How Much For The Set? How You Can Monetize Your Datasets
Data science websites overflow with tips for organizing and handling your data, improving visibility and gaining insight into your company’s performance. The question is, though: how do you turn this into cold, hard cash?
Buckle up, because you’re about to find out.
How Does Data Monetization Work?
Before we take you through the process of monetizing your data, let’s clarify the main ways you can boost your bottom line with data.
Here are three common use cases:
1. New revenue models
What insights can you provide that no one else can? That might be to your customers, to your customers’ customers, or even to organizations that are currently partners or peers.
Note that the real value in your data probably won’t come from simply selling raw datasets but by using that data to answer pressing questions for parties that can’t access this information on their own.
2. Boost retention
One of the most expensive and resource-intensive activities for any company is attracting and onboarding new customers.
It’s far more cost-effective, and will lead to greater ROI per customer, to keep hold of the ones you have and upsell to them when you can. Close analysis of data from your CRM or ERP, combined with external data from sources like social media platforms and review sites, will help you create more accurate predictive models, pre-empting churn and enticing customers back into the fold.
3. Prevent revenue leakage
Do you ever discover late in the day that you have an unpaid invoice? Or that you’ve forgotten to bill for a particular service?
It’s easy to let things slip through the net, but over time these small costs add up to very big losses. Using the right analytics approach, you can uncover common patterns in types of unpaid charges or late-paying customers and nip the issue in the bud. This should soon lead to stronger cash flow.
10 Steps to Monetizing Your Data
Now that we know what we’re aiming for, let’s see how it’s done.
Step 1: Map and value your data
Begin with a thorough audit and appraisal of the data you have in or available to your organization. Data on your processes and sales/transactions? On customers and partners? On your assets? Your security and compliance depends on you managing and utilizing this data in the best possible way, especially if you plan to share some of this (or insights derived from it) outside the business.
You need to know where your data is kept, how datasets interface, how relevant and valuable they are. Be specific: in what way is the data valuable? Will it help you cut costs? Streamline your operations? Could you sell it?
Step 2: Clean up the quality of your data
You need clean, consistent data that’s shared strategically across the organization and based on robust IT architecture. Don’t try to monetize anything until you’re confident about the quality and reliability of your raw data!
Here are a few pointers to get you started:
- Standardize your process (as well as the way you enter all data in-house).
- Assess whether the data you plan to use will fit neatly into the boxes and categories you’ve designed.
- If in doubt, run a pilot with a sample of data.
- Monitor where errors are coming from to help you pinpoint and fix corrupt data.
- Version-control your data into four distinct categories: Raw Data, Cleaning In Progress, Cleaned Data, and Data For Analysis. That way, if something goes wrong along the way, you can trace back to locate the problem without starting over.
- Scrub for duplicate data.
- Keep “collected” and “calculated” data values separate to avoid confusion.
- Make a note of precisely what different categories and codes in the data mean and consider whether values entered as text need to be converted to an integer before you can run calculations on them. For example, let’s say you’re a university working with data on grades achieved by students. Can you work with “A, B, C..” etc, or do these grades need to be converted to a numerical value?
Another important aspect here is metadata. If your data inventory is to be searchable and usable, it must be properly labeled and accounted for. Tackling this now will save a lot of frustration later.
Step 3: Start internally
Before you sell data products to anyone else, you need to apply them to your own business problems to make sure they work.
Extensive research by McKinsey shows that the most promising areas for data monetization are: Sales and Marketing, R&D, Supply Chain Management and Distribution, Workplace Management, Capital Asset Management, and Manufacturing. Discuss with these departments what their biggest data-related challenges or untapped resources are, and start there.
Need some inspiration? Take Uber, which monetizes location data (from picking up and dropping off customers) to create spin-off business Uber Eats. Marketing was able to closely target and personalize deals, vouchers, and specific food outlets to customers based on where they live. Or imagine you want to improve your R&D. By mining data from social media posts that reference your products, you can generate insights on things that frustrate customers — and use this to create software patches or simple accessories that improve their experience.
Step 4: Decide what kind of data you need
You can have the best machine learning algorithms known to man, but if you don’t have the right data, you won’t get decent results.
Think carefully about what dataset will be able to answer your question. Sales data is useful for defining and simulating success parameters. Competition data helps recreate market conditions to test your performance against others in your field. Marketing data helps evaluate how past strategies influenced customer decision-making. Customer intelligence gives a clearer picture of how the target audience thinks, so you can better predict their future behaviors.
Chances are, you don’t generate all the raw information you need in-house. In which case, you’ll need to bring in external data sources. That could mean a mix of public, open-source, and proprietary datasets, or it could mean connecting to a platform that already links to many different streams. The more independent sources you incorporate, the richer and more useful your insights will be.
Step 5: Pick the right tools for data science
There are all kinds of data-related platforms, tools, and technologies out there, including natural language processing tools, machine learning, and deep learning packages, BigML (for processing machine learning algorithms), MATLAB (a multi-paradigm computing environment that supports matrix functions, implementing algorithms and statistical modeling), Apache Spark (an analytics engine designed to accommodate batch handling and stream processing), right through to the humble Excel spreadsheet. Other platforms are used to bring together different data streams or even add new ones that may enrich your insights.
Make sure that whatever platform or combination of tools you go for is powerful enough to combine internal and external data streams seamlessly. It should use automated machine learning for harmonizing and compatibility tasks, fixing issues relating to old or out-of-sequence data quickly and reliably. This is no time to get bogged down in months of admin, compliance headaches, and cleaning up sloppy data. You have money to make.
Step 6: Build your machine learning
You’re now ready to dive into building and generating the best machine learning algorithms to help you tease out insights and tackle pressing problems.
Step 7: Test and prove your concept
Here’s the nerve-wracking part: based on the results of your data acquisition (or augmented data discovery if you’re using a data discovery tool), you now need to put the insights you generated into action. If they do work (by boosting revenues, cutting costs, improving retention, etc.) you can start thinking about how to translate this into externally-focused models and businesses.
… If not, tweak and try again.
Step 8: Explore the possibilities
How can you use your new ideas to tap into new markets or create new value propositions for existing customers?
Don’t put all the pressure on IT for this. Involve management from across the business. After all, they’re thinking from a sales and marketing point of view, while IT and data scientists are responsible for figuring out how to use data for machine learning in order to make this happen.
Step 9: Establish a business model
Next, drill down to an approach or precise product that works. It may be helpful to form partnerships with other entities in your data ecosystem that have already developed new tools. For example, providers that can help you tap into external data sets, or “white label” platforms you can fill with your own data and provide to customers.
Step 10: Take it to your clients
Congratulations! Once you’ve got your unique, proven data product or service figured out, you’re ready to start offering it to customers and monetizing it externally.
Remember that this is just the beginning. Data sources are multiplying all the time. Tools, strategies, and best practices are evolving. Your analytics needs, and those of your customers, will continue to change.
This is no time to rest on your laurels. You need to keep refining your product, developing new uses and adding new data sources into the mix. You’re sitting on a gold mine — so keep digging!