What challenges do BI vendors face in South Africa today? Is BI just another acronym, full of sound and fury, but ultimately signifying nothing?
Brainstorm invited a number of BI vendors and integrators to a round table discussion to discuss customer attitudes, data quality and how to win quick and win big.
Present were Sean Wainer, country manager, Microstrategy, David Maron, business development executive, Business Objects SA, Martin Rennhackkamp, technology director, Prescient Business Technologies, David MacWilliam, Cognos SA, Ashley Ellington, MD Softline Enterprise, John Olsson, director Ability Solutions, Chris O'Connell, COO at KID, Andrew Connold, MD Synergy Computing, Estelle de Beer, BI manager at Sybase SA, and Carel Badenhorst, BI product manager at SAS Institute SA.
Brainstorm: How mature are current BI solutions?
Wainer: In terms of usage, not particularly, no. People know what business intelligence is but customers haven't grasped its true capability.
Maron: There's a mature uptake of the technology in terms of where it exists and which of the organisations in South Africa is deploying it locally, but there's a large issue around the ability to deploy what's been purchased. Companies need to be able to derive value from what's been deployed.
Rennhackkamp: The toolsets are mature; it's the implementation of them in customer bases that is lacking.
Ellington: Previously, the uptake that we've seen has been from larger organisations because BI cost of ownership has been an issue. In the mid-market it's evolving. Systems need to become more user-oriented rather than tech-focused. Larger organisations have been able to afford the investment. Also, BI is seen as an IT-focused thing rather than something that is user-focused.
Olsson: The problem is the definition of what BI really is. Everybody here has an informed opinion about the uptake among the user community. People tend to think BI in terms of the high-end tools, SAS, Cognos and Business Objects, but if you take the knowledge worker concept down to the debtor's clerk level, BI is just an extension of an age-analysis with a bit more intelligence, whereas in middle management, it's Excel with pivot tables and is at a much more strategic level. The cost element is true for the larger companies that have invested millions, but the common denominator in our industry is that we're notorious for messing things up. We came in with the first wave of CRM, we had the whole backlash against CRM and now there's a backlash against BI because of lack of ROI. If I was a company right now looking to invest in BI, not only would it be critical to have the right toolset, but the right partner to assist in the deployment.
O'Connell: The uptake is certainly bigger among the larger organisations, but even there you find that the deployment is limited to a select group of knowledge workers, such as the business analysts, rather than getting it on to the desktops of every decision-maker. In terms of the type of solution that gets built, it is hugely sophisticated but that mass rollout isn't there, I don't believe. That is a function of cost but also, if you look at past implementations, they didn't work and, therefore, customers are now more cautious.
Maron: Do you come across instances where everyone is happy about the BI paradigm, but the level of education about the business isn't there? So you put the tool on the desktop and people don't know what to do with it?
Connold: On the one hand it's a bit of a management issue. The cost element does come in but there are ways of working around that. But because the lower level decision-makers haven't bought into BI, they haven't been able to relate the BI solution they've been given to their business. The other thing that happens is that because businesses are dynamic (as we all know), people move on, others come in, and the focus is lost. That is something that must be handled by management.
Brainstorm: So, for it to work, BI is something that must be driven by management?
MacWilliam: Management has to have a vision, yes, but not of business intelligence per se. Management needs to have a vision of an information strategy that supports a business strategy. And business intelligence tools become a natural choice. In our experience and relating our experiences to similar-sized organisations overseas, those information strategies and the business intelligence usage that flows from that is a strategic business choice. Here in SA, business intelligence tools are bought by the IT guy. He then takes them along to the business manager and says: 'Here, I've given you the data and the tool', and the business manager says: 'But I don't need that. I'm doing fine on the back of my cigarette box'. The maturity of an organisation depends on the maturity of business intelligence deployment, which is a consequence of having an information strategy in the first place. I don't believe you can have a BI strategy - there's no such thing.
