I ultimately held this too long, but the decision to exit was a wise one. There's a great saying which is that "sentimentality about an asset leads to a lack of discipline." I suffered from commitment bias and also suffered on Appen from straying outside of my circle of competence. It was a massive lesson for me. 

Macquarie has been dead right on this, unlike Citi. Unless Appen can dramatically pivot, their involvement in this industry (human led AI data labelling) is a dead end. This I why I think a safe entry would need to be on a single-digit PE and even then, the industry is moving so fast that such an investment would hold risks. 

To be honest, I think Appen are doing their best in a difficult situation (although they acted way too late to start shifting their business and thought that US tech would be a goldmine indefinitely). The growth from the Chinese market (whilst not secure/sticky long term) is showing a very impressive trajectory. But, it is likely to eventually end up following a similar fate to the US tech giants. 

It is still a hot industry right now, with US tech valuations of similar firms through the roof. A friend of mine was involved in launching a data labelling firm recently, called Encord (https://encord.com/). They raised a $12.5M USD Series A round last year. However, they are more like Scale AI than Appen: replacing manual processes (that use crowdsourced workforces) that make AI development expensive, time-consuming, and difficult to scale. 

Had Appen's management been ahead of its time, it should have acquired a firm like Scale AI in 2018 or 2019 before it was too late. In 2019, Appen was valued at over $2b AUD and in August 2019, Scale AI raised a series C financing round that put the company’s valuation at $1 billion USD. Today, Scale AI is worth north of $7b USD. Appen's situation could have been avoided. Sadly, this is a story of the rise and fall of a former market darling.

I've recently made the decision to cut my losses on APX, which was down 40%.

In a Kenny Rogers song, 'The Gambler', the American singer-songwriter highlights his encounter with a ‘gambler’ on a ‘train bound to nowhere’. If understood as counsel about life, the general advice of the song isn’t straightforward. However, it does highlight an important truth – that some of life’s most important decisions come down to using accumulated experiences and knowledge to make a judgement call based on probabilities.

Analogies with poker aside, Kenny is right. Indeed, “knowing” what to do and when with long-held positions is one of the most challenging and important aspects of successful investing. In the below note, I expand on my own investment framework for selling and put it into context with ASX:APX. It was a difficult decision to make, but on the balance of probabilities, I think it was the right decision to make, for me.

https://www.tepinvestments.com/blog/knowing-when-to-hold-em-and-knowing-when-to-fold-em-asxapx-amp-asxqpm

Wishing all $APX holders the best going forward. I'm not certain that Appen is "down and out" — rarely are positions that black and white in the markets. Yet, I’d prefer to be on the sidelines right now and put my capital to work in better opportunities. It is difficult to swim against the current (sentiment) and I have the optionality of returning if I think the situation is changing. In my eyes, that is prudent investing.

February 2022 (H2 CY21 results release) will provide us with some further insights into the situation. But, for the moment, there are too many red flags to stay invested.

Mondrian acquires an interest in Appen

The London-based international investment firm, Mondrian Investment Partners, purchased a sizeable stake in Appen last week.

The firm bought a 5.1% interest or 6,272,348 ordinary shares in the company. This puts Mondrian as Appen’s fifth- largest stakeholder, behind notable companies such as HSBC, JP Morgan, Citicorp and C & J Vonwiller (ex-Chair and Founder).

The date of the share acquisition (on-market) commenced back on 20 October and finalised on 19 November.

Mondrian manages over US$60 billion in both diversified equity and fixed income asset classes.

Appen completed its restructure in May, focusing on its core business interests.

While its half-year results were underwhelming, the company expects an improved second half. This is supported by a strong order book, higher confidence in its pipeline, and the expected second-half revenue skew. The latter is due to its customers’ delivery schedule for e-commerce, digital advertising, and search programs.

I'm curious as to how Facebook's developments of a metaverse may impact Appen, we already know that Facebook is Appen's number 1 customer by revenue...

Understandably, many here are frustrated with management due to the share price decline and the manner with which they have allowed it to happen. But, my view is that Appen's management team have acted in the same neutral, low-key, understated manner all the way since 2015. This is a continuation of what we have come to expect from MB. That style works tremendously on the way up when things are going well (50c to $40) but horribly on the way down when things are not going well ($40 down to $8).

I'm continuing to hold because the original investment thesis of mine has not yet had the chance to play out over a full cycle (initiated by COVID where US big tech reprioritised work). Do structural issues concern me? Yes. What concerns me is Facebook's developments in SSL; this absolutely could be a structural issue for Appen. This is why the share price was at $8. But, what reassures me is the number of new projects for Appen (albeit small in value) that are springing up within the very same big tech companies. The number of new AI use cases is staggering. What also reassures me is the hundreds of millions of dollars of smart money being poured into industry players like Scale AI from some of the best investors in the world. Sure, they are a competitor, but my point is that Scale AI plays in the same arena as Appen. If the data annotation industry was going to 0 (i.e. be displaced) anytime soon, I don't think we would be seeing investment into new data labelling players from firms such as Dragoneer and Greenoaks Capital (both San Francisco-based investment firms), and Tiger Global, the prominent technology-focused investment company based in New York.

