A Very Thin Post on Data Trust

Here at IData we have been introduced (or, in some cases, re-introduced) to the work of Charles Feltman, specifically the ideas collected in The Thin Book of Trust. While this monograph mainly focuses on building trust and forging productive working relationships between colleagues and across teams, many of the key tenets seem as though they apply in a more data-centric context.

Feltman identifies four dimensions of trust in an organization: care, sincerity, reliability, and competence. When trust exists, we believe that our colleagues care about our needs and concerns, that they speak the truth (at least as they understand it), that they will fulfill commitments they make to us (and others), and that they are able to perform the tasks they are given (or that they accept).

Feltman suggests that in many cases where there is a lack of trust, the situation isn’t a complete breakdown, but rather that something is lacking in one or more of these dimensions. Maybe we believe that others don’t really say what they mean, or that they don’t care about the success of the company beyond their personal sphere or business unit, and no doubt everyone has had coworkers who seemed, if not incompetent, then at least underqualified for the work they were expected to perform. Maybe we ourselves have been untrustworthy by failing to deliver what we promised, or by delivering it with significant delays or cost overruns.

At IData we’ve worked with hundreds of clients, and with thousands or even tens of thousands of individuals at those client organizations, and it has not been uncommon over the years for people to tell us they can’t or don’t trust the data at their organization. Sometimes this has really been a way of saying they didn’t trust the people collecting and managing the data, or the people who manipulate it into products, or the people who fulfill data requests.

It's easy to see where these interpersonal trust dimensions could come into play. We suspect many of us have said, heard, or at least thought something like the following:

• The business analyst didn’t provide me with the information I needed, or they didn’t provide it to me when I expected it, and so now I consider them unreliable. In the future I won’t ask them for work, or I will discount its validity when they provide it to me.
• The people who capture data don’t care what I want to do with it and won’t store a certain set of facts about certain people or transactions that I think is important. They don’t care about everyone’s success and so I can’t or won’t trust them.
• The managers who say they want to use data are insincere. They don’t really want to use data to inform their decisions, or they only want to use data that supports their preconceptions.
• The developer said they could build a data warehouse, but the data coming from it doesn’t seem any more reliable and queries don’t run any more quickly—are we sure they’re competent to do that job?

As much as we might wish otherwise, data doesn’t exist in a vacuum, unaffected by human prejudices, blind spots, and shortcomings. People make decisions about what data to capture, where and for how long to store it, how and when to share it, and so on. They decide what data will be included or excluded from analysis, and they decide whether a given analysis will be used in planning and decision making. It can be difficult to establish and maintain trust with other people, and it’s often quite sensible to remain skeptical of data, which after all reflects so much human input.

It's one thing to bring to bear a rational skepticism, and another to distrust data altogether. Most of the time what we hear when someone says they can’t trust their data is not that data professionals don’t know what they’re doing or don’t care about their work. What we hear instead, when we listen closely, is that there are barriers and gaps and deficiencies at nearly every stage of the data lifecycle.

• Sometimes “I don’t trust the data” means I don’t get the data I’m looking for, or I don’t get it soon enough, or I don’t get in a format that I understand, or—and it takes trust to admit this to ourselves and our colleagues—I don’t know enough about the data to interpret it or to determine how it might help me. This is something of a reliability issue: we think that our data has promised us something, but we don’t get that something according to expectations.
  
  
• Sometimes “I don’t trust the data” means that the data doesn’t confirm my hypothesis. (And we’re being charitable, since it’s far too rare that people can articulate a hypothesis around the data they’re using.) More often, there’s a general sense of how things are, and an expectation that if the appropriate data is assembled in the proper fashion, then the data will comport with that general sense. While it’s a bit of a stretch, to be sure, we could think of this as a situation where the data doesn’t seem to care about your needs or desires.
  
  
• Sometimes “I don’t trust the data” means “I don’t think we have (access to) the data that would answer this question.” This could be a lack of raw data, it could be a slow buildout of a warehouse, it could be a data structure your tools aren’t equipped to handle, it could be a situation where you need some data science skills but your organization doesn’t have enough of them. In situations such as these, we might say that the data isn’t really competent to perform the tasks you’d like to assign to it.
  
  
• Our final dimension for trust is sincerity, or the lack thereof. When we say, “I don’t trust the data,” and we’re really accusing the data of insincerity, what does that look like? Insincerity, after all, is when you say one thing, but don’t mean it, or perhaps mean something else altogether. One example might be poor dashboard design, or misleading data visualizations. Pie charts, for example, despite being notoriously inefficient at conveying complex information, are often the first graphic available in visualization tools. 
  
  

When a speaker doesn’t seem to speak plainly, it’s easy to wonder about their sincerity, their concern for our needs, their reliability, and their competence. When our data products exhibit the same inscrutability, is it that much of a stretch have similar doubts? It’s definitely a stretch to say that data is insincere, or that it isn't concerned about our organization's success. But can we say that our data management processes are insincere, or unreliable, or simply not up to the task?

If I make a request in good faith for something to be done to/with our data so that I can answer a question, and what I get back either doesn’t shed real light on the question, or the results, however promising, can’t be verified, it sounds like maybe my organization is not doing all it could to ensure the trustworthiness of its data.

Whatever our business is, we’re going to fall behind if we’re not using data to improve our decisions, evaluate our efforts, and drive our strategies. We must trust data in order to use it! How do make it so that our data earns and rewards our trust?

Let’s go back to our four pillars of trust, shall we? 

Are we collecting, maintaining, and perhaps enriching data that cares about us, that speaks to our needs and concerns? Frequently there’s a disconnect between what a business is trying to accomplish, and the data it collects.

When we collect, store, and maintain data, what steps are we taking to ensure its reliability? Do we have a business glossary for our data terminology? Does that glossary guide our standards for acquiring data and preserving its integrity? Is everyone responsible for data quality? (Is anyone?)

What does our organizational data competency look like? Have we devoted enough resources to our data management processes? Do we have enough staff? Do the staff receive training to keep their skills current? Are our data management tools sufficient for our needs right now, or, as is often the case, are they only barely sufficient for what our needs used to be?

Do we treat our data with sincerity? Do we have reasonable expectations for using our data, and have we articulated those expectations clearly and widely across our organization? Or have we just shared the highlights with the analysts toiling away in the data sub-basement and assumed they’ll just figure it out?

We hope you've found this a fun--but not entirely whimsical--thought experiment, mapping these dimensions of trust and untrustworthiness onto our data. We think that our data governance, data catalog, and data intelligence solution, the Data Cookbook, can be a critical element of your efforts to build trust in data, and to leverage that trust into organization success. Our data management practices look to establish a foundation for meaningful engagement with your data, first and foremost by focusing on pragmatic activities that address real needs. We'd love to talk with you about your situation. We can help to identify where your data trust needs shoring up or complete repair, and we can suggest useful techniques you can take to get started or maintain progress in this effort. 

Hope you found this blog post beneficial.  To access other resources (blog posts, videos, and recorded webinars) about data governance and data intelligence feel free to check out our data governance resources page.  And additional resources on data trust can be found here.

Feel free to contact us and let us know how we can assist.

 Contact Us

(Image Credit: StockSnap_2C05UHUIR8_frustratedgroup_thintrust_BP #B1298)