De Beer: I agree. Six years ago, none of us would have been here because we [business intelligence vendors] didn't exist. And six years ago, everyone wanted point solutions - MIS under a new name. Now they want a strategic BI solution but that has to live on a very strong integration backbone. And that's where the foundation of our BI systems lives now. It's not the front end delivery that counts, although that is where the success is measured, it's the complication of putting the backbone integration layer together. That has elevated it into a strategic layer. People are paying a lot more attention at the board level but it's still a difficult pitch to emphasise that it's the integration that counts. Customers just want to see the pretty reports.
Wainer: What I have seen is that it's a bit of a double-edged sword. On the one hand, BI has traditionally been sold to IT and IT staff are very risk-averse with their integration layers. The business manager at the top doesn't care about that. Some of us are to blame because we have been talking to the wrong people. I went to the manager of a big vendor the other day and he told me that his BI solution is a telephone: he picks it up and asks the right questions of the right people. So I think that BI has to be both a top-down and a bottom-up approach.
Rennhackkamp: There are a couple of very large institutions in this country that are flying their Boeings by looking at the stars. The top management don't realise that they need information to steer that Boeing. In those companies, it's the IT departments that are trying to get BI put in place but there's no drive from the top.
Ellington: It's a technology issue, especially to do with integration: how you correlate all the information that's coming from many different applications.
Brainstorm: Do vendors that have database and application layers have a distinct advantage over pure BI vendors? If a client was wall-to-wall with a particular vendor's offerings, would they have any advantage?
De Beer: But how many customers are really wall-to-wall of any vendor? Everyone has a bit of everything.
O'Connell: Even if you are monolithic today, tomorrow you buy a competitor and you suddenly have multiple vendor solutions. We see it over and over again in large corporates. They make acquisitions and then they have to integrate different IT systems, different databases and different CRM environments. In that situation, technology comes second to integration.
Olsson: Let's openly accept that the major players - the SAPs, the Microsofts, the Oracles - all see BI as a natural extension of their enterprise systems and they see it as a source of revenue and so on. They may even ship starter versions of those products with their own applications. But the very early point we made about knowledge and acceptance among senior executives still stands. Even if the integration is already in place, there's still a need for education and evanglisation to take place. That the integration is already there because, say, it's part of a total product, doesn't mean it's a recipe for success; you are still going to need somebody to take that starter pack and deploy it in an organisation so that there's added value.
It's amazing when you go into an organisation. Many of them have all the products on the shelf in their nice, pretty packaging. They've never deployed it. The delivery tool is not the issue; the issue is the education process.
De Beer: The penny dropped for me a few years ago when I was dealing with a big mine. We dealt with production managers and technical managers and spent a lot of time with shift supervisors, asking them what they wanted. Then we sat in front of the mine manager and said: "We're pulling out our hair here - what exactly is it that you need? How do you know now that you have a successful business?" And he said: "I look out of my window."He kept repeating this until, eventually, we walked around to his side of the desk and looked out of the window for ourselves. Outside there was a big board that was manually updated with how many tons were blasted, how many injuries there were and how many workers were on shift. So we asked him what his problem was, given that he already had this real-time dashboard. And his problem was that he only had a once-off view of it. If he wasn't there when the shift changed, then he missed it, and that information wasn't kept anywhere. He ended up being stuck at his desk five hours a day to answer questions by phone - other people asked for those quantities and he had to give them. So all we had to do was take that information and give him an electronic version with a built-in history. Of course we had to do the hard work of integrating 15 different systems together, but that was the value we provided.
O'Connell: So what you did was present him with the right information at the right time in the right format and not much more than that. You took the appropriate steps and then changed his behaviour from looking out the window to looking on his desktop, and telling him he could accept the intelligence provided by the system. That is a behavioural change that requires training and change management.
