You know the janky data solution your startup currently has in place just isn't cutting it. But when you think about building what you need, you get a pit in your stomach. It feels like you've been asked to stop patching over holes in the wall and start building a fifty story skyscraper.
Listen to that pit in your stomach. It's telling you something important. Maybe you could get lucky and pull off switching from wall patches to a fifty story skyscraper. But it's not a good idea.
There's a better approach you can use, one you've seen everywhere if you've been working in the world of startups: a Minimally Viable Product (MVP).
In this article, I'll show you how to take the idea of a product MVP and morph it to adapt to the needs of a data solution—and in doing so, help your startup reach new heights.
Why data matters to your startup
To understand what it means to build a minimally viable product when you are building a data solution, first let's make sure we are clear on what your "product" needs to do for your customers, aka the folks who’ll be using the data.
- Velocity/agility: Above all else, startups need to be able to move fast and quickly adjust to changing circumstances. To pull that off, your users need more than an intuitive, vague sense of how your startup is doing. At any time, they need to know what’s going well, what’s falling flat, and what opportunities it’s worth investing in. And that requires data.
- Avoid big misses: Because things are moving so fast, your users don't just need data, they need the right data at the right time. That’s one reason the use of One Metric That Matters (OMTM) is so prevalent among startups.
- Make investors happy: Before investors are going to bet on—or continue to invest in—your startup, they need assurances that your startup is viable. And that means you need data that helps your founder make a strong empirical case.
What is a data MVP?
The goal of a product MVP is to quickly get something into the hands of customers to get real world feedback on what customers actually want. Similarly, a data MVP allows you to quickly get a bare-bones data solution up and running so you can discover what users actually need.
This approach makes a lot of sense for startups. If nobody knows exactly what your startup will be doing a year from now, there's no sense in trying to build a comprehensive solution; there's no way to know if all that effort will address users' future needs.
But there’s another benefit of a data MVP: it’s a great strategy for any organization where either users don't have much experience working with data.
It's not that uncommon, for example, for data folks to build a complex data warehouse based on a detailed, thorough analysis of what users say they need, only to discover that once users start actually working with the data, they realize that they have a very different set of needs. Users may also discover that a lot of assumptions they had about their customers are false, so they need to radically change direction.
In short, if there's a lot of uncertainty—either about the direction of your company or about users' needs—a data MVP is a great way of addressing these risks.
How to figure out which data matters
How do you decide what should go in your data MVP? Figure out what analysis you users can and can't easily do right now, then build what you don't have but need today.
One of the best ways to pinpoint those needs is to ask what your company's most pressing pain points and greatest opportunities are.
The other question you need to ask is, what can I build relatively quickly and affordably?
Once you have a beginning list of where you can get the biggest bang for the smallest buck, whittle it down. Drop everything that would be really nice to have but isn't critical. Keep going until you've got a list that's as short as possible but still very useful.
As you are whittling down the list, keep reminding yourself that an MVP isn't about building a full-on data stack, it’s about making a solution that fits where your startup is right now. It's totally understandable if a part of your brain keeps asking, "what about scalability?" Just keep reassuring it that you will get to scalability once you know what your users actually need. And remind it about the scalability paradox; by focusing on scalability too early, you can decrease the odds that your startup will thrive to the point where your solution needs to scale.
Identifying key metrics
Once you have a general sense of what you’re tracking—e.g., which departments, which requirements—it's time to figure out the details. That means identifying the most critical metrics.
If your startup is smart about how they use metrics and is crystal clear about what their most important metrics are right now, you're all set. If not, you might use the MVP as an opportunity to help the team sharpen and clarify what metrics they are tracking and why. One way to help them do that: mock up nonfunctional dashboard prototypes.
Now is also a good time to research which metrics similar businesses have used. There’s no sense in reinventing the wheel—or, if you live in San Francisco, the parking brake—when you can learn from other folks' painful experience.
Just make sure not to copy and paste what others have done. For example, DAU (daily active users) and MAU (monthly active users) are useful metrics for many startups, but that doesn't mean they should inherently be your startup's focus right now.
You can also use your data MVP to help surface conflicts over business logic. If there are differences in opinion about how to measure critical metrics, now is a good time to nail down an answer. Not only will getting everyone on the same page help you right now, it can also pay off down the road. For example, being able to demonstrate consistent metrics over time can be very helpful during M&A...and it’s a lot easier to build them into your reporting now than to unravel a mess three years down the line.
Make a plan for measuring your MVP
Before you roll out your data MVP, there's one more step: get meta. There’s no point in building an MVP if you don't know how you're going to measure what's working for users and what's not.
You don't need to get fancy with your metrics about metrics. All you need is a simple way to get a rough sense of how you're doing. Even a quick and dirty SurveyMonkey poll could give you a good enough sense of whether users are happy with your data MVP and where there may be some rough spots.
Alternatively, if your company has a self-service BI tool, you might want to make sure there’s tracking in place about how it’s currently being used. If your data MVP increases its usage—whether in the number of users or the number of queries they run—that could be a signal of success.
Next steps
Of course, the data doesn’t stop there. Once you've built your data MVP, you can continue to iterate with a deeper dive into your startup's data.
And now for our shameless plug: If most of your data comes from one of the bazillion common data sources that integrate with Panoply, you could jumpstart that data MVP by getting pipelines and storage up and running in minutes...all without a single line of code.
Sound too good to be true? Give it a shot with a free trial—no credit card required.
