BAD data. BAAAAAAAAD data.
Okay, maybe “less-than-stellar-quality” data, if you want to be PC about it.
But you see the “way-less-than-stellar” impact this data is having on your ostensibly data-driven organization. On business-critical questions like:
- Which product line should we invest in – or adjust – or market differently?
- Which customers are the most valuable, or most vulnerable to competitors, or have the most potential to up their spend with us?
- Which sales strategies bring in the most customers, or the most loyal customers, or the highest revenue?
When business users complain that they can’t get good enough data to make these types of calls wisely, that’s a big problem.
And when business users don’t complain, but you know the data isn’t good enough to make these types of calls wisely, that’s an even bigger problem.
How are you, as a data quality evangelist (if you’re reading this post, that must describe you at least somewhat, right?), going to convince top-level management that adopting a data quality strategy pays big dividends?
Tie data quality directly to business objectives
“Better data quality? How will it directly impact our bottom line?”
If you can’t respond with succinct, specific answers, you’re nowhere near ready to approach top-level management with a data quality strategy.
What kind of answer qualifies as good enough?
“Currently the data available to our sales department regarding customer contracts is not timely and complete enough to facilitate reaching out with relevant upgrades right before contract renewals. If we could improve the completeness and timeliness of the relevant data sets, we could increase annual revenue and customer LTV.”
THAT’s the kind of answer you need.
But, assuming you can’t already rattle off 10 examples like that for your company, where are you going to get that information?
It’s time to do a little bit of qualitative research.
(If you are such a data quality evangelist that you shy away from all but the most cold, hard, quantitative data, this may be a little difficult for you. But keep a stiff upper lip. It’ll be worth it.)
The goal is to conduct interviews with representative stakeholders in the area to which your data quality strategy would apply, whether that is enterprise-wide, relevant to one particular business line, or related to a shared repository. Stakeholders might include data owners, senior business data experts, senior data architects, and IT program managers. You will want to cover the picture from all relevant angles while keeping the number of interviews small enough to take no longer than two or three weeks.
Interview questions should elicit descriptions of current data problems, the business challenges they cause, what changes or improvements they want to see in the future and what practically that would enable them to accomplish.
Remember the “good enough” answer above? (If not, scroll back up and look for the italicized paragraph.) That’s the kind of information you’re aiming for; let that be your compass in designing your questions and guiding the interview.
After you’ve finished conducting the interviews, it’s time for analysis and organization. We recommend recording the interviews and then transcribing them so you can easily analyze the complete set of information.
Based on the responses, map out:
- Areas where poor data quality has a concrete impact on business goals
- Concrete business benefits that would be realized by improving data quality
Then order these problems and benefits by greatest impact to your organization and its objectives.
Now when you’re asked, “Better data quality? How will it directly impact our bottom line?”, you’ll have an answer.
Get cross-departmental support
Using these stakeholder interviews to Identify data quality issues and their business applications does more than just get you information. It gets you cooperation, interest and support.
Even if your interviewee from the Customer Relations department hadn’t so far put lots of thought into the impact poor quality data was having on his department’s success (or lack thereof), it’s certainly going to be on his mind now.
Even if the Head of Product didn’t heave daily sighs over the inefficiency of her department’s new product version iteration process, now she may find herself looking at progress reports and shaking her head regretfully.
The wider the net of stakeholders that see a concrete value in improving your organization’s data quality, the more frequently top-level management will hear about it.
Hearing (non-stop) is believing.
Do a Data Quality Pilot
Before you go to top-level management with a proposal for an enterprise-wide data quality management strategy, consider running a data quality pilot. A pilot involves picking one small, specific – but significant! – dataset, and taking concrete steps to analyze and improve its quality.
The ROI for a data quality pilot can be huge:
- Raising the awareness and profile of data quality in your organization
- Measurable business benefit for the dataset and department involved
- Creation of processes, tools, and templates that can be reused for all future data quality projects
- A concrete example to bring to management of data quality importance and impact
After you find a business department interested in volunteering for the pilot (and they’re going to get a measurable benefit out of it, so it’s totally worth it for them), execution of the pilot includes the following steps:
- Define the scope (which dataset is to be included and why, i.e. the practical business objective to be gained by improving its level of data quality)
- Create a working group involving representative data producers and consumers for this dataset
- Define quality rules, ideal quality targets and acceptable quality levels
- Select relevant data quality tools
- Develop a data quality assessment strategy and profiling plan
- Implement the data quality testing strategy to conduct the profiling effort
- Review results, analyze according to defined quality rules and levels, and do root cause analysis to determine the source of data quality issues and how they can be addressed
- Establish a quality baseline for this dataset with concrete metrics and explanations based on business objectives
- Write a report detailing results and recommendations
- Present report to your organization’s data governance body for review
- Implement approved data quality improvement strategies and standards
And, of course, milk the data quality pilot’s success for all it’s got when it comes to promoting a more comprehensive data quality strategy to executive management!
For more detailed information on running a data quality pilot (and how you can leverage it to create reusable tools and templates), this is a fantastic resource.
From data evangelist to sales expert
For most of us, our organization’s culture has not yet advanced to the point where a data quality strategy can sell itself. So for now, it’s up to you to sell it.
And if you do a good sales job, you’re sure to get buy-in.