Business Analytics vs Business Intelligence

  “Business intelligence is an over-used term that has had its day, and business analytics is now the differentiator that will allow customers to better forecast the future especially in this current economic climate.”
Jim Davis SVP and Chief Marketing Officer, SAS Institute Inc.

The above quote is courtesy of an article reported on Network World, the full piece may be viewed here.

Analytics vs Intelligence

In the same article, Mr Davis went on to add:

I don’t believe [BI is] where the future is, the future is in business analytics. Classic business intelligence questions, support reactive decision-making that doesn’t work in this economy because it can only provide historical information that can’t drive organizations forward. Business intelligence doesn’t make a difference to the top or bottom line, and is merely a productivity tool like e-mail.

The first thing to state is that the comments of this SVP put me more in mind of AVP, should we be anticipating a fight to the death between two remorseless and implacably adversarial foes? Maybe a little analysis of these comments about analytics is required. Let’s start with SAS Institute Inc. who describe themseleves thus on their web-site [with my emphasis]:

SAS is the leader in business analytics software and services, and the largest independent vendor in the business intelligence market.

It is also worth noting that the HTML title of is [again with my emphasis]:

SAS | Business Intelligence Software and Predictive Analytics

Is SAS’s CMO presaging a withdrawal from the BI market, or simply trashing part of the company’s business, it is hard to tell. But what are the differences between Business Intelligence and Business Analytics and are the two alternative approaches, or merely different facets of essentially the same thing?

To start with, let’s see what the font of all knowledge has to say about the subject:

Business Intelligence (BI) refers to skills, technologies, applications and practices used to help a business acquire a better understanding of its commercial context. Business intelligence may also refer to the collected information itself.

BI applications provide historical, current, and predictive views of business operations. Common functions of business intelligence applications are reporting, OLAP, analytics, data mining, business performance management, benchmarks, text mining, and predictive analytics.

and also:

Business Analytics is how organizations gather and interpret data in order to make better business decisions and to optimize business processes. […]

Analytics are defined as the extensive use of data, statistical and quantitative analysis, explanatory and predictive modeling, and fact-based decision-making. […] In businesses, analytics (alongside data access and reporting) represents a subset of business intelligence (BI).

Rather amazingly for WikiPedia, I seem to have found two articles that are consistent with each other. Both state that business analytics is a subset of the wider area of business intelligence. Of course we are not in the scientific realm here (and WikiPedia is not a peer-reviewed journal) and the taxonomy of technologies and business tools is not set by some supranational body.

I tend to agree with the statement that business analytics is part of business intelligence, but it’s not an opinion that I hold religiously. If the reader feels that they are separate disciplines, I’m unlikely to argue vociferously with them. However if someone makes a wholly inane statement such as BI “can only provide historical information that can’t drive organizations forward”, then I may be a little more forthcoming.

Let’s employ the tried and test approach of reductio ad absurdum by initially accepting the statement:

  Business intelligence is valueless as it is only ever backward-looking because it relies upon historical information  

Where does a logical line of reasoning take us? Well what type of information does business analytics rely upon to work its magic? Presumably the answer is historical information, because unless you believe in fortune-telling, there really is no other kind of information. In the first assertion, we have that the reason for BI being valueless is its reliance on historical information. Therefore any other technology or approach that also relies upon historical information (the only kind of information as we have agreed) must be similarly compromised. We therefore arrive at a new conclusion:

  Business analytics is valueless as it is only ever backward-looking because it relies upon historical information  

Now presumably this is not the point that Mr Davis was trying to make. It is safe to say that he would probably disagree with this conclusion. Therefore his original statement must be false: Q.E.D.

Maybe the marketing terms business intelligence and business analytics (together with Enterprise Performance Management, Executive Information Systems and Decision Support Systems) should be consigned to the scrap heap and replaced by the simpler Management Information.

