We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.
Start a Discussion

Should BI Strive for a 'Single Version of the Truth?'

Vote 0 Votes
Should BI strive for a 'single version of the truth?'  Is such a thing still possible?

12 Replies

| Add a Reply
  • A single version of truth is impossible in many cases, unfortunately. If employee A worked 52.5 hours this week, that's a single version of truth that is crucial to the employee and the employer for wage purposes. On the other hand, a single version of truth for large companies is sometimes impossible to ascertain without enough time and effort. For example, if you are the VP- Sales, how does it matter that as of this morning, your sales was $1.456M this quarter or $1.432M? The price of ascertaining this precisely is more than its value! o it's a question of what is its value and what is the price of getting there? In 99% of the cases it may not be worth pursuing. If it is easy to get there, you should always do it. Rarely is it easy or cheap even in small companies!

  • Any efforts to create canonical model of company will result in a model that is too brittle and inflexible. BI should strive to deliver correct "local" versions of the truth and not a "global"/single version. "In BI, truth is the first casualty."

  • Yes, a "single version of the truth" is the goal to strive for -- at least for the 20% of information that is most critical to the company as a whole -- revenues, sales, inventory, customers, etc. And its a vision that requires a lot of ongoing work, calibration, and input from across the enterprise. And that "single version" will change as the enterprise changes. But all parts of the enterprise need to be working with the same, consistent information that is important to the organization.

  • The BI department in a company should strive to facilitate the single truth, but enforcing or guaranteeing the single truth is absolutely futile.

    Definitions of key metrics, easy to understand data models, and good navigation to finding the right data to use are the types of things BI managers should provide.

    But it will always be up to the users of the data to stick to the truth. Even when provided with the "true" data or metrics, it's when the data is manipulated and presented, which will almost always be outside of the BI environment in a slide deck or a spreadsheet, that the single truth cannot be controlled by BI.

  • There can never be a single version of the truth – it’s a myth!!!
    It’s all about relativity – I guess we all, at some point of time studied relativity – a guy throws a ball up while in a moving train. To him it goes up and comes down straight back to him. A guy watching this event from outside the train will see the ball travelling in a parabolic path.
    So in a nut shell it depends on who is viewing a piece of information!!!

  • My experiences have been that organizations striving or making this single version of the truth an end goal of their BI strategy have lost both visibility and insight into their data.

    I agree with Soumadeep; the truth is relative; an example would be the 52.5 hours worked and recorded. That total number may be crucial for the employee and organization; but as a project manager I may be analyzing billable hours for that given employee which may only be 20 of those 52.5 hours recorded. Its a matter of time and who is looking at the information; and the question they are ascertaining.

  • user-pic

    Isn't the phrase 'single version' an oxymoron?

    Of course there is a single version of the truth, you just can't create it using most BI or even master data management tools. The truth is, as others have already alluded to, the further away you get from single points of data (which may be absolutely true) the more difficult it is to say which interpretation is the correct one.

    The truth is, most tools designed to enable a singular reality in an enterprise are ironically not equipped to manage the ambiguous truth: that marketing's interpretation and legal's interpretation and human resources' interpretation - all of which are 'true' - need to be reconciled to that singular reality.

    Lest you think I am just being philosophical, take the case of one company in two geographically separate markets. Is the English version of the collective information assets more or less 'true' than the Urdhu version?

  • I think that when companies identify a goal of developing a single view of the truth, they do so to minimize confusion and disparate numbers within finance, sales, etc. The fact remains, however, that each department looks at and interprets information differently. Consequently, I think it's more important to ensure high levels of data quality and that the data being analyzed is valid and valuable for the individuals using BI as opposed to working on developing "one version of the truth" that may only benefit very few people.

  • I feel that BI systems should strive for a 'consistent common version of data' rather than the 'absolute truth' because 'truth' depends on the frame of reference. Also, trying to build a single version of truth results in 'Analysis Paralysis' where organizational departments endlessly argue on what is the truth and what is not. Having said that, there are some absolute truths in any organization (say Sales Targets for example) that should be defined unambiguously within the BI system.

    Bottomline, identify the organizational level unambiguous 'truthful' metrics (which should not be more than a handful) and for all other data elements define a consistent version that all business users can relate to and understand how it was derived. (Did I say 'Metadata'?)

    As they say, don't try to build a perfect system. Try building a successful one instead!

  • The "single version of truth" lies in the ability to dynamically offer the right information to the right person, at the right time. This is where a "Dynamic Service Enablement Platform™" comes into play, where the information is managed and collected in a variety of persistent storage mechanisms transformed into the appropriate ontology for the task at hand.

    Being able to dynamically and rapidly offer new versions of services coupled with the appropriate information gives us the separation of concerns that make the "single version of truth" the illusion, much as the master carpenter makes his "single version of truth" the illusion of perfection.

  • (1) A "single version of the truth" is like saying "a single version of politics". No matter how good our facts are, they are only an approximation of "truth". Different people can look at those facts and come to different conclusions about what is really happening, and should be done about it. (Crowdsourcing "truth" is one of the big next frontiers of BI -- http://12sprints.com is a step towards it)

    (2) We should strive to get a "single view of the facts" -- through master data management, data quality, data marts, data warehousing, metadata management, etc. BUT we should never expect to actually achieve this goal. There will always been new data available, and new ways of looking at it. Indeed, the whole point of BI is that it should always be changing -- once you've used BI to spot a problem or opportunity, you should implement more BI to monitor it, spot more opportunities, etc.

    (3) There's a big opportunity in the industry to bridge the gap between the "corporate truth" in the data warehouse and the "business user truth" contained in local spreadsheets, etc. For example, by letting individuals upload their data to a central corporate tool (like http://explorer.ondemand.com), and "bringing shadow IT back into the strategy".

  • Why do we hear companies talking about a "single version of the truth"? It is because of the frustration they have experienced where multiple people argue about which number is correct rather then focusing on what the metrics mean in terms of improving operational performance and business results. They want data consistency so they can understand trends, variances, cause and effect etc.

    Pragmatically it means using a consistent data source for each base metric, ensuring that KPIs are calculated the same way for everyone, and that everyone interprets these metrics in a consistent way at all levels in the organization. Accuracy is important, but consistency is more important. Relevance, low-latency, actionability, interactivity and usability of metric data are also key to information value.

Add a Reply

Recently Commented On

Monthly Archives