Maron: If you take technology out of it - all of us here at the table have something to offer, some of us compete, some of us complement and some of us do both - then it becomes an alignment between people and processes. If management goes away and decides that the company needs to be in a certain position by a certain time on the calendar, and that the alignment of the processes are correct, then the BI system will deliver. If it's not done correctly and the people aren't educated to do it, then the BI system won't deliver. It's a technology that must be driven by people and processes.
O'Connell: And frankly, that BI system could be as simple as an Excel spreadsheet.
Olsson: That's the point I was going to make. Too often we paint this picture of BI as only relevant to senior executives at a strategic level. It isn't. It flows all the way down the organisation. In the latest versions of modern ERP systems, even the components at the lowest levels have intelligence behind them that allows users to go behind the figures and give them the underlying information about something as simple as an age-analysis.
The other important point is that everybody has a different view of the business. BI systems need to be flexible enough to give a different view of a common set of data.
I've been personally embarrassed when visiting a boardroom and the MD compares his general ledger with the latest BI report and the two just don't tie up.
O'Connell: That is where the integration strategy becomes important. Regardless of what you plug in on top - it could be analytical cubes, it could be data mining - do you have a consistent layer underneath on which you're reporting?
Brainstorm: How many big companies in South Africa have that?
O'Connell: Very few.
Rennhackkamp: If you look at the number of routes by which data can flow to the top, it's shocking. And it's the same data. Well, supposedly the same data.
Brainstorm: So what are BI vendors doing about that problem?
Badenhorst: What we're doing is just making sure that all the disparate BI systems that are hanging around organisations - disparate because HR has bought one system and finance has bought another - are taken out of the silo of business units and up to enterprise level. So when a query comes in or when data mining is happening, then whatever is being done is being done against a consistent view of the organisation, be it finance or HR or whatever.
Brainstorm: How do you prove to a customer that there is a consistent view of the organisation?
Badenhorst: You sit with them and educate them as to the process you go through to ensure that they are getting good, quality information. There are systems reports you can give them to show that data is flowing through the system properly, that calculations are being performed correctly, and that way it becomes strategic rather than unit-focused.
Maron: So it comes back down to people and processes. You need the right people assigned to the right processes: getting the HR and finance people in a room, closing the door, dimming the lights, sliding the pizza under the door and asking them the right questions.
For example, one of the cellphone providers here wants to know: who is our customer? The guys in the sales team who sold the SIM card will say those buyers are our customers. But the activations division will say no, we don't consider you a customer until you've popped that SIM card into your phone and made a call. And the guy at the top is saying 'it's whatever my view is'. You need to get these people to sit down and decide what they need.
Badenhorst: This is not something that software can provide. It is a business decision that the customer must make: this is my view, this is the one I trust.
De Beer: That to me is the new wave of BI: metadata and business rule capture. I don't think any of us are terribly strong on that yet, but there is a lot of need in the market for it. How many customers do we know of out there with mature metadata repositories? We need to able to ask business process questions: where was something changed? Why was it changed? How can we standardise in the future? That information is becoming as critical as knowing what our revenue is.
Connold: The re-implementations that we're seeing in some of our customers, to me, is an indication of growing maturity in certain areas, particularly from the business side. Customers are learning, and have learned some lessons over the past few years and now they are reimplementing more effectively.
Brainstorm: Do customers have a realistic attitude towards BI? Are they willing to be educated?
O'Connell: Customers are extremely cynical. If you approach the IT side of the business, as we tend to do in many cases, their first attitude is 'what are you coming to sell me now? What front end or back end or data quality or integration tool is on offer?' They've been burned in the past and they already know the truth. In many cases, they claim to know it better than the vendors.
But when you approach the guys out there making the business decisions and educate them about what benefit they can get, there is large acceptance. Certainly, there are lights that go on about what is possible.
Wainer: Once again, it's selling the value to the business rather than the technology. If you walk in and say, 'I've got BI', everybody has a different interpretation of what BI is.
