In both B2C and B2B markets, machine learning and artificial intelligence have established a strong foothold in the Sales and Marketing technology stacks, helping businesses to forecast, measure, and optimize outcomes in these domains. Tools like Salesforce and HubSpot (now virtually household names) surface actionable insights that help leaders make decisions– like who is likely to close versus abandon a sale, and what content is most popular and generating the most leads. Strangely, the Product department has been widely ignored when it comes to measuring and optimizing product success by leveraging machine learning and AI.
Product data historically has not had an analogous tool, even in the most sophisticated product analytics suites. Instead, enterprises (that have the scale and budget) grow their own data science departments to mine insights that can improve product decisions to drive the business forward.
So where does that leave the majority of product-led companies that want to make data-based decisions to drive growth? In a time when everyone is focused on building the right thing (rapidly) with shrinking resources and smaller teams, it’s all the more critical for product owners to ensure they’re making the right bets shaping the product roadmap.
Meet AI-enabled product-led growth (PLG)
This, in essence, is why Fuzy was built. It brings the power of machine learning and AI to the product department––literally delivering Product Science as a Service (PSaaS) to empower product leaders to turn their product analytics into valuable insights that connect the dots between product data and core business outcomes.
Fuzy’s VP of Engineering Triet Le explains, “It comes down to measurement and accountability––being able to answer questions like ‘What impact did releasing this feature have on our KPIs?’ or ‘Justify why we should spend $300k of engineering resources to do this.’ There are some exceptionally good product owners out there with a lot of experience and great intuition; PSaaS enables them to enhance their domain knowledge, expertise, and qualitative processes with rigorous quantitative analysis.”
5 Reasons why product analytics and Fuzy are best friends
For PLG companies that have a maturing tech stack but not a robust product science capacity, Fuzy could be your product owner’s new best friend. Far from replacing product management or the existing tools you love, Fuzy is an indispensable sidekick that brings rich context to make data more useful and your team more efficient and strategic. These are five key reasons why:
Getting to actionable, impactful insights on your own is extremely difficult.
When it comes to best-in-class product science departments, Indeed stands out. First, they have a culture that revolves around data. Their CEO mandated that everything they built had to be metric-driven and data-backed. Their deep pockets and desire to hire extremely experienced product leaders saw them bring on 50 product scientists and teams of data engineers to create their capacity for analyzing data. Clearly, it takes a great deal of money, hours, experience, and the right culture to accomplish what Indeed did.
To illustrate why this might not be feasible for some companies, Triet Le paints this picture: “Once you have the ability to collect data, at some point a product leader will say, ‘We need to hit metrics XYZ,’ then a PM will start working on how. They will go to Mixpanel or Amplitude to look at the data (which inevitably won’t make a lot of sense). They won’t know what to do with it. Companies that can afford it might hire a data scientist to work with the data and deliver actionable insights. It’s expensive, but it’s a viable option. Next, the data scientist will need to export the data into a warehouse of some kind, and for that you need a data administrator. Then you have to deal with the data format, which often requires a data engineer to clean it up. All of these steps require significant costs and an extended timeline in order to get to the original goal. And that’s too slow for the pace of most technology companies.”
Fuzy shortcuts all of the above and builds that resource stack for you. Furthermore, the PSaaS solution provides deep reduction in cognitive load because you don't need to come up with every hypothesis to test; the hypotheses are served to you. This saves you from searching for needles in a haystack in your continuous quest for understanding what matters and what to build next. Fuzy sifts through the hay and points out the needles for you.
With Fuzy, you’re able to quickly correlate a target with an insight. The built-in AI links relevant user behavior (what people do in the app) to a target outcome that matters to the business. Good intuition might allow a PM to achieve some of this, but machine learning can do it better, quicker, and with perfect discipline, which leads to a more consistent result.
You couldn’t ask a PM to hold all your product and business data streams in their mind, but Fuzy does this by design, even mapping important pathways and patterns that were once invisible. Fuzy can take hundreds or thousands of rows of data and produce a visual map showing how users flow through your app, and even how those flows have changed. This reduces the data analysis burden for your PM and allows them to process it in an intuitive, visual way––helping your team get to the right answers faster.
Fuzy sits on top of existing analytics.
