Data Sufficiency – What Is Enough Data?

In Data Governance by IRM UKLeave a Comment

Print Friendly, PDF & Email

When creating a new app / product or system, how do you know when you have enough data to go live?

Liz Henderson, Associate Director, Global Data Governance Lead, Turner & Townsend
Liz spoke at IRM UK’s
Data Governance Conference & Master Data Management Summit Europe 2019 on the subject, “What Is Enough Data
The next IRM UK Data Governance Conference & Master Data Management Summit Europe will take place from 11-14 May 2020, London.
This article was previously published here.

You don’t want to disappoint or provide a poor initial experience, you also may not have the time & resource to have 100% complete data. Here I will provide an insight to different techniques to defining enough data & some considerations.

Data when rearranged and ordered becomes information
Information becomes new and useful insights
Insight leads to intelligence maybe artificial
Intelligence leads to better quality for all…the circle of life…

How much data do you need to generate that intelligence?

If you think of data like animals on safari, we don’t want to see a single lion or just a group of lions.

You want to see a variety of different animals and more than one of each I would expect…never been on safari!

But how would you know when you have seen enough animals.  Back to data, how would you know when you have enough data in your new solution to satisfy the needs of the users?

Let’s look at 4 methods that can be used to assess how much data you need.

  • User Experience
  • Confidence Level
  • Sector by Sector
  • Common Basket

User Experience

We want to delight the user by enabling 9/10 searches successful. This is along the common basket theme. Do you know what your common basket is made up of?

Do you need to consider regional variations of data?

Although enabling good search response rates, is this high success rate needed.  How often do you google something, and the results returned are not helpful? 

Consider the amount of data that would be needed to enable this high search rate of success – Maths needed here.

Do you have the time and resource to collect and publish this required amount of data?

Confidence Level

Taking the Six Sigma & American Quality Assurance (AQA) methodology, to achieve a 95% confidence level.

30 items for each of your categories are needed.

Depending on how many categories you need to fill with data, this could be quite high.

Maybe consider reducing the 30 items down a little. If 30 is 95% confidence. What would you need for 80% confidence?

Again, a little maths, how many categories, how many items per category and do you need to consider regional or other variations of the data could be different colours or styles.

Sector by Sector

Taking a sector by sector approach to data collection and system launch may allow you more time for data collection.

  • Identify the sector order & data location
  • Triage data to use most suited for the items needed
  • Prioritise the data for inclusion
  • Gap fill any categories within the sector you are working on

Is the data sufficiently different in each sector to warrant a sector approach? What other variants need to be considered?

Common Basket

Taking the Pareto principle 80/20 rule – 80% of the items searched for come from 20% of the items.

How can you identify the top 20% of the most common items – for me this insight came when looking into the sector approach as it came to light that the data was not different in each sector.

Factor in any variants required and get your data ?

Lessons Learnt

Lessons learned along the way during the journey for collecting and defining what is enough data.

  • It’s difficult! – Digital transformation, data collection, quality assuring it are all a challenge and the project should not be considered lightly in terms of budget, time, resource, complexity
  • Planning – you need to plan and plan again after you found out the quality and quantity of data available
  • Preparation – goes with planning / think of potential pitfalls and mitigate and allow extra time
  • Know where your data is!
  • Expect challenges – expect the unexpected and allow time for these challenges to be resolved
  • Team skill set and seniority – you may need to change the skill and experience mix of the team part way through the project.  Add that any unexpected challenges so you are prepared should it occur.  Changing the team mix will impact your budget.
  • Mitigation plan – what challenges do you need to mitigate for?  Don’t forget the motivation of your team, how are you going to keep that going. Try a dream board technique so you know each individuals’ motivators.
  • Manage expectations – things will go wrong and you will face unexpected challenges. How will you ensure you can manage the expectations of those above you, below and at the side of you. What is your communication strategy?
  • Working group – from the business – commitment and time to advise on the challenges to help you towards a successful delivery
  • User group testing – periodically test your solution, with a clearly defined test script to understand if you are delivering what is needed and what tweaks are needed to ensure it hits the button to be successfully adopted.

It’s not easy, but with good planning, communication, and support you can achieve success.

Liz is an Associate Director for Turner & Townsend leading their Global Data Governance. She has over 15 years experience, creating a data-driven culture for organisations, to enable monetisation of their data. From developing the vision, designing the strategy, providing strategic leadership and, advising and executing a broad range of corporate compliance and digital data transformations. With her passion for data governance she has solved problems which many organisations experience with duplicate, inconsistent and incomplete data, on multiple siloed platforms. Prior to her current role she was a consultant delivering Europe-wide programmes for example; Reduced regulatory compliance costs by 40%, Accelerated the start-up of new capital projects enabling top quartile performance and while leading the European graduate induction programme delivered revenue of $2m by expediting the deployment of new consultants. She has spent a number of years in the branded office products industry where she led the business to revolutionise the standards and governance for global product data and was an active member of the industry federation influencing the direction of the industry for data. She is currently applying her skills & experience to transform the construction industry bringing digitisation to accessing past project information. When she is not governing data, she enjoys gardening and travelling, has a data blog , is a STEM ambassador and a non-executive director for the charity Focus who support the development of young people.

Copyright Liz Henderson, Associate Director, Global Data Governance Lead, Turner & Townsend

Leave a Comment