All areas of the somewhat splintered discipline that I work in use the past to influence the future, be that via predictive modelling or looking at whether last week’s sales figures are up or down. Pigeon-holing one element or another as backward-looking and another as forward-looking doesn’t even make much marketing sense, let alone being a tenable intellectual position to take. I think it is not unreasonable to expect more cogent commentary from the people at SAS than Mr Davis’ recent statements.

Continue reading about this area in: A business intelligence parable and The Apologists.


56 thoughts on “Business Analytics vs Business Intelligence

  1. This was interesting. Here’s my 2 cents. I’d have to say that I think that analytics is a subset of BI as you point out. BI actually ties to driving business strategy and change by linking data with buisness decision and objectives. Analytics can exist with out this. It’s the difference between a science experiment (analytics) and real business practice. Analysis support intelligence for greater things.

  2. The point David was trying to make by VALUELESS is that the value of Analytics, within context of BI, is much greater that business users can assume CLASSICAL BI value is much less.

    Mathematically looking at data does provide higher insihgts into the business of the enterprise. If you look at from SAS’ perspective rather than CLASSICAL BI vendors, you will find business solutions in all industries rely on analytics or mathematics to solve business initiatives, just in Banking you can see Compliance Officers use Baysian Analysis to detect Anti-Money Laundering or Fraud initiatives. Marketing Offers use clustering/association techniques to segment their clients.

    Peter consider this:
    Just compare the value a business user will get from segmenting their customers based on income, age or any other measure with drill-up/drill-down capabilities. Relatively to using mathematical models that will study the behaviour of a customer over time and then run the segment for a business content such as a campaign. You will find that Business Users will appreciate the later and relatively the CLASSICAL BI (Query/Reporting) will be valueless and hardly used.

    Leader in BI and Analytics is because Consultancy firms like Gartner have 2 different assessments and SAS is a leader in both.

  3. Hi Fadi,

    Are you the Fadi Jaber who is Business Development Manager at at SAS Institute (as per If so, it would probably make sense to disclose this in your comment.

    I have an MSc in mathematics and probably don’t need to have its value in some business circumstances pointed out to me.

    I don’t agree with your definition of “classical bi” as query/reporting either.


  4. Maybe its Fadi, Maybe its someone who is using Fadi’s name. I like Akismet too – I got spam attacks after I wrote some posts on the open source debate .However there are literally billions of dollars at stake here , as BI seems to be the next round of consolidation , mergers and alliances.Also in analytics there are new challengers like R (both open source as well as commercially supported), WPS ( A Base SAS clone – Peter you need to check these guys — they figured SAS as a language can not be copyrighted so created a clone at 1/10 th of price ) and other traditional sources.SAS Institute is the market leader in Analytics for three decades now, thus provoking passionate responses both for and against. And it is probably the only private entrepreneur company left in this space ….

  5. […] Post from Peter Thomas, Business Intelligence Guru. Bottom line, it’s all fluff. I don’t like the term business analytics; it doesn’t tell me anything. Frankly, I think business intelligence as a term is downright laughable, too. What does that mean? […]

  6. So historical data is useless? Try telling that to an insurance company or a bank.

    I see as a broader term Business Intelligence being more focused around the distribution of information, whereas business analytics more about how an individual can delve into/contrast and compare the information once it arrives.

    • Hi Barney,

      No BI expert will ever say Historical Data is useless. On the contrary, Business Analytics relay on years and years of historical data to come up with a model. For example, Credit Scoring for Credit Cards will need at least of 12 months of applications and their scoring result to come up with a model to predict the delinquecy probability.
      Business Analytics is a sub-set of BI and they both serve to better understand the BUSINESS of an enterprise, i.e. give more intelligence.

      Comparing BI to BA, is the way & means each analyse data. BI as in CLASSICAL BI reports looks, tells and explains what happened in the past and BA looks into the future.

      By CLASSICAL BI I mean slice & dice, OLAP, query & Report generations.
      BA is the use of statistics, Descriptive Analysis, Predictive Modeling, Forecasting & Optimization.

