It has been a wild year for Emerging Companies Portfolio stock Appen, with the artificial intelligence tech enabler seeing its share price plummet below $20 when Covid hit in late February 2020.
Previously the Appen stock rallied hard with many tech peers over the next few months, clearing $40 in August 2020, on the expectations of a strong result and upgraded guidance from management as they had historically done. However, this didn’t occur, and management were muted on the outlook for the business due to Covid impacts which have continued since resulting in the share price falling over 60% from its highs.
Where did it go wrong for Appen?
To answer that it is worth taking a step back and looking at the business model. Despite being lumped in the basket of “tech stocks” on the ASX, Appen is not a true tech stock compared to peers such as Altium, TechnologyOne or Xero who develop a suite of software solutions for customers and exhibit the characteristics you would expect from that business model, primarily scalability with high incremental margins on new customers.
Alternatively, Appen’s core business is centred around the “Crowd” which is their term for the over 1m people they contract to provide human annotated data sets. As an example, a search engine provider may engage Appen to process millions of search terms and apply each term to the most applicable result. While a human would initially perform this process, over time as data is gathered the search engine provider can feed those results into their artificial intelligence platform which will improve over time and “learn” on its own. It is clear from this business model that as the business grows, the costs of the Crowd must grow with it and Appen can’t achieve the same scalability as pure software peers.
While Appen has historically been a well-run business and providing the “picks and shovels” to the artificial intelligence boom has been very profitable, the flaw in their business model is they are extremely reliant on project-based revenue from large customers. In the latest annual report, Appen disclosed they receive 89% of their revenue from five customers, and while they are not disclosed it is safe to assume, they are the major US tech players who are leading the race in AI (the likes of Google, Facebook, Microsoft, Apple, Amazon, etc.).
When Covid first hit, most businesses deferred discretionary costs until they got more certainty on the post-Covid outlook for their business and economy in general. This was the message Appen management provided the market in August 2020 at their half year results, but they were generally upbeat about the future believing the projects that had been deferred would emerge in the second half of the year.
However, in December last year Appen management shocked the market with a trading update that suggested the impacts of Covid had continued through the second half and the business was on track to substantially miss the financial guidance it had previously provided. This seemingly went against the narrative and data that the US tech giants had maintained their discretionary spending as their businesses were far less impacted by Covid than initially feared. I have compiled the quarterly R&D spend of each business from last year, which shows a steady increase over the year:
While R&D spend of these customers is not a perfect corollary for Appen (as it depends on where they allocate their R&D budgets) it is a useful data point to track. At the recent full year results, management were again muted on the outlook for the business and expect the same headwinds to persist until the second half of 2021. This has now caused the market to seriously ask the question of whether Appen is a business impacted by transitory Covid effects, or is there some fundamental change to their industry that management are missing? Perhaps customers are allocating projects to competitors to maintain balance in the industry, or maybe even internalising some of Appen’s business themselves. Ironically, Appen may end up competing against the very artificial intelligence platforms they serve as they improve over time.
Certainly, Appen’s margins paint the picture of a business experiencing competitive pressures and poke a hole in the tech stock narrative (remember the characteristics of scale and improving incremental margins):
Unfortunately, Appen has been a holding in the Emerging Companies portfolio throughout this turbulent year.
Looking to what lessons we can take from the last six months, the following stand out to me:
Downgrades generally come in groups. Management is almost always either too optimistic on the future of their business or too confident in their ability to turn it around which is why the first downgrade is rarely the last.
Identifying if a cyclical weakness is becoming a structural one is paramount. When Appen first missed market expectations in August 2020 the share price fell in line with the lower expected earnings, but the business continued to trade on a relatively high multiple (35x). However, as the market now believes the issues to be structural rather than Covid weakness, Appen’s multiple has compressed significantly (20x).
Given we can’t change the past we can only look to the future now to dictate our holding in Appen. With the business priced much more reasonably compared to it’s business model (tech adjacent rather than pure tech) and growth rate at ~20x earnings we continue to hold our position.
Want to learn more about Oracle Emerging companies portfolio, click here