This is the first in a multi-part series on launching successful data products. At Juice, we’ve helped our clients launch dozens of data products that generate new revenue streams, differentiate their solutions in the market and build stronger customer relationships. Along the way, we’ve learned a lot about what works and doesn’t. In this series I’ll take you through what you need to know to design, build, launch, sell and support a data product.
Part 1: Getting Started
The first step in building a great data product is to pinpoint a customer need and determine how your unique capabilities will solve for that need.
A successful data product lies at the intersection of the three circles in the following Venn diagram:
- Your customer’s pain point, an urgent problem they want to solve;
- The characteristics of your data which can be brought together to solve that problem;
- The capabilities you have to enhance the value of the data to make it as useful as possible.
Take the Academic Insights data product we designed and built for US News and World Report as an example of finding this intersection. (1) Their customers, university administrators, needed to understand how they compare to peer institutions and where they could best invest to improve their performance and stoke student demand. (2) US News was sitting on decades of detailed survey data and rankings to compare universities of all types. This data was unique in its breadth and historical coverage. However, the data was essentially stored in old copies of the paper magazine, not a format that was conducive to delivering insights to their target audience. (3) That’s where our data visualization and user experience capabilities helped them turn this data into a web-based analytical tool that focused users on the metrics and peer groups they cared about.
Let’s dive a little deeper into those three elements:
1. Pain Points
We’ve noticed a temptation with data products to forget the cardinal rule of any product: it needs to solve a specific problem. Without this focus, a data product comes in the form of a massive 100-page PowerPoint deck or a collection of raw data tables. There may be value in the data, but it is clear the product manager hasn’t thought deeply about their customers and what the data can do to solve their problems. I spoke to a credit card executive recently who mentioned how his bank spent huge sums of money on benchmarking reports. Despite his deep experience, he was unable to make sense of the reports he was sent. These are lost opportunities to deliver powerful data products.
“Your users are your guidepost. And the way you stay on the right path in the early stages of a startup is to build stuff and talk to users. And nothing else.” — Jessica Livingston, co-founder of Y Combinator
With data products the core question of your user is: What information or insights will let you make better decisions and perform better in your job?
Look for those unique situations where indecision, ignorance, or lack of information are blocking smart actions. Rather than solving your user’s pain, you need to enable them to solve their own pain. Physician, heal thyself.
2. What’s unique about your data?
The foundation of your product should be data that is somehow unique, differentiating, and valuable. In our experience, the right raw materials can come in a few different forms:
Breadth: Do you have visibility across an entire industry? Or population segment? Breadth allows you to provide benchmarks and comparisons that aren’t otherwise visible to your customers. One of our clients has data on the learning activities of more than 60% of all healthcare workers.
Depth: Can you explore deeply the behaviors of individual people, companies, or processes? By drilling into these activities, you may have the power to predict future behaviors or find correlations that aren’t visible to others. Fitbit tracks massive amounts of personal activity data from each individual user.
Multiple data perspectives: Are you in a position to combine data sources across industries or connect disparate data sources? By bringing together different perspectives on your subject, you may be able to answer new types of questions or explain behaviors through a multi-faceted perspective.
Naturally, having breadth, depth and multiple perspectives is best of all. Companies like Google, Apple and Amazon have profound data assets because they can both see human behaviors across a large audience and they know a lot about each individual.
3. Your value-added data package
It is seldom enough to create a data product that is simply a pile of data. That isn’t to say we haven’t seen many companies that believe that a massive data extract represents a useful solution to their customers.
People don’t want data, they want solutions.
How are you going to turn that data into a solution? There are many paths to consider:
- Visual representations that reveal patterns in the data and make it more human readable.
- Predictive models to take descriptive data and attempt to tell the future.
- Industry expertise to bring understanding of best practices, presentation of the best metrics, analysis of the data, and thoughtful recommendations. Bake your knowledge of the problem and the data into a problem-solving application.
- Enhancing the data through segmentation, pattern recognition, and other data science tools. For example, comments on a survey can be enhanced with semantic pattern recognition to identify important themes.
- Enabling users with features and capabilities to make them better in their job. The user’s ability to analyze, present and communicate insights can be a value-add to the raw data.
If you can determine the right recipe of customer need, data and value add, then you’ve gone a long way toward defining the foundation of your data product. But before getting down to designing the data product, you’ll want to get the right people in place.
4. The right product manager
We’ve helped launch data products in many industries including healthcare, education, insurance, advertising and market research. The most important factor in turning a concept into a business is a quality product manager. The best product managers have a vision for the product, understand the target customers, communicate well, are definitive in their decisions and recognize the reality of technical trade-offs. For a more complete list of general product manager skills, check out this Quora answer.
For data products, we’d emphasize a few more skills. The product manager needs to understand the data, what it represents and the business rules behind it. It helps if she is a subject matter expert, but if not, she should know when to bring in more expertise. Finally, she needs to understand the technical challenges involved with building a data product and be able to weight the impact of changes (which are often necessary as you learn more) against the benefits of launching sooner and gathering customer feedback.
5. Get stakeholder buy-in early
Kevin Smith of NextWave Business Intelligence (a consultancy focused on data products) warns: “Get the critical stakeholders involved and in agreement early or you’ll end up reciting the history of the project and why key decisions were made many times for many people.”
Launching data products is a journey that doesn’t end at the product launch. It also can push your organization into new and uncomfortable ground. These realities highlight the need to build broad support early in your process. Ask yourself:
- Is IT on board to provide development support, data access and data security resources and sign-off?
- Is the COO ready to provide resources after launch to support and maintain the product?
- Is your legal team confident that the data you’ve been collecting and incorporating into your data product is available for this new purpose?
- Is the marketing team ready to support a product launch that includes all the resources, collateral and creativity required of any new product?
- Is the sales team in place to understand the product, the target audience and establish the sales framework for pushing the product?
Originally appeared on Juice analytics blog.