      This is only my view that I would like to share I hope it provides some insights on the difference between BI & BA from different perspectives.
      Kind regards,

      Fadi Jaber – Business Development Manager, SAS

    • Hi Fadi,

      Thanks for coming back – glad I didn’t scare you off!

      My point – if I have one – is that one reason (and there are others*) people look at the past in OLAP etc. is to take decisions about the future. For example, if the profitability trend for Product A is up in the last six months and that for Product B is down, then we should put more resources behind Product B. I’m sure that you can think of other such examples.


      * E.g. assessing the performance of units / countries / individuals / marketing campaigns etc. etc.

  7. Hi Peter,

    Thanks for your comment, I am certainly enjoying this discussion.
    I would just like to elaborate further on my above Credit Scoring example to illustrate the use of BA to gain more insights into the future.

    Credit Scoring uses Predictive modeling techniques, and after it’s put into action. All new credit card applicants will get instant approval or disapproval. The output of every application is a score “The likelihood of the applicant to default on payments”.
    So here, a bank knows prior to accepting the application whether or not the applicant will default on payments. Here we are talking about the future, we know based on previious applicants the percentage & likelihood of defaulting “Delinqeuncy rate”. Assuming the model works, the percentage should accurate enough on 5% tolerance.
    The model being accurate means that the bank knows defaulters before they accept them. Hence reducing credit risk.
    The example you illustrated above, does not know anythink about the future. It knows based on past performance of Product A & B that they need to move resources around.
    But it does not provide the performance increase in product B if we moved resources.
    BI helps business manager to better make decisions and it would definetly help future performance but it does not predict anything in the future.

    I believe we are on the same page now. You have explained it perfectly above “Is to take decisions about the future”. I would add, does not predict the future.
    In the case of BA, the future decision is already given to the business user in the BI report. And that is the value of BA David was explaining. I disagree with the VALUELESS term he used.

    I hope I have been helpful in this discussion.

  8. Fadi,

    In this I’m using the notation that BI = BI\BA. In truth, BA ⊂ BI but it will get too confusing if I write that way.

    Both BI and BA use historical data – there is no other kind. Both BI and BA produce information that can be used to influence future decisions. Neither BI nor BA “know” anything about the future (nor does anything or anyone else, except maybe Nostradamus). BA can be used to make statistical inferences about the future, based on the past. BI can be used to estimate future trends, based on the past.

    As far as I am concerned, the only difference between BI (as defined above) and BA is the latter’s use of statistical modelling techniques. Then the data that is used in these models tends to often come from BI in the first place.


  9. I’ve seen some evidence that there is a rebranding effort a foot- where certain vendors are trying to distance themselves from the term “Business Intelligence” or at least “Big BI”. As those of us in the profession know, there are a lot of Business Intelligence project failures, and as with many things they get higher viability than the successes. But changing the words or debating definitions on and on isn’t as useful as working on the underlying challenges. Nor as fun.
    On the subject of forecasting- although of course its worth trying, and can have value, it’s important to remember that on the whole it’s never been very successful. Read the Black Swan, a good overview of how successful predictive models can be. (Or just look at the current financial situation.)

  10. Thanks for the comments James.

    As a Pure Mathematician by background, I have at least a rudimentary understanding of statistics and agree about the mixed benefits / limitations of models. The failure of LTCM , with Nobel laureates coming out of their ears, comes to mind even before the current problems.

    Like anything else, quants are a useful tool, but relying solely on them is a mistake, as many organisations are currently finding out.


  11. This is a good thought provoking article which illustrates perfectly why IT is so often alienated from the business. They argue about things that make other people shrug shoulders.

    Take a hundred C-level people and ask them what the difference between BI and BA is. I would bet the family farm that perhaps a handful will venture ot guess. It is just technical marketing bubble (and yes, a new name helps to sell software to the gullible).

