UserTesting Stock: Using AI for Customer Market Research

Last month, we wrote about TaskUs stock (TASK), which offers a way to invest in a little light human slavery digital offshoring services. But the company relies heavily on a Philippines-based workforce to power its software-as-a–service (SaaS) model. There are references to artificial intelligence and automation technologies, but cheap or crowdsourced labor isn’t exactly what we want to invest in. We covered TaskUS at the request of one of our sharp-eyed readers who also helps support our team of very real human MBAs. Another reader recently tipped us off about UserTesting stock (USER). It’s another SaaS tech company that also compresses two words together to form its name. But that’s where the similarity ends.

About UserTesting Stock

Founded way back in 2007, San Francisco-based UserTesting had raised more than $150 million in funding before it had an IPO last November and added another $140 million. Just like it says on the tin: UserTesting conducts user testing on anything under the Silicon Valley sun, using its Human Insight Platform. The 30,000-foot view is that the company uses video to record customers interacting with a product, service, brand, app, or whatever. It uses artificial intelligence (AI) to not only help capture and process the video but also analyze it at scale. Some actual human analytic brains are also in the mix.

Back to the basics about UserTesting stock: It hasn’t escaped the whirlpool that has sucked value out of public companies this year, especially in the tech sector. Its market cap has plummeted by nearly 60%, dropping below $1 billion, as of early June. Normally, we avoid investing in any company below that threshold, but UserTesting has some interesting things going on.

First, the company is significantly increasing revenues. In 2021, it had revenue of $147.4 million, up 44% from the year before at $102.2 million. Second, it derives more than 90% of its revenue from yearly subscriptions, billing for its non-cancelable services in advance. That’s very much unlike TaskUs. Third, market research is a big market itself, with Statista claiming a total addressable market of more than $75 billion. Bonus: UserTesting went public through a traditional IPO rather than yet another dodgy merger with a special purpose acquisition company (SPAC).

Before we dive further into the financials, let’s try to understand a bit more about what UserTesting does and how it does what it does using AI. 

What Does UserTesting Do?

Simply put, UserTesting is digitizing customer market research. Its platform captures “highly contextualized customer perspectives from targeted audiences who have opted in to share their thoughts” – presumably for some sort of compensation – about digital or real-world experiences or some combination thereof. This video-based technology enables UserTesting users to record and analyze “human signals” from these Customer Experience Narratives (CxNs) outside of the verbal feedback, including things like intonation and tone of voice, facial expressions, body language, visuals, and behaviors. The company has qualified a network of 600,000 contributors and can usually deliver insights within a day.

How Does UserTesting Use AI?

That sort of on-demand customer market research suggests a pretty high degree of automation is baked into the platform. Indeed, UserTesting claims that its extensive video archive – augmented with audience profiles, questions, and task data – has enabled it to cook up several proprietary machine-learning models for audience targeting and distribution, as well as for analyzing content and producing insights. 

Here are a few ways the platform uses AI:

• Natural Language Processing analyzes transcribed audio to identify positive and negative sentiment, identify reactions such as confusion, as well as to find questions and their answers in a CxN.
• Audio-based models rate the quality of a CxN using volume, tone, inflection, and other factors. 
• Demographic- and behavior-based models keep contributors optimally engaged.
• Other models can identify intent in a CxN and even “recognize indicators of engagement within a CxN, including joy and excitement as well as frustration and disengagement.”

Seems like the last AI model could be useful for building a video dating app, if this whole customer market research thing doesn’t work out. And, like any good AI algorithm, the system gets better over time. For instance, the platform uses voice data to identify a contributor’s mood changes during a CxN, and then uses that information to train a “clickstream algorithm” to recognize how these mood changes are reflected in cursor movements and clicks. The insights help improve the platform’s ability to recognize behavior and adjust in order to engage better with contributors. 

