One of the greatest challenges blocking your team can be traced back to your data. But it’s more insidious than not having the right tools, or not hiring the right people to ask innovative questions. It’s whether or not you can even trust your data in the first place.
At Mammoth Growth we’ve worked with almost 900 companies on projects related to growth marketing, product analytics, and data infrastructure. And 93% of the CEOs we’ve spoken with said their teams did not have the data required to make informed decisions. But once we dug a little bit deeper, we discovered that in the vast majority of cases, they already had the data. They just didn’t trust it.
So let’s talk about the elephant (or mammoth) in the room: your data is everything because it dictates how you grow your business. And that means whenever we work with a client, we have to go beyond people and technology to include the benefits of data trust in the conversation. In this article we share how high data trust impacts what companies can achieve, and how it can lead to better decisions.
Why Should You Care About This? Why Invest Time and Energy in Building Data Trust?
One of the simplest answers to these questions is the drop in efficiency that stems from a lack of data trust. As an example of this, consider what happened with Barre3. They had big plans to expand their network of fitness studios, but they had little to no visibility into customer acquisition. In addition to their brick and mortar studios, they also had an e-commerce shop and a subscription site, so deciphering their customer data would have been a monumental task even in the best of circumstances.
However, what made their challenges particularly onerous was the lack of data trust. They had implemented Mixpanel years ago for behavioral analytics, but somewhere along the way their team began to regard their legacy data with skepticism and suspicion. It got to the point where team members would rather rely on manual spreadsheet reporting than Mixpanel’s analytics dashboard.
Experience had shown them they couldn’t trust what Mixpanel was telling them. To regain some semblance of faith in their numbers, they resorted to doing everything by hand. A lack of trust in their data not only slowed them down, it also meant they were paying for a tool they rarely used. And all of this was directly at odds with their expansion plans.
This type of situation repeats itself over and over at companies all around the world. If employees don’t trust the data they deal with every day, they’ll find workarounds that turn into the new status quo.
Not trusting the data is worse than useless, it is a source of risk. That’s because making the wrong decisions based on inaccurate data can lead to losing customers and investing resources on the wrong things.
Now, consider this simple example of how mistrust can creep into your data with little warning. Imagine your team has created an app which performs simple analyses on digital pictures through machine learning and AI. In order to work, the app must be connected to WiFi so users can easily upload their pictures.
One of your product managers wants to check the app’s onboarding flow. They create a simple report to see how many actions were attempted while WiFi was on, and how many were attempted while WiFi was off. The results are shown below:
With so many people attempting to onboard while WiFi is off, clearly there must be a problem with the app. Your team would need to drop everything and do a sprint to fix this issue. Right?
Not necessarily. By changing the data type, the product manager discovers an entirely different issue:
Data taken at face value is only trustworthy if it’s accurate. If you don’t have the right data, you might spend all your time on the wrong problems. Or, you might not even know you have a problem at all.
So, let’s outline some signs of low data trust, and discuss some steps you can take to rebuild it. Trusting your data is like knowing how to speak the local language when you arrive in a new country for the first time. Without it, life is far more difficult than it needs to be.
“This Can’t Be Right!” and Other Signs of Low Data Trust
Lack of timely insights - Do you need to open a ticket to the BI team and wait a week to get an answer? Or maybe you just released a new feature and you have no clue how it’s behaving? We see this bottleneck over and over with companies that rely on SQL to answer all their product and behavioral analytics questions. Granted, SQL is incredibly flexible. But it’s also very slow and expensive, because the talent pool is so limited. In this case, your team might have extremely accurate data. The lack of trust comes with the “black box effect” of relying on a very small number of people to interpret or manage the data for you. No transparency and long delays are early warning signs of low data trust.
Not trusting the data you have - In this situation, different tools give you different results based on the same data set. For example, we often see this when companies compare their advertising analytics from Google Analytics 4 with tools like Mixpanel and Amplitude. Take care when comparing results from different solutions, since every comparable platform handles data in a slightly different way. The lack of data trust sneaks in when we insist that one tool’s output is more reliable than conflicting results from elsewhere, without questioning why. It might not be the data at all, but how you treat it.
Data is siloed and inaccessible to the rest of the company - Ever hear the terms “marketing data,” “sales data,” or “product data?” Every team has different KPIs, but ideally everyone should share the same North Star metric. If different teams don’t have an easy way to understand what the rest of the company is dealing with, is it enough to tell them, “just take our word for it?” Siloed data can also arise when different departments define the same term in different ways. How do your different teams define the term, “active user?” Are all of these definitions consistent?
Bad data formatting - Finally, poor data trust can be traced back to something as simple as poor formatting. Using the wrong units for currency or any other KPI can create lasting confusion, or worse, the failure of your mission. And ‘null’ or missing values can impact even the simplest query, as we saw in our WiFi example above. And if you want a real-life example of this, consider the slow roll-out of Apple’s new SKAdNetwork 4.0. This was partly due to a bug which caused conversion value updates to return ‘NULL’ instead. Ad networks realized they couldn’t trust their data, so they quickly reverted to SKAN 3.0.
How To Rebuild Trust In Your Data
Almost every company has experienced one or more signs of low data trust. Here’s what to do about it.
Usability - Make sure different teams are pulling from the same data source. A Customer Data Platform (CDP) can make this much easier to achieve, as long as your data are organized in the proper format. And if you’re sending data from a data warehouse to your CDP, be careful that this transformation doesn’t distort the data along the way. Adding a new tool to your tech stack won’t do anything to improve data trust on its own.
Granularity / hierarchy - When defining events for product and behavioral analytics, there is a sweet spot between too specific and too generic. If your events/data models are too specific, then finding the data you need is like searching for a needle in a haystack. It's much easier to filter / segment data than it is to combine multiple events and data models. On the other hand, if your event data are too generic, it becomes difficult to get meaningful insights. If you can find the balance between too specific and too generic, that also means you and your team won’t need to manually update custom events each time you add another page / button.
How intuitive is your data? When categorizing and naming your data, try to be as clear and consistent as possible. Ideally, anyone new to your organization should be able to understand all your data labels based solely on context and your data naming conventions. While this might seem a bit extreme, it happens all the time when experienced employees leave and new employees take their place. Any steps you can take to avoid confusion will save your team tons of time down the road. More specifically, make sure that KPIs are clearly defined so that anyone in the organization can understand their place in your data hierarchy.
Optimize data collection channels - Create clear documentation on how to collect data from different sources. Set up a testing mechanism so you don’t ingest data with errors or data that’s not formatted properly. You can also check that the data set includes all of the information needed to serve its purpose in your analyses downstream. Finally, just like we’ve mentioned above, it’s vital that the same data values from different sources are the same. A customer data platform (CDP) can be very helpful here.
The Benefits of Data Trust
Earlier, we compared trusting your data to knowing the local language when you encounter a new culture for the first time. If everyone in your organization has high trust in your data, the benefits are akin to a universal translator from science fiction. Anytime you attempt to analyze completely new data, there’s zero confusion and zero misunderstanding. You know exactly how that data fits into your larger schema, and what role it plays in guiding your team to success. The most fundamental benefits of high data trust are clarity, accuracy, and speed. And every company is capable of getting there. Talk to us to learn more about restoring data trust in your organization.