Global CIO: IT Execs Worry Too Much About Data Quality
How's that for a bit of heresy?
When we asked IT leaders what's the single biggest opportunity facing CIOs, the one cited the most among the nine choices we gave was "using data to drive new products and growth." That response jibes with the rising value we see CIOs putting on data analytics.
But you know what can foul up those analytics effort? Too much focus on data quality and other forms of data management, especially as those efforts get started.
Heresy, you say. Brush before bed, don't swim for 20 minutes after eating, and always strive for clean data. Are no truths sacred anymore?
But listen to two authoritative voices on this subject--one from vendor IBM and the other from practitioner Procter & Gamble--for perspective on how the quest for perfect data can get in the way of progress.
Andy Warzecha, IBM VP of information management strategy, spoke recently at the Fusion CEO-CIO conference in Madison, Wis., run by WTN Media, on how companies can get started using data in a way that sets them apart from the competition.
Step 1, Warzecha offers, is to focus on the biggest and highest-value opportunities, the ones that will sell more cars, reduce more fraud, save more lives. Focus on that thing that is most central to your company, that will get the top talent involved and excited.
The biggest barriers to broader use of analytics, according to IBM's survey of nearly 3,000 executives, are lack of understanding about how to use analytics to improve the business (38%), lack of bandwidth because of competing priorities (34%), and lack of skills in lines of business (28%). Concerns about the data (21%) and the ability to get data (24%) are less of a barrier. "The data element that many of us have been so concerned about--the data integration, the data quality, the single view of whatever--are far down the list," Warzecha says.
Warzecha's observations reminded me of the insights P&G CIO Filippo Passerini shared with me during interviews last fall. Passerini has been a champion of analytics at P&G, in a way that literally changes the conversations company employees have, from the very highest executive levels to more than 30,000 knowledge workers. He has audacious goals for increasing the real-time data P&G workers have access to in the coming years (more on those goals below).
Passerini discussed his thinking on "good enough" data when getting analytics efforts started. Even when it comes to data presented to CEO Bob McDonald and his leadership team, it's OK if some of it isn't perfect. As I wrote in my profile of Passerini, when we named him InformationWeek's 2010 Chief of the Year:
"We intentionally put the cart before the horse, because it is a way to force change," he says. For years, companies have tried to gather all the right data, build the high-powered data marts to support it, and only then build decision-support tools to exploit the data. And that approach hasn't worked.
Instead, Passerini's team gives executives a clear view of what's possible, to "use it as a catalyst to drive the right data convergence," he explains. If a data point is getting discussed on these screens [in executive briefing rooms], but it's based on models or projections rather than hard data, the pressure grows to acquire that data. Within two years, Passerini wants P&G employees to have access to seven times the amount of real-time information they have access to today.
This is how CIOs must build an analytics-driven culture. Do they need to be aware of the data's limitations and potential inaccuracies? Absolutely. Wrong data that leads to wrong decisions, or people equating pretty charts to gospel truth, will sabotage analytics efforts (and business results). And sure, some data's accuracy is more vital than other's--financial results, transactional data, and safety-related data come to mind.
But for many decisions, rough directional data is a great starting point for discussions. Is this data point trending up or down? Has it changed dramatically one way or the other of late? Might these deltas suggest a problem to look into? As long as leaders know the limitations of their data, that's a great start. Then comes the decision of how much to polish the data.
[We'll have more (polished) data from our Global CIO survey in an InformationWeek article and InformationWeek Analytics report March 12.]