Fuzy doesn’t require retooling or instrumentation of your existing tech stack. It’s likely you already have and track gobs of data; sync that data and Fuzy takes care of the rest. Head of Engineering Triet Le shares the thinking behind Fuzy’s value proposition: “There is a wealth of data out there; there are mature, proven product analytics platforms. One of the complaints we hear most is, ‘Collecting data is not the problem––we have too much of it, and we don’t know where to look.’ Fuzy sits seamlessly on top of these existing platforms and basically provides a signal detection layer. Our goal is to take that wealth of data and discover what’s really important and what’s not, find hidden and unexpected relationships between the data, detect when patterns fundamentally change, and help explain what changed and why.”
In this way, Fuzy allows PMs to do their real jobs: managing products while Fuzy manages data. Our PSaaS approach distills multiple data streams using machine learning and other techniques that bring invisible insights to the surface while eliminating communication silos and guesswork.
Fuzy delivers on the ‘last mile’ of measuring product performance.
Unlike product analytics, which are only able to report on outcomes contained within the product, Fuzy’s PSaaS connects the dots from customer attributes to in-app behavior to core business outcomes. That makes the difference between reporting on how many features shipped to understanding which feature resulted in increased upgrades (for example) and why.
Triet details how: “We maintain two data sets from the customer. The first is in-product usage data (what a user sees and does in the app, including clicks, page visits, etc.). The second is business data, like that contained in Salesforce, Stripe and HubSpot. This allows us to understand additional metrics (like when an account churns, when a user upgrades, and other transactional data). This allows us to link leading product actions to lagging business metrics. We also use demographic data to create additional connections and surface patterns that provide more context. This helps product leaders make better decisions. Context is what makes the result of analytics useful. For example, with Fuzy, you could predict when an account will downgrade or convert based on data patterns. You might observe that if the user does these five things in a certain sequence, they’re more likely to convert to a paid user.”
Triet goes on to say, “Typical SaaS products are continually evolving. New features are being released; enhancements are made. So at the very base level, we have systems running to look for fundamental behavior changes. If you are doing any kind of data science with machine learning or even basic metric projections on top of your customer/behavioral data, any pattern change can invalidate your models and projections. So detecting when this happens and being able to explain what happened is extremely important if you want everything else to be accurate, especially how they impact business outcomes.”
Fuzy protects against bias in data and decisions.
While many companies strive to be data-driven, this is actually hard to achieve in reality. Having access to data doesn’t mean teams are using it to make product decisions. Due to velocity pressure and management structure, organizations can fall victim to top-down or “squeaky wheel” decision-making. And while PMs often do have excellent intuition, Fuzy brings objective third party data to help bolster the PM by enhancing the decision-making process with evidence.
As Triet puts it, “Fuzy helps PMs validate and/or focus their thinking so they don’t have to boil the ocean. It can also surface invisible and unexpected insights. We give you a foundation of data, which means you’re less likely to have biased thinking. Without a data-based foundation, you start the decision-making process using gut instincts, creating bias.”
Additionally, data science for product analysis requires cycles of hypothesis testing. This experimentation has the potential to overlook important signals in favor of what the team is biased to look for due to historical reasons or personal philosophies. With Fuzy, there’s no cherry-picking of data. Instead, our algorithms find the patterns, then serve them up for the product manager to consider and act on if they are useful. Fuzy also helps you find important anomalies that you might otherwise overlook. The system can focus PM attention on an important event or window versus manually hunting through dashboards. Thus, the process of figuring out ‘what happened, when?’ becomes more automated (and less tedious), and ultimately less error-prone.
Finally, Fuzy inherently increases your data literacy quotient by offloading much of the manual testing and calculation that traditional product analysis requires. With Fuzy, you don’t have to be a data scientist; the system performs data analysis and interpretation without the danger of human error.
Fuzy is budget-friendly compared to homegrown data science departments.
For the variety of reasons discussed in this article, spinning up a data science center in your company is very expensive––and cost prohibitive for most growing organizations.
Take a look at the total overhead of growing your own product analysis department. Implementing Fuzy’s PSaaS not only saves cost, but significantly shortens your time to value and boasts far more operational efficiency than building and maintaining a dedicated product science department.
Right now, Fuzy is embarking on an interview series with experienced product leaders and data scientists who have looked at product analytics from every angle. Many have been a part of large tech firms with in-house teams who have created custom tooling and processes; others have grown businesses from the ground up and share lessons learned from making rapid decisions with limited information. Across this spectrum, there is a ton to learn about what makes sense for each stage of product development and company growth. Follow the series to learn what considerations are critical at each stage for building product science capabilities, and walk away with wisdom from experts who are practiced in combining evidence and intuition to make the right bets in product development.