Most of Appen's revenue comes from relevance projects for Facebook and Microsoft. Revenue for these projects was down 9% YoY in H1CY2021 v H1CY2020. Appen claims that this is because H1CY2020 had no COVID impact (true at the time for big tech) and H1CY2021 experienced a major pivot of big tech away from ad-related revenue into new AI projects (which initially are less data-intensive). If revenue for content relevance with Facebook and Microsoft does not return by the end of CY2022, then we have a more evidence-based data point to suggest Facebook are removing their needs for data annotation in the relevance space (rather than re-prioritised temporarily). If this is the case, the management team of Appen need to face the public and explain why their explanation at the time deviated from the truth and whether they misled investors. This would be a serious event. But, we don't have to worry about that yet because that is purely a hypothetical at this point in time and Appen maintain that this revenue will return gradually. Management has been truthful along the journey since 50c and I trust that they are doing their best currently to maintain that reputation.

For the moment, Appen's guidance (for flat/increasing revenue) suggests that although there are some amazing breakthroughs taking place in SSL right now, even at Facebook and Microsoft, Appen's services are still needed. What gives me the confidence to continue to hold my major position here and not take the loss is that Appen remains working hand in hand with Microsoft and Facebook to this day. China/Govt/Enterprise are also important and are progressing well, but we want the core of this apple to not go rotten. Encouragingly, Appen has more projects live right now than at any point in their history (many of them are low in value but can grow in time). Appen is also doing increasing amounts of work with major auto companies. This is a competitive space, but great to have a foothold here. Time will be the true test. Solid growth and/or upgrades in CY2022 and the share price has a clear path back to $20+, where I will be in profit on my APX holding. There is of course the possibility of further downside if Appen's business deteriorates, but at this stage, I like where the company is at (making acquisitions, investing heavily, acquiring new talent, and with cash in the tank and no debt). I'm happy to wait another few years here to see how SSL evolves, it is definitely a threat, but I don't think it will displace Appen (or Scale AI/Samasource/Labelbox/Hive/Cloudfactory/DataLoop) any time soon. One of my favourite quotes is that "It is often never as good as it seems, but also not as bad as it seems. The truth lies somewhere in the middle." I think that is applicable to Appen today. The greed to fear to greed cycle continues, but I am staying the course...

P.S. I will update my sentiment to buy if Appen can meet their guidance for full-year CY21 and guide higher again in CY22 by at least 20%. For now, it remains a hold for me.

Firstly, H1 was always going to be a challenging period on a comparison basis because if we remember back to CY2020, Appen wasn't actually impacted at all by COVID in the first half of CY2020 (this is part of the reason why the share price raced to $40+ mid last year). It was only in the second half of CY2020 that Appen's customers' (US big tech) priorities starting changing (a shift away from traditional advertising-based revenue due to diversification efforts and legislative changes). Appen had already made it clear that CY2021 would have an earnings skew to the second half and so we shouldn't have been expecting much for H1.

Despite having a COVID impacted period (H1 CY2021) contrasted with a non-COVID impacted period (H1 CY2020), Appen managed to keep revenue flat (up 2% versus the pcp), that's not a bad effort, but could have been better. Profit (EBITDA and NPAT) however was significantly down because Appen is now deeply investing in new products, including the Quadrant acquisition. Of course, on the surface, the result looks horrible with margin compression and profits falling. Either Appen is in a structural decline (due to competitive threat and insourcing/SSL at US big tech) in which case margins will continue to be under pressure and the share price has much further to fall or this is a short term COVID induced blip and a transition point in the lifecycle of the company. To be fair, I think the truth lies somewhere in between but closer to the latter.

I think Appen does have a viable future (at least across the next 10 years) and importantly what I liked about the result is Appen is signalling to the market that it is up for a fight against players like Scale AI. Appen at this stage remains a "star business" (refer to the BCG growth-share matrix); it is in a fast-growing industry and is the leader in the industry. The acquisition, new executive appointments, increasing spend on employees to attract top talent, and increased spending on R&D/product (10% of total revenue) is Appen's way of saying, "we're in this for the long haul, we are the market leader and we aren't going down without a fight". Part of their competitive advantage is that they are a one-stop shop -- almost every data modality is covered by them (including geospatial via Quadrant). They have a size, scale, and breadth advantage compared to competitors and are increasing spend to stay the course as the leader. The industry is evolving fast and Appen needs to keep investing to stay current.

