All business users, when asked, will always ask for "everything". This excites a vendor, but the budget and timelines scare the user back into quick-fix mode. This is the cycle the industry has been stuck in for quite a while now. The current market is driven by: "I want a quick solution and I want to make sure it can scale into a full enterprise solution." This, strangely enough, is the way forward.

* Keith Jones is MD of Harvey Jones.