Data beats intuition in a fair fight
You’re asking your people to take a leap of faith when they use data. Data analysis feels a lot more abstract than experience-based insight, so they have to trust the data is accurate and relevant. If they don’t, they’ll have a hard time trusting it.
When people use data, they want to know:
How reliable it is — data quality, doesn’t contain duplicates, etc.
If it’s the right data for the job
The definitions of terms
How metrics were calculated
How the data was produced
That it’s recent enough to be relevant
Who owns the data and how they’re maintaining it
Whether anyone else has used the data, and with what results
And most data consumers need all that information to be clear and accessible — or they won’t understand what’s going on.
Productizing datasets is a great start for most organizations, but that takes gathering a lot of information about the data, establishing clear visibility and accountability, and making the data widely accessible.
The work doesn’t end there: You also need to maintain reliability and monitor for new issues, as well as support data freshness.
TELUS improves trust in data with the help of Collibra
1New Survey: Data Will Make or Break Workers’ Trust in AI
2Trusted business reporting starts with trusted data
And you need a way for risk, compliance and legal stakeholders to monitor use of the datasets, from reviewing new use cases to applying policies.
These three principles can help you instill trust:
Unify governance so data silos don’t get in the way of consistent governance — delinking control and visibility from specific data systems, sources or compute platforms
Apply pragmatic automation to address manual tasks like monitoring data pipelines
Bring technical and nontechnical users into the same governance system
Boost confidence in business intelligence
IDC reports only 44% of people trust the data used for business reporting. Governance can change that. You can increase confidence in data-driven decisions by using a system that lets people easily determine whether the data fits their use case, and certifies the data, metrics, models and related terms with verified definitions and rules.2
54% of knowledge workers say they don’t trust the data used to train AI1