One of the risks from here is that Appen has once again signaled a skew; this time towards Q4. This again puts Appen on a tightrope. There is a chance that Appen will miss guidance again come the H2 results in Feb 2022, although Mark appeared more confident than previously on the investor conference call citing an increasing order book with a stronger proportion of late-stage deals. If Appen this time deliver on what they have guided, half 2 is going to be massive to make it to US$80m EBITDA (A$110m EBITDA) for the CY21 year, given H1 only reported underlying EBITDA of US$27.7m. It is also going to look really good if they achieve that because the prior corresponding period (H2 CY20) was a shocker. $110m AUD of EBITDA, if achieved, with guidance of c.20% growth in CY2022, probably puts Appen on a 15x CY22 EBITDA multiple for a market cap of $1,980 for a share price of $15.84 (SOI of 125M). If growth in CY22 is again achieved and again signaled for CY23, then we have a path back to $20-$30. So it is a long time frame that we are now looking at for those that hold with higher AEPs.

With every investment and every month that passes, I'm trying to become a better investor (and person) and Appen has certainly been part of that journey: I've made plenty of mistakes here. One of my key learnings from this experience has been to use stronger technical analysis filters alongside my investment thesis. For example, going forward I will be implementing a hard and fast rule to not buy a company in a technical share price downtrend, no matter how oversold or cheap one believes the valuation to be. I've caught a falling knife here, with an average entry point below $20 but still a position where I am significantly underwater. So far, I've taken 1 step back for every 3 steps forward with my investing journey... I'm ok with that, I think it is normal. Overall, investing has been a thoroughly rewarding journey for me and I expect that to continue; for those who keep learning and keep moving forward, this journey will work out well in the fullness of time. Everyone needs to be humbled by the market every once in a while to keep them honest. What matters most is what you learn from it.

The biggest question is probably -- has the investment thesis broken? One thing I have learned is to not fall in love with a stock and if the investment thesis breaks, it is time to say goodbye. I have no emotional attachment to Appen, I'm here to make money. Has Facebook, Google and Microsoft turned away from outsourced content relevance work completely? Will Appen need to rely solely on new markets revenue now, which may have a limited runway, despite growing at an amazing clip at the moment? Will China's revenue continue to ramp up at an unbelievable pace (up 5.5x YoY) or will China too eventually shift priorities and strategies to deal with outsourced data labeling?

The easiest way to tell the difference between an amateur and an experienced investor is that an amateur will tell you that an outcome is a "sure thing". For Appen, no one can proclaim to know what will happen next and how this situation will play out, not even Mark Brayan himself (for what it is worth I think he has been given a hard time and is working hard to keep APX as the market leader during a tough COVID induced transitory period). Whether this is the beginning of a structural decline or a short-term transition point as part of a much longer dynamic journey at the beginning of the AI uptake curve is yet to be seen. I think it is more likely to be the latter, but I'm far from certain. Lou Holtz — 'Nothing is as good as it seems, and nothing is as bad as it seems. Somewhere in between lies realty.'

I think Appen has a chance to evolve and continue to be at the forefront of the AI data labeling industry even as the industry may evolve to the detriment of historical business lines. There are plenty of green shoots: Indeed, Appen's China division is looking particularly interesting right now, if that run rate continues (60% per quarter), it will be on an annualised run-rate of $118m USD by this time next year. The number of new project wins is also very encouraging.

Perhaps the biggest risk of all is taking no risk: for the moment, I am going to stay invested. Switching my sentiment to HOLD until I see a reversal of fortune in the share price and in results (H2 CY21).

Below are the financial profiles of two (very different) ASX technology businesses: Appen and Domain. One is valued at circa $1.5B AUD (Appen) and the other is worth twice as much at circa $3B AUD (Domain). These companies are very different with different revenue models. But, for the sake of the exercise, bear with me. If you saw these financial profiles below, without the context of the business, you would be crazy to think Domain is the $3B business and Appen is the $1.5B business.

• One has been consistently profitable since 2017, the other has not
• One has been consistently growing the top line since 2017, the other has not
• One has circa 2x the revenue and circa 1.5x the profit as the other

Does Appen deserve to be trading at $3B and Domain at $1.5B? Do both deserve to be trading at $3B? One is deemed to have a decades-long runway of growth ahead of it and one is deemed to be facing a structural decline. If you were looking from afar and were told one company was operating in the AI industry and the other in the property listings market, you would be crazy to think that it was the AI business that didn't have a runway ahead of it.

If you were told that one business was the market leader in its industry (Appen) and that the other business was the #2 in its industry (Domain), would you be crazy to think that Domain is the $3B business (trading on 11x revenue) and that Appen is the $1.5B business (trading on 2.5x revenue)? If you were looking from afar and were told that one was confident enough to release guidance on earnings growth of 18%-28% YoY and the other had no guidance, would you be crazy to guess it was Appen with the guidance and not Domain? One has a PE of 23x and the other has a PE of 83x.

Sometimes the market is crazy.

I think this is an excellent appointment; Richard Freudenstein should be ideally placed to help turn around some of Appen's perception issues that have arisen in the past 12 months. If Appen can win back the trust and support of the instos, a re-rate of APX's multiples alone (back to historical levels) will send the share price back to $25+ levels. It will take a concerted effort, strong underlying results, and some carefully positioned 'marketing' to the Street. Also curious as to how Mr Vonwiller will treat his shareholding once he retires from the board. He sold $58m worth (2m units) in June 2020, so hopefully, that is enough cash to fuel his lifestyle across the next decade and he won't need to dip his toes back into his remaining shareholding in a meaningful way (and hopefully not at these deflated levels).

