This is a very interesting question “How much does a BI Project cost?”, and one that many people want the answer to. The standard response from IT is “how long is a piece of string?”, and that is a valid response in many senses. ie how would you know how much it will cost without understanding the detail?
The trouble is that this response doesn’t help anyone. I saw a post on LinkedIn about this very topic, and I thought it was a very good read – hence I am sharing the original blog post by SolidQ here, and I have been prompted to write this post.
The synopsis of the post above is that a BI project costs between $400k and $2m, and the biggest drivers of the cost are the experience of your BI implementation team, and the size of the user community. It is very difficult to estimate the cost, and a project can over run by 100% – 200% quite easily.
From my perspective, there is another layer of important detail to answer this question. A BI Project that uses PowerPivot and DAX will be “orders of magnitude” cheaper than a traditional BI project that uses traditional SQL Server Analysis Services.
The advantage of PowerPivot and DAX is that it is integrated on the desktop (via Excel), and can be used as a prototyping tool to build a BI solution in a much more agile way by actual users of the data. It my view, it is not unreasonable to expect that a PowerPivot BI solution would cost 1/10th of the cost of a traditional BI project. This may mean that the PowerPivot solution is less robust and less automated than a traditional BI solution (not always, but it can be true) and in some cases this may be an issue. But in many other cases, a $40k – $200k project is much more affordable, compelling and ‘fit for purpose’ than traditional BI. Some simple desktop BI solutions using PowerPivot are much cheaper even than this and can be built for as little as $10k with external assistance or can be built without any incremental cost by skilled internal staff (if they have the skill) but that does depend on the length of that string, of course!
What do others think about my ‘order of magnitude’ cost estimates for PowerPivot projects vs Traditional BI?
Great comments and contribution Dan. I am also engaged with a client at the moment that doesn’t have a good history of data management. In this case, it is due to the fact that it was difficult for anyone to get any reports in the past, and hence there has been no ‘reason’ to keep the data clean. I have spent a number of week’s preparing the data so I can use it, and now the reports can be punched out in just a hand full of hours.
Your blog about “How much does it cost…?” is very real to me, and I thought that I might add to the discussion.
I am co-owner of a small business consulting firm based in Halifax, Nova Scotia, Canada (think 400 miles north-north-east of Boston, as the crow flies). I started using Cognos when it first became a Windows-based product over 20 years ago, and started using Power Pivot about a year ago. We focus on very small to medium organizations, so people are as important as the data, at least in our world.
The cost issue always comes up, and your response is very much what we say; it depends.
We like to start with a pilot project, prove that the investment is worthwhile, and then move forward.
Let me give you the two extremes.
A local manufacturer, ISO-certified, in the pilot project, wanted us to speed up the searching of production issues, sifting through their 15-year database of past problems and finding solutions implemented for similar historical reports. This allows the maintenance/repair work to find a solution faster. Their data is clean, well organized, and very detailed. It was taking 45 minutes, on average, to search for an historical problem/solution; if the quality engineer got a match, they could fix the issue from there. If they did not get a match, the search was started again; another 45 minutes.
Power Pivot changed the search time to sub-3 seconds, using slicers to pre-screen and pre-sort the data. Additionally, charts allow analysis of costs, down-time, and the highlighting of machines and processes that need upgrading. This pilot project took one day of billable time and resulted in more work for us. The organization was already using data-driven-decision-making (we call this D3M), and their corporate culture already matched what we were hired to do.
Across town, we are working with a charity, where the data over the past 20-plus years has been mostly entered by volunteers. The “standards” for entering names and addresses were not tight and going back to make corrections to the database was a “someday” promise.
Creating what should be a fairly simple report that provides analysis of donor retention by community and/or postal code has taken about 30 days so far, and we are not finished yet. (Our first project was to analyse their direct mail campaigns, saving them enough over 12 months to pay for 120 days of our time.)
Before we took the contract, we ensured that management knew that their corporate culture was going to have to change, that we could improve the accuracy and speed of reports, but that their staff needed to better understand and embrace an overall D3M process. While we could move faster on our end, the people at the charity need time to absorb the cultural changes, and to rely more on data and less on instinct.
As their level of understanding grows, we have seen remarkable changes in their approach to data and to key performance indicators.
So, how much does it cost? It really does depend. We feel that change will only occur when the TEAM total becomes more expensive than the status quo. TEAM is Time, Effort, Aggravation, and, of course, Money. It is clear that a non-D3M organization will spend a great deal more money implementing a BI project than the actual cost of the project; human and dollar costs that should be identified up front.