Analytics is the key to great business decisions.
But is it possible to rely on data too much?
According to one of the world’s biggest companies, yes. Facebook uses data to drive much of its decision-making, but they’re also known for pushing past immediate metrics for real answers.
How is that possible? Because analytics does a great job at tracking what you’re measuring. But data doesn’t tell you if you’re tracking the right metric. It can’t suggest changes.
And perhaps more importantly, data can’t predict where there are areas for huge opportunity.
That’s why Facebook’s insight into how to use—or not use—analytics is key to any business owner. Truly understanding the pitfalls behind analysis and metrics can help you grow your business.
Today, we’ll look at suggestions from Facebook’s own team on how to use analytics the right way to grow your business.
How to build your data analytics team
Way back in 2010, Facebook’s then head of product Adam Mosseri gave a talk on how Facebook structures its analytics teams.
(Today Mosseri is now head of Instagram, so clearly his strategy worked well for growing Facebook.)
The title of his talk was “Data Informed, Not Data Driven” which gives a general overview of the framework through which Facebook sees data. You can watch the full speech here.
While this was nearly a decade ago, the details are helpful for any startup in the early stages of scaling their product. There’s a lot to learn, and it all starts with how to build a team.
Mosseri starts off by explaining how Facebook sets up teams. Small teams are best, he says, because they’re fast.
Because of that, teams at Facebook are only made of six or seven people. They don’t have a hierarchy of managers, either, since that can slow down and cloud the decision-making process.
Instead, a team makes a decision and the only other person who must approve it is Mark Zuckerberg, the CEO.
Teams are made up of the following people:
- Product designer. Responsible for visual, interaction, and product design.
- Researcher. Conducts qualitative and quantitative research.
- Engineer(s). Typically one to four engineers per team.
- Product manager. Responsible not just for project management, but also ensuring products ship on time and maintaining product quality.
Working together, these small teams are able to find a problem, iterate quickly through solutions, and create an improved product based on analytics.
Here’s how they do it.
Optimize small-yet-critical interactions
When looking at data, Facebook decides areas to focus on based on their potential value to improving the user experience.
Ken Rudin, The head of Facebook’s analytics, said that analytics must be relevant to a real problem and have real-world implications to be worth exploring.
“If you can’t imagine how the answer to a question would lead you to change your business practices,” he was quoted as saying in 2018, “the question isn’t worth asking.”
That means Facebook doesn’t waste time looking at meaningless statistics—they’re focused on what really improves the user experience.
As a simple example, the old photo uploading tool in the early days of Facebook was slow and clunky.
It required multiple steps and only 42% interested in uploading a photo actually did it successfully.
To improve this, the team worked to replace the photo uploader. They started with a hypothesis: people were having trouble uploading photos.
To solve it, the team reduced some of the steps, added better clarifications on other steps, and dramatically improved the usefulness of the tool as measured by successful completions.
According to 2018 data, only 55% of bloggers regularly check their analytics, and much of that data isn’t actionable.
To really start putting your analytics to work like the social media giant, you need to understand what really moves the needle and focus on changing those areas.
Make guesses, then test with data
When it comes to designing a product—especially one like Facebook that has created an entirely new industry—you can’t rely on data for all your decisions.
At a certain point, you have to make guesses and use the data to see if they were correct. Analytics won’t guide you into new and creative areas, which is exactly what you should be doing.
A simple example is the Facebook deactivation page, where users can shut down their account if they’ve decided to leave Facebook.
Lee Byron, a designer at Facebook, designed, built, tested, and shipped a new version that didn’t just ask why a user wanted to leave, but gave them a reason to stay.
This wasn’t shown by data, but Byron had a hunch showing other friends on Facebook would encourage people to stay with the platform. He was right.
Upon testing the experiment, data showed that it reduced deactivations by 7 percent, which at the time meant millions of users stayed on Facebook.
Collect data and use it to inform decisions
When it comes to data collection, Facebook is at the forefront.
Every day they collect massive amounts of data. According to one source in 2017, Facebook stores over 500 terabytes of data daily.
But it’s not just the statuses, comments, images, and videos people upload every day.
Facebook also tracks interactions, clicks, friend requests, and a host of other points they can later mine and explore to improve the product.
Without the data itself, Facebook couldn’t make a lot of decisions they do.
Ken Rudin, analytics chief at Facebook, has been quoted as saying, “Big Data is crucial to the company’s very being.”
Because they collect so many types of data, Facebook is then able to use a variety of metrics to guide product decisions.
This includes quantitative and qualitative data, user interests, competition, and regulation, among other metrics.
And according to a 2017 analysis, Facebook also uses extensive deep learning to explore this data as deeply as possible, providing new types of analytics from the raw collected data.
For example, Facebook has developed an algorithm that’s more accurate than humans at determining if two pictures are of the same person.
They’ve also created a tool that can explore the meaning behind text posts and decide what people truly mean based on the subtle language clues they’ve left.
As a result, Facebook is in a place to use a huge variety of data types to solve nearly any problem they face or product change they want to implement.
Accept mistakes and focus on the big picture
When conducting new experiments and analyzing data, Facebook makes sure they don’t miss the big goals.
It sounds simple enough, but it’s easy to get trapped in smaller changes and lose sight of the big picture. They call these “micro-optimizations.”
As Mosseri says, “a micro-optimization is when one interest over-optimizes for itself at the expense of another, and this is a very difficult thing for us as we scale.”
While all of the teams are trying to optimize for a better feature, sometimes the product as a whole loses out. Keeping an eye on the primary goals trumps data every time.
And finally, there are often failures and experiments that don’t go as planned.
When these come up, there’s a time to ignore the data and focus on the product—but also a time to recognize that a great idea just isn’t worth pursuing.
When that’s the case, learn from those mistakes and keep moving on.
When it comes to data and analytics, what you see isn’t always what you get.
Instead, there are oftentimes hidden factors beneath the raw numbers that tell a deeper story about who’s buying, how important your engagement is, and where you can grow.
Using these strategies from Facebook’s data analysis process, you can get a deeper understanding of the metrics behind your site.
Remember to find areas that might seem insignificant, but lead to disproportionate results. This is the best place for your marketing efforts since you’ll get the most “bang for your buck,” so to speak.
Another way to find areas for growth is to look at customer pain points. What do customers struggle to finish? What areas are difficult for users? How can you improve the experience?
Perhaps the biggest blind spot of analytics, however, is that it can’t predict a feature until you’ve created it.
That’s why the Facebook team creates a hypothesis first, then an experiment, and uses data to help find the result instead of driving the result from analytics.
It’s a proactive instead of a reactive approach.
And finally, focus on the big picture. Data can bring you lots of micro-optimizations on smaller and rather insignificant factors of the business.
But without a clear idea of what the ultimate goals are for your brand, these micro-optimizations can eventually tank your company goals. Look at analytics in the scope of your mission.
Analytics provides a wealth of information to guide our decisions. But at the end of the day, it’s human ideas that keep your brand growing, not data points.
How will you use analytics to grow your brand?