The content relevance division should be growing at the moment given advertising spend figures by the big tech players in the US. The Content Relevance division was given a significant boost by the transformative acquisition of Leapforce in late 2017 for A$105M. This was funded primarily by debt but also a small ($30M) capital raising – the only time the company has raised additional capital since its IPO. Leapforce was a global leader in Content Relevance with a crowd-sourced workforce of >800,000 people – and the deal immediately established APX as the #1 global player.While Appen typically does not disclose the identity of its clients, it has revealed that it works with 8 of the top 10 global technology companies, plus we know that Facebook and Google generated the vast majority of Leapforce’s revenue, and that Microsoft was APX’s largest customer prior to the acquisition of Leapforce. Based on the customer relationships graphic from the most recent investor presentation, these would still seem to be large APX customers.

The upcoming earnings announcement will also be interesting. Prior to the last 12 months, Appen had an enviable track record of earnings upgrades. In Appen's short history of listing since 2015, the company has generated 13 earnings upgrades and just 1 earnings downgrade (2020). This included upgraded guidance at four of the company’s half-year result reporting dates (FY15, FY16, FY17 and FY18), two upgrades on the release of full-year results (FY17 and FY18) and the remainder during the normal trading year (including at the time of the material Leapforce acquisition). The company's forward P/E inflated from ~20x in 2016 to between 30x and 50x across 2017 to 2020 and then back down to ~20x in 2021. If we expect CY21 underlying NPAT of $70m, at a $1.44b market cap, Appen currently trades on a forward PE of 21x. With earnings (EBITDA) expected to grow between 18% and 28% in CY21, that equates to a desirable PEG ratio of less than 1.

Xero (XRO) is priced for perfection (PE of ~500) and Appen (APX) is priced for a structural decline (11x FY21 EV/EBITDA).

Which is the better business? 99/100 analysts will tell you it is Xero due to their strong competitive moat, high gross margin, network effects, and pure-SaaS model. But, when faced with the choice of buying Xero at a $17.4 billion valuation or Appen at a $1.37 billion valuation, I choose the latter.

If you can see moats where others don't, you'll pay bargain prices for the great companies of tomorrow. Of equal importance: if you can recognize no-moat businesses that are being priced in the market as if they have durable competitive advantages, you'll avoid stocks with the potential to damage your portfolio.

I think Appen has more of a competitive moat than most analysts are willing to believe right now amid all of the uncertainty (rising competition, regulation of big tech, changing COVID priorities, the rise of self-supervised learning etc.). Appen's growth is volatile; it always has been.

Buying shares in a company means you own a tiny piece of the business. The value of that business is equal to all of the cash it will generate in the future. A business that can profitably generate cash for a long time is worth more today than a business that may be profitable for only a short time.

Return on capital is the best way to judge a company's profitability. It measures how good a company is at taking investor's money and generating a return on it. To see if a company has an economic moat, first check its historical track record of generating returns on capital. Strong returns indicate that the company may have a moat, while poor returns point to a lack of competitive advantage -- unless the company's business has changed substantially.

If historical returns on capital are strong, ask yourself how the company will maintain them. Buying stocks with low valuations (multiples) helps insulate you from the market's whims, because it ties your future investment returns more tightly to the financial performance of the company. If the return on capital is higher than the weighted average cost of capital (WACC), then the company will be profitable over the long term.

As physicist and philosopher Niels Bohr once said, "prediction is very difficult, especially if it is about the future". Yet, that's exactly what we need to do when assessing the durability of a company's competitive advantage. But, sometimes the future throws you a curveball, and that's when you need to reassess whether a company's moat is still intact or the unexpected turn of events has done permanent damage to the company's competitive advantage.

I will leave that up to all of you to decide.

T.E.P.

The broker price target figures I have attached below for you are from prior to Wednesday's business update (19th May). Even before this update, Appen was oversold. I expect upgrades from here, particularly post the August half-yearly results.

This is a major turning point. Appen has provided some detailed visibility into the business on a segment level, that they have never revealed before. Over time, this will highlight the transition of the business to being an AI product-led business with recurring revenue (SaaS).

Over the past half-year, fear, uncertainty, and concerns have been highly exaggerated and this has been accentuated (/been made possible) because of the historical lack of visibility into the business. This restructure and change to segment reporting lines changes the game.

Even very smart people who I have spoken to (who are leaders in the technology sector) have been unfortunately swept up in fear around Appen's competitive moat, considering Appen "a glorified labor-hire company". Today's update and revelation of the revenue split by product/service debunks this myth. But, this is what creates opportunity in the market. Fear creates opportunity. One needs to form a position grounded in deep due diligence and hold it unless the thesis changes.