    But who cares? This is all mere semantics which gets in the way of much bigger and worthier challenges in this field, such as making right business decisions based on data.

  12. You can analyse ’til the cows come home. If you’re doing the wrong things it still doesn’t help. You need to be doing things that support your strategy (often assumed, rather than the reality) before you get any real value out of analysis.

  13. Peter,

    I agree with Adrian’s point to a large degree. What I have witnessed all to often is that the information goes no where (information for the sake of curiosity) and we lack any discipline on how to use the tools (cubes, reports, dashboards, analytics, etc).

    What to me is far more important is to use the tools in the right place in the right manner. The tools should be used to support a management process or in conjunction with one. The management processes should support the strategic objectives and goals of the organization.

    There are some clear and extremely valuable uses for predictive analytics, and there are some areas where predictive analytics could be of additional value to the organization. There are a number of areas where a company just needs fairly simple reporting to get the job done. We do not need to drive a tank to the store to get milk.


  14. A couple points:

    1) “Business analytics” can be viewed as a descriptive term (ie: the type of analytics performed by businesses) or a marketing term. “Business intelligence” seems to have lost all meaning as a descriptive term and is used solely as a marketing term. “Business analytics” is still in the process of being defined as a marketing term.

    2) Marketing terms get defined differently by each player in the market. Ultimately, a market consensus develops as to the accepted definition of the term, though this definition can change over time.

    3) Whether we call it “business analytics”, “decision intelligence” or something else, a consensus is developing that the traditional BI platforms lack a category of capabilities that are being provided by “analytics” vendors.

    4) My personal view is that “business intelligence” focuses on tools to provide information, while “business analytics” focuses on tools to provide insight. BI certainly includes some analytic capabilities and business analytics certainly includes some BI capabilities, but this does not negate them being separate categories.

    5) As a “visual analysis” vendor focused on the analytics side, we divide end-user interfaces into three groups: reporting, dashboards/scorecards, analysis. While dashboards and scorecards can be viewed as a type of report, the interaction metaphor is distinct enough to consider it separately. Likewise, analytical interfaces use a different interface metaphor than reporting interfaces.

    In the market, there are visual analytics vendors like us (Lab Escape), Spotfire and Tableau, collaborative analytics providers like GoodData and Swivel and statistical and predictive analytics providers like SAS and Netezza, which provide a category of capabilities not included in the existing traditional BI packages. The goal of all these vendors is providing the processes and tools to create insight, not simply provide information.

    Placing these new capabilities under an umbrella term like “business intelligence” only serves to dilute a term which is muddled enough already. So whether we call it “business analytics” or something else, we need a new term to call these new capabilities.

  15. Trevor,

    Thanks for your post, which certainly ran to more than “a couple of points”!

    Perhaps insight versus information is rather splitting hairs. What is information for if not to provide insights?

    My central point is nothing to do with whether BI and BA overlap, are subsets of each other, or wholly distinct. It is that it is pretty stupid to claim that BI is useless because it relies on historical data, whereas BA is somehow different as it relies on… er… historical data as well.

    Will take a look at your site and see if anything strikes me as interesting.


  16. This article is worth reading.In my perspective, Business Analytics is the heart of BI that doesn’t incorporate the real business decision making, however the steps that lead up to the decision.

  17. Late to this one – been a little busy; one thought is enough. How can I do predictive or prescriptive analytics without descriptive (and some would add diagnostic) analytics? Seems a little odd we are even having a discussion like this, guessing some of us are really not as close to metal as needed. BI? Analytics? Different? Really? See the INFORMS taxonomy or better yet try to build a predictive model or attempt auto-classification with Machine Learning (as an example) without some existing BI and see what happens. Thx for bringing this to my attention Jim Davis knows better, thinking this quote was taken out of context they (SAS) are usually better than this ;

    -jdp (Parnitzke)

This site uses Akismet to reduce spam. Learn how your comment data is processed.