Companies need to create a help desk for data, similar to the help desk they created for hardware, software, application, network, and user problems.
Can you imagine if companies didn’t have a computer help desk and each department had figure out their own computer issues? If each department had to find, load, configure, and troubleshoot their own hardware and software?
But isn’t that how most companies operate when it comes to data and data projects?
Everyday Data Challenges
Each department has to find, obtain, understand, clean, and transform data, as well as troubleshoot their own data challenges.
So each department spends many hours of frustration trying to serve their own data needs, and many of them do a poor job*, take too much time getting projects done, give up on projects, or don’t even start them in the first place. All of this impacts productivity, current and future.
*Because these companies don’t have data standards and offer assistance regarding how to find data, obtain it, store it, validate it, transform it, and analyze it properly, citizen data scientists (a fancy term for people who really don’t know how to handle data properly, at least most of them) make all kinds of stupid mistakes that lead to management making poor decisions based on flawed analyses, not to mention all the security issues the project introduced.
Some departments appoint or hire their own data champion** or team and start building their own data practice, documenting what they’ve learned for use in future projects; eventually, they gain traction and get better, automating some of their work.
**In internal audit, we call this an analytic champion. See that post for more info.
The smarter champions seek out other champions in the company and share struggles and solutions, and sometimes, even data.
Sure, some companies have high-end data analytic departments that serve departments that generate revenue or manage compliance, but try to ask them for help. They tell you to get in line, and then walk away, and then quietly laugh to themselves because they know you’ll never make it off their backlog.
Why Data Help Desks are Few
I realize that a data help desk would be expensive and the cost would show up clearly on a balance sheet. But that ignores all the soft dollars that are hidden in each department’s budget as they churn through the same exact problems most other departments are having, all on their own.
The problem is that the main cost is hidden in the salaries of the people in each department that have to deal with these data issues. The bigger the team, the more hidden cost.
But you also have to consider the labor cost and future productivity lost due to all the projects abandoned or not started simply because the labor and time couldn’t be justified. When you consider the automation that could be put in place, that’s a savings that is being lost each year.
Eventually, some senior VP is going to figure this out and create a data help desk, and eventually collect a big bonus as productivity across the company increases and data governance becomes easier.
Depending on the size of the company and the data challenges it faces, the data help desk may not be just one team, but several teams or departments working in concert. In some companies, it could be a process that flows through several areas that are already established.
But at least one team has to take the initial call and route it to those who can provide value quickly, not stick you on a backlog for 2-6 months or more.
Let’s think about computer help desks again…most of the problems are solved quickly, and the more complicated ones take more time. Priorities are assigned.
I’m not saying the data help desk needs to be staffed to give everyone an answer the same day, or in more complicated cases, the same week. But so many times, I’ve been on data projects where knowing one simple answer to a simple question delayed a project for weeks while I searched for it; other projects I would not have started or would have shut down much sooner if I could have discovered a few things early in the research stage.
What is NOT Needed
First of all, the focus needs to be on helping departments gain business value, not making their departments data driven. Your processes can be driven by data and still not produce value (e.g., see my citizen data scientist comment above).
Data-driven companies tell you where the data is, what you need to get it, and send you links to all their standards, governance requirements, and security policies, but they don’t help you solve your problem so you can provide value to the business..
It’s kind of like how no one really wants a drill; what they want is a hole. Data-driven companies don’t focus on the hole, they focus on the drill.
Value-driven companies focus on the hole. Drills come in all shapes, sizes, and costs. The challenge is the hole: where, how big, and how deep, and at what angle. The hole IS the value add.
What is Needed
A value-driven data help desk needs to be able to work with the real data nerds and less knowledgeable citizen data scientists to assist with these these types of questions:
- What data is available, and how do I get it?
- What fields and tables are available, what do the codes and values for each field represent, and how are the tables and fields linked together?
- Where does the data come from, how is it transformed, and what rules are applied? Why?
- How was this data initially validated? How often it is re-validated? What data edits are in place to catch data entry or transformation errors? Why should I trust this data?
- What are appropriate and inappropriate uses of this data? For example, some medical data cannot be used for marketing, and some internal data cannot be shared to anyone not employed by the company.
- How can I improve my data project and deliverables?
- Has anyone else in the company requested this data or doing anything like my project that I can leverage?
- What are the access rules? Such as:
- Can I access the data with my personal ID or do you only grant access to system and generic IDs?
- What training is required to access and use this data? How must it be protected?
- What company policies, laws, and regulations apply to my project?
- How do I request access?
- How often do I have to certify that I still need the data and protect it appropriately?
- Who can answer questions regarding the technology involved (how to connect, configure ODBC, REST, etc.).
- Who can help me write or troubleshoot this query? When the database fails over to a secondary location, how do I stay connected?
- Who can answer the data questions (e.g., how does this field relate to that field?, or if these people are active employees, why is a termination date present?)
- What tools has the company approved and made available to analyze this data?
- How do I learn and get help with those tools?
- How do I reduce false positives?
- How can I automate this?
- To whom should I report control failures and other problems I find as a result of analyzing the data?
- How do I report errors in the data? How fast will they get fixed?
- How am I notified of changes to be made to the data, underlying technology, and anything else that might affect the business process I built around this data?
I’d love to know what you think?
- How does your company help people with data?
- How many of the questions above can your company answer? Quickly?
- Do you agree with my definition of the problem and/or the solution?
- Do you think the cost of a data help desk provides enough ROI?
- Have the auditors in your company audited data governance in your company?