UserTesting Use Cases

UserTesting claims it has more than 2,300 customers, including more than half of the world’s top 100 most valuable brands, according to Forbes. These organizations use the platform to do things like improve their customers’ browsing and purchasing experiences; vet market opportunities and understand product-market fit; sanity check marketing campaigns; test new designs and prototypes; and just generally understand how to help customers reach a new state of buying bliss. 

So, how does this translate to return on investment (ROI) for a typical customer? Take the example of BRP (DOOO), which manufactures a line of well-known weekend boy toys like Ski-Doo watercraft and Can-Am ATVs. BRP had hired Adobe to completely redesign and build a new digital framework (ie, website, app, email) for customers to interact with its brands, starting with Can-Am. UserTesting provided customer market research in near real-time on what worked and didn’t using a network of contributors consisting of riders, enthusiasts, and prospective Can-Am buyers. The result? A 30% increase in conversion rates across the products portfolio.

How Does UserTesting Make Money?

As we noted earlier, UserTesting makes nearly all of its money from customers like Adobe/BRP on annually recurring revenue (ARR) subscriptions for its platform services, which means the company has a very consistent way to measure and predict its ongoing revenue stream. For example, more than 300 customers account for at least $100,000 ARR each – up from 190 at the end of 2020. In addition, net-dollar retention rate – the percentage of subscription revenues it retains from existing customers – has averaged 116%. That means current customers are renewing and expanding their contracts every year. It also added some new customers last year, including British Airways, Canada Post, DocuSign, HP, and Volvo, among others.

Meanwhile, less than 10% of revenue is based on professional services, such as research studies, training services, and strategy workshops. 

Should You Buy UserTesting Stock?

So far, there’s a lot to like about UserTesting, beginning with its ARR business model. The reliability is evident in the company’s 2022 guidance of revenue between $197 billion and $201 billion. That’s a pretty tight projection. In Q1-2022, the company continued its current trajectory with total revenue of nearly $46 million, an increase of 47% compared to last year. Subscription revenue accounted for nearly 95% of all revenue at $43.2 million.

We also like that UserTesting boasted a gross profit of about 75% last year. We’ve been using gross profit as a way to measure tech companies in the current bear market, rather than just revenue growth. It simply means that UserTesting only spends about 25% of its revenue on expenses directly related to the products and services they sell. That suggests that the software is doing a lot of the heavy lifting, as you would expect from a company based on a SaaS enterprise model. In terms of its subscription business, there are three main expenses: paying its contributor network, operating its platform, and paying its employees who directly support the business. 

Now, that doesn’t mean UserTesting is profitable. In fact, the company lost more than $50 million last year. It does tell us that the core business is sustainable, but other costs like sales and marketing have nearly tripled while R&D has doubled. Over time, a mature, well-run company will be able to trim the fat (administrative costs also doubled) and turn a profit. In the meantime, grow and grow fast.

Another plus: No single customer accounted for more than 10% of total revenue over at least the last couple of years. One downside: More than 80% of revenue is concentrated in the United States, though the company is trying to expand overseas, with outposts in Scotland and Singapore. 

Conclusion

We like UserTesting, but it’s almost too small to find a place in our portfolio and the value proposition seems to lack durability. Does the company create enough economic value for their customers that the offering will be resilient during times of economic turmoil? Like many tech companies, UserTesting stock is trading pretty cheaply based on a simple valuation ratio (current market cap/annualized revenue based on the most recent quarter) of less than 5. Anything over 40 is generally too rich for our blood. 

But if we are interested in this theme of digitized customer market research – and, frankly, that in itself is up for debate – is UserTesting the best in show? Among its competitors is a big data startup called Qualtrics (XM) that we covered years ago when they were still a startup. Today, the publicly traded firm has a market cap of about $8.5 billion on $1 billion in revenue. It seems to play in a similar sandbox as UserTesting, but with a platform that extends beyond customer experiences to employees and more. But as our last date will tell you, bigger doesn’t always mean better. We’ll follow up with a deep dive on Qualtrics in the near future.