You need to talk about the end result, which is much more than dashboards and scorecards and reports. Maybe there's a bad reputation because of failures in ERP, CRM and whatever other three-letter acronyms that customers tried in the past without success.
Maron: Our traditional approach has been to sell value on the reporting side and what we've found quite often is that the BI was deployed on top of some data source, and when the business or IT manager saw what the data was really like, he fell apart.
He couldn't believe what utter nonsense was sitting in his back end.
The approach now seems to be around building the solid data foundation first: integration, enterprise information management, data quality, understanding what's there and understanding that front-end systems don't always capture what you expect them to.
O'Connell: You need a unified approach. Whether it's an enterprise data warehouse - I know they have a bad reputation - but whatever it is, you do need a single source of truth and that all the sourcing is done in a consistent way.
Badenhorst: This is the major difficulty we have as business intelligence vendors: the data sucks. The client often has a year's worth of work to do on their data before they can see the pretty pictures. That is why BI has a bad reputation: we are working with notoriously bad data.
Brainstorm: Is there really so much bad data in customer back ends?
De Beer: Since we represent different technologies, we get a good feel for what's happening in the market and last year, the data quality component of our business more than doubled. That says to me that that is where the problem sits.
Rennhackkamp: There are also a lot of external factors like legislation and reporting requirements that are increasing awareness of data quality and forcing companies to look at BI.
Brainstorm: Are there any quick wins to be had in BI?
Connold: There are. We had a customer who took a snapshot of the data out of his ERP system and was able to gain a huge understanding of how his business worked, and we sold lots of those products. That analysis product was enormously popular - it made us millions of dollars - and now you can't get the users away from it. But then you can't get them away from Excel either and there's no greater corrupter of data that's ever been invented.
Ellington: Yes. We deliver a BI solution with our ERP application that comes with pre-configured OLAP. It's an information tool for ordinary users. They don't necessarily understand the technology behind it but it's in a format they can understand using their own data.
Badenhorst: I would go so far as to say that you cannot implement an enterprise BI solution unless you do it in a quick-win manner. You need small iterative steps, business unit by business unit, person by person. You need a methodology that can accommodate small, quick wins. BI cannot afford to stall for a year while data quality issues are sorted out, for example.
Brainstorm: How do BI implementations vary by size of organisation? Are we different from the rest of the world?
Connold: In bigger, more sophisticated companies, the purchase of BI tends to be business-driven. It's the smaller guys who say: 'What are you trying to sell me?' The bigger guys are a bit more worldly, they've travelled, they've seen BI working overseas, whereas the smaller companies tend to be more operational.
O'Connell: Interestingly, we've seen the opposite where the smaller company driven by a visionary sees the big picture but is limited by his budget, whereas you go into some of the bigger companies and it's so departmentalised that you battle to get a solution that anybody can agree upon.
Connold: It also varies by the maturity of the organisation. The big cellphone providers, for example, are actually very immature companies. To call them federations is a compliment. Because they grew so quickly and have to invent new products all the time, it's not one company but rather hundreds of individual companies and that's why you get silos of information.
O'Connell: Our market is not very different from the rest of the world. I think we have an advantage in that many organisations are quite willing to take up technology on the leading edge. We compare very favourably to the rest of the world, it's just a smaller market.
De Beer: We tend to push technology to the absolute limit. International representatives are always amazed at what customers are doing with our technology. That pushes us as vendors to deliver because we then have to perform miracles on very small deployments.
The sum of all fears
Clearly, business intelligence has tremendous potential for growth. The figures bear this out - by 2012, Gartner estimates the vast majority of companies of all sizes will be spending money on specific BI solutions, compared with the minority today. But, as always, the hardest part of change will be the people that have to use the technology in their day-to-day jobs.
The tech may be ready; getting the users to embrace it and leverage it properly will take time. And that will take management with the vision to see what's needed to implement it correctly and the persistence to see it carried through.