Here is what the now revealed facts suggest: a) 30% of Appen's revenue is now committed (i.e. recurring). b) The New Markets division (AKA Appen's training data products) comprises 20% of the business (and rising). c) This AI product-led section of the business is at an FY20 run rate of $85M USD and growing at 34% YoY.

If the 'new markets' division (Appen's training data products) of Appen was spun out and listed on the NASDAQ, I would guess it would be valued at $1 billion USD or more, alone, on a circa 10x PSR. Scale AI (a similar sized business to APX's new markets division and similar in tech capability) suggests that it could go for a lot more than that (Scale is valued at $7 billion USD!). This is the tech-enabled component of APX, where the majority of the future growth of the business will stem from.

The global division of APX is also a valuable division of the company, producing consistent mid to high single-digit profitable growth. This part of Appen is producing all of the profit currently (the new markets division has been cash-flow negative but is now at breakeven as of H2CY2020). I'd conservatively value the global division component of APX (i.e. the global tech giants relevance work) at a 10x EBITDA multiple for a value of around [($42m USD H1 + $46m USD H2) x 10 EBITDA multiple] $880m USD = $1.135 billion AUD.

Adding the two components together, using this sum of the parts methodology, I arrive at a current fair valuation of Appen of circa $2.425 billion AUD ($1.29 billion AUD + $1.135 billion AUD). With 123 million shares on issue, this equates to a current fair value share price of circa $20 per share. If you are comfortable forecasting out one year into the future and assume that the global division grows at mid-single-digit figures and that the new markets division grows in line with the AI industry average, one arrives at a price target at least $4 higher than that (i.e. $24+).

A high volume day (such as that we experienced) is highly likely to have been driven by institutional purchases. My own analysis for a 12-24 month view falls somewhere between Citi & Ord Minnett. I believe that achieving a share price in the $25 to $31 range within that timeframe is very achievable.

Lastly, let me say this. If Mark Brayan's understated style (measured, calm and trustyworthy) (whom I support FYI) leads to an upgrade over the next 12 months, then a return to $30+ is well and truly in sight. I'd much rather have a CEO like Mark at the helm, who has a history of under-promising and over-delivering, than a brash and bold CEO who will claim anything to move the share price higher in the short term at the expense of trust, longer-term.

As a part-owner of this business, I am not a trader and I am investing capital here for consistent profitable growth in future cash flows. In the short run, sentiment and storytelling are all-knowing. But, in the long run, cash flows are the only thing that truly matters for the share price.

APX Multiples -- Are They 'Fair'?

Let's start by understanding exactly how Appen is being valued right now versus how it has been treated historically.

Computed below are the key multiples for understanding Appen's valuation over Jun 2017 to April 2021, reported quarterly.

These are: TEV/LTM total revenue, TEV/NTM total revenue, TEV/LTM EBITDA, TEV/NTM EBITDA, TEV/LTM EBIT, TEV/NTM EBIT, P/LTM EPS, P/NTM EPS, P/LTM Normalised EPS, P/BV, P/Tangible BV, P/BV, P/Tangible BV, P/NTM CFPS, TEV/LTM Unlevered FCF and Market Cap/LTM Levered FCF.

Across 2017 to 2021, Appen has traded across a range of:

• TEV/LTM Total Revenue: 2.67x to 8.44x (3.3x currently)
• TEV/LTM EBITDA: 17.33x to 48.82x (25.61x currently)
• P/BV: 4.15x to 19.35x (4.15x currently)

What does this tell us?
Firstly, Appen currently trades at the bottom end of these historical valuation-multiple ranges. In other words, sentiment towards the stock is very low relative to historical levels.

Secondly, anything that affects the (perceived) timing, likelihood or duration of future cash flows will affect these valuation multiples. Thus, I don't think the situation we find ourselves in is as simple as classifying Appen as a cyclical v growth v high-growth stock.

Ultimately, there are a large set of factors such as competitive dynamics, pricing power, the nature of revenue (% recurring revenue), management, etc. that are all influencing broker estimates of future growth rates, margins and subsequently cash flow. This is reflected in the valuation multiples.

But, what valuation multiple is 'fair'?
I consider Altium & Wisetech the closest comparators to Appen on the ASX, albeit they are very different businesses. (The best comparator to Appen globally is Lionbridge AI).

These two businesses trade at enormously higher valuation multiples. Currently, they trade 340% (ALU) to 540% (WTC) higher on a revenue multiple basis and 55% (ALU) to 194% (WTC) higher on an EBITDA multiple basis.

There a few factors that dictate why Altium and Wisetech are being rewarded with higher multiples. The multiple premium afforded to Altium and Wisetech is significantly higher than that of Appen's primarily because of two main factors. 1) Earnings quality and 2) earnings reliability.

1) Earnings quality is reflected below in the operating statistics of ALU v WTC v APX, across gross margin (%) and EBITDA margin.

ALU has a gross margin approximately 1.9x higher than APX. WTC's gross margin is 3.3x higher than APX's. The EBITDA margin premium over Appen is between 2.2x (WTC) and 2.9x% (ALU).

2) On the other hand, earnings reliability can be reflected in the percentage of revenue that is recurring:

• 65% of Altium's revenue is recurring
• 91% of Wisetech's revenue is recurring

What could cause a re-rate in Appen's valuation multiples?
As discussed previously, analyst growth rate expectations are currently rock bottom and well below Appen's guidance. Confirmation from Appen that their long term earnings growth trajectory is sound (probably towards the end of this year with CY2022 earnings guidance) will lead to a normalisation of future growth rate expectations by analysts -- this will lead to broker upgrades and a rising share price. This alone should lead to a 50%-100% increase in share price if the valuation multiples return to APX's average historical levels across 2017 to 2021.

Long term, could Appen reach valuations as lofty as Altium's or Wisetech's?
Possibly. They might not be able to reach 15-20x revenue (which is extreme), but they are making important steps towards transitioning to a higher quality SaaS-based business model. There is one key chart to watch. The percentage of Appen's revenue that is recurring.

In 2H20, Appen reported a 343% increase in committed revenue versus 2H2019. This $92M of committed revenue in 2H20 is 31% of the total, up from $36M in 1H20 (12% of the total). In my opinion, this has been underestimated by the investment community.

Appen's acquisition of Figure8 was a game-changer for the company strategically. Figure Eight materially increases the quality of Appen's revenues and the breadth of its customer base via high growth, high-gross margin recurring revenue from annual platform subscription fees (SaaS model) earned from its ~200 customers.

The acquisition combined the scale, quality and language expertise of Appen's leading global crowd, supported by its efficient crowd management platform, with Figure Eight's innovative data annotation platform, to create a unique end to end solution.

We are only beginning to see the fruits of this acquisition. Accounting for differences in gross margin, EBITDA margin and recurring revenue, it is my view that Appen deserves to trade on at least half of Altium's revenue multiple -- which would be a revenue multiple of ~7x (versus ALU at 14.6x). On 2020 revenue of $600m that suggests Appen currently deserves a valuation of circa $3.5 billion AUD -- for a share price of $28.

If Appen could reach ALU's 65% recurring revenue threshold, continue growth at a materially higher pace compared to ALU and increase EBITDA margin to close the gap towards ALU's 34% margin, it would be reasonable to expect that Appen too could trade on 10x+ revenue for a market capitalisation north of $6 billion AUD ($50+ share price).

Appen remains the largest holding in my portfolio -- I am grateful that Mr Market has provided me with a chance to enter at this level.

Time will tell if I am right.

T.E.P.

A variety of factors are conspiring against Appen at the moment (and may continue to over the coming handful of months/quarters):

1. The ongoing fall of the USD (Appen's revenue is almost entirely generated in USD)
2. The evolving regulatory landscape of the US NASDAQ tech giants (Appen's customers)
3. The evolving product landscape of the US NASDAQ tech giants (Appen's customers) brought about by COVID -- shifting away from existing mature AI products into newer AI products
4. A conservative company outlook for CY21 (with earnings back-ended into H2) [a consequence of #3 above]
5. Extremely frothy valuations across the board leading to nervous and skittish investor behaviour
6. The rotation of fund managers away from growth stocks back to value stocks
7. Broker downgrades (a consequence of all of the above, primarily #3 and #4)
8. Increasing competition and pricing pressure (claimed by the institutional investors), but I do not see material evidence of this in the relevance division and Appen have also disputed these claims)

Personally, I believe all of the 8 factors above are temporarily and not long-term structural trends that are here to stay (with the exception of maybe #1). So, in these moments, I like to remind myself of 2 of WB's favourite quotes:

• "All there is to investing is picking good stocks at good times and staying with them as long as they remain good companies." - WB
• "The best thing that happens to us is when a great company gets into temporary trouble...We want to buy them when they're on the operating table." - WB

To have confidence that Appen will return to its long term growth trajectory (in line with the industry growth rate), one needs to have confidence on all (or at least most) of the 8 points above. In terms of #3, all indications suggest to me that Appen has entrenched itself into the new emerging projects that the tech giants are now beginning to grow. These will take time to mature but Appen will be a trusted partner of the tech giants as they pursue new AI products, I simply doubt Appen will get left behind as the tech giants change their product focus (away from advertising - in the light of increasing regulatory pressures.)

My predictions for the upcoming annual results are as follows:

• CY2020 EBITDA = $106M AUD
• CY2021 EBITDA = $131M AUD to $140M AUD
• CY2022 EBITDA = $162M AUD to $185M AUD

The low end of the range assumes 24% growth in EBITDA YoY. The high end of the range assumes 32% growth in EBITDA YoY. Astute observers will note that this is a +/- 4% range from APX's guidance around growing earnings in line with industry growth (28% p.a.)

Anyone who is expecting more than $140M AUD for CY2021 EBITDA is perhaps a bit too optimistic. Or perhaps I am being conservative by discounting the power of operating leverage and expanding margins. Regardless, in my eyes, anything above $140M EBITDA for CY2021 would be a spectacular performance (that would be up 32%+ YoY).

If the midpoint of the above EBITDA ranges are hit, on current prices, APX would be trading on 19x CY21 EV/EBITDA and 15x CY22 EV/EBITDA (noting $126M in cash as at 30th June 2020 and a market cap as it stands today at $2.8 billion AUD).

Is that too high? Perhaps. Historically, across the markets as a whole, the general rule of thumb is that EV/EBITDA under 10x is attractive and undervalued. However, that is for the market as a whole, not the tech sector, which commands higher multiples due to stronger growth.

Worldwide, the average value of EV/EBITDA in the technology & telecommunications sector as of 2020 was a multiple of approximately 21.1x, an increase from 15.1x in 2019.

• Afterpay is at 971x EV/EBITDA.
• Tesla is at 128x EV/EBITDA.
• Wisetech Global is at 102x EV/EBITDA.
• Xero is at 95x EV/EBITDA.
• Altium is at 58x EV/EBITDA.
• NextDC is at 55x EV/EBITDA
• SEEK is at 42x EV/EBITDA.
• CarSales.com is at 32x EV/EBITDA.
• Appen is currently on 25x EV/EBITDA (assuming $106M AUD in CY20 EBITDA)

Today, there are some examples of eye-watering high levels of EV/EBITDA in the tech sector, but APX, at current prices, is not one of them.

T.E.P.

https://www.tepinvestments.com/blog/appen-apx-entropy-dcf-models-amp-investor-behaviour

“The trick is not to learn to trust your gut feelings, but rather to discipline yourself to ignore them. Stand by your stocks as long as the fundamental story of the company hasn’t changed.”
– Peter Lynch

Appen: An Aussie Star on the World Stage

From the 5 golden Aussie technology stocks (the ‘WAAAX’), at current prices, we like the look of Appen most. Given our investing style and Appen’s recent fall from grace, the smallest member of the 5 WAAAX giants is looking particularly alluring right now. In this research note we will evaluate why, but first, let us lay out the context:

Appen collects and labels images, text, speech, audio, video, and other data used to build and continuously improve the world’s most innovative artificial intelligence systems.

Appen’s competitive moat includes having a global crowd of over 1 million skilled contractors who speak over 180 languages and dialects, in over 70,000 locations and 130 countries, and the industry’s most advanced AI-assisted data annotation platform.

It is this reliable and cost-effective training data that gives leaders in technology, automotive, financial services, retail, healthcare, and governments the confidence to deploy world-class AI products.

The companies utilising Appen to deploy AI products are the real deal. We are talking about the BIG end of town. Although Appen does not disclose their customer’s identity, they do reveal that they work with 8 of the 10 largest Western AI technology companies on Planet Earth.

We believe those customers to be Microsoft (Azure), Google (Google Cloud), Facebook, Twitter, Mozilla, IBM (Watson), VMware and American Express, in addition to smaller AI application developers. Not bad for a somewhat little Aussie company that was founded in 1996.

Working with the giants of the tech world has propelled Appen’s growth. Since 2015, the share price is up an astonishing 1368%. Indeed, Afterpay is the only member of the WAAAX that has outperformed Appen across the past half-decade.

The top line is up 546% across 2015-19, from $83M to $536M. Moreover, with the underlying EBITDA margin holding in a range between 16% and 20%, EBITDA is up 622% over the same 5-year period from $14M to $101M AUD. Over this 5-year period, the PE multiple has ranged from 44 to 114, reflecting its status as a high growth firm.

Appen’s sell-off: sizing the opportunity

On the 10th of December 2020, Appen downgraded its FY2020 earnings guidance on the back of its heavyweight US-technology customers confronting new lockdowns in California and shifting their resources to new projects away from data-intensive mature projects in the wake of COVID-19.

Appen is still projecting overall YoY growth in both revenue and earnings, but this growth is now signalled to be slower than was previously indicated – the company noted that it was not experiencing the usual ramp-up in sales that has been experienced historically at this time of year.

Consequently, the EBITDA range for FY20 was downgraded from $125M-$130M to $106-$109M AUD, which includes an FX adjustment (headwind) from US$0.70c to US$0.74c.

The key piece of information for investors to understand is that the earnings update relates to the re-allocation of work by Appen’s customers. Some of Appen’s US tech clients have temporarily shifted resources (engineers) away from major existing projects, which are data-intensive, to focus on new project opportunities (that enhance their long-term resilience and value), which are less data-intensive. The mature projects have not been abandoned completely.

The share price is down circa 43% from the all-time high of $43.50 to $24.66 at the time of writing, for a market capitalisation of circa $3 billion AUD. We have taken this opportunity to add Appen to the TEP Investments portfolio and it is now our largest holding.

“Whether we're talking about stocks or socks, I like buying quality merchandise when it is marked down.” – Warren Buffett

We take a long-term view on value creation, seeking companies that have a disciplined growth strategy, a laser-like focus on their core market and are the leader in their industry. This process gives one the opportunity to look past the short-term valuation metrics and short-term performance if we can have confidence a business has the platform and management team to create a more valuable business on a medium and long-term basis.

What we are looking to identify, is a business that might look expensive on short term valuation metrics (i.e., one year forward PE multiple) or look ‘down and out’ because of a temporary earnings downgrade but offer value in 3 to 5 years on the back of strong compound revenue and profit growth.

Will this earnings downgrade be short-lived or is it the beginning of a structural decline?

For earnings downgrades, if the information released to the market causing the downgrade does not impact your original investment thesis, then your margin of safety has improved, potentially dramatically. For any earnings downgrade with a high-quality company, we consider if the change is a result of cyclical or structural changes.

Our view is that this is not a structural (long term) change in demand for Appen’s products and services. The largest western technology firms who are Appen’s customers (e.g., Microsoft, Google, Amazon) continue to grow at pace. Moreover, Appen has an opportunity to secure blue-chip Asian technology clients in the future, such as Tencent or Alibaba, and we believe this opportunity is not priced into the share price. Appen also has an opportunity to expand into other verticals such as finance and healthcare.

On the Appen earnings update investor call, CEO Mark Brayan highlighted that the company believes this COVID-related development could be positive longer-term in terms of presenting new revenue avenues that could complement a return to growth on their historical programs of work. However, currently, Appen has a lack of visibility on this and their near-term work orders and this is something we are vigilant of as a risk. This is the first time we have seen evidence of Appen’s customers struggling to cope with multiple R&D programs of work. Our view is that ramp ups in engineering, product and software hiring at the US tech giants may indicate that this is a challenge that will be managed through over the next 12 months and beyond and is temporary.

The long-term trends for Appen remain exceptionally favourable. Spending on artificial intelligence is growing rapidly at 28% annually and AI adoption should accelerate in a post pandemic environment according to market observers such as the Boston Consulting Group (BCG). Further, online advertising, a major source of revenue for Appen’s key customers and a reasonable indicator of their spend, is forecast to rebound strongly in 2021 according to analyst forecasts. In the Q&A section of the trading update conference call, Mark Brayan declared that he expects Appen to resume a growth trajectory in 2021 in line with market growth. The structural tailwinds, as well as the strength of the existing pipeline for 2021, support a return to strong growth rates in 2021 in line with industry trends.

Our conclusion is that this is a short-term cyclical earnings impact not a structural headwinds earnings downgrade – AI is here to stay.

How much is Appen worth?

Valuing Appen is notoriously difficult.

The challenge of this task is evidenced by the fact that broker fair-value estimates for Appen i) swing wildly and ii) vary wildly among the sample set. When a single-year earnings downgrade of circa 15% shifts a broker estimate by more than 25% despite the value of the business being the present value of all future cashflows across the lifetime of the business – we have an overreaction.

It implies that buy-side research firms have no real indication as to the length, likelihood, and size of Appen’s future cashflows and are tinkering not just with short term earnings but modifying key long term inputs to their Appen valuation models. Changes to the long-term growth rate or the terminal value of Appen leads to enormous variation in the fair value of the business.

With pessimism and uncertainly clouding investors’ judgement right now, short term fluctuations appear to have crept into long term assumptions. This forms part of our investment thesis and is part of what we aim to take advantage of over the next 12-24 months with the Appen share price as long-term growth assumptions normalise again once Appen proves this current set back is temporary.

We have tested 3 scenarios across the TEP Investments APX discounted cash flow valuation model: i) the bear case, ii) the base case and iii) the bull case and determined a fair value target price using a blended valuation across the 3 scenarios.

Across all 3 scenarios, the following has been held constant:

• A corporate tax rate of 30%

• A weighted average cost of capital (WACC) of 9.9%.

  • This is based upon a cost of debt calculation that reflects i) the risk-free rate, ii) the country risk premium and iii) the sector’s (technology) specific cost of borrowing

  • Annual inflation rate of 2.40%

• An EBITDA multiple method for calculating the terminal value

• Assumed total shares on issue remains constant at 122.35M

• Assumed the underlying EBITDA margin is 17% in 2020F and rises gradually (linearly) to 21% by 2030F (across 2015-2019, the EBITDA margin has ranged from 16% to 20%)

• Assumed the EBIT margin is 11% in 2020F and rises gradually (linearly) to 16% by 2030F (across 2015-2019, the EBIT margin has ranged from 12% to 16%)

The Bear, The Bull & The Human

In the bear case, we assume that Appen is unable to meet its promise of returning to an industry standard growth rate (28% p.a.). We assume Appen’s top-line grows at 20% in 2021F and then gradually declines (linearly) to 12% growth YoY in 2030F. This is a low-growth scenario in which Appen would be underperforming the industry growth rate by a significant margin. We consider it a highly unlikely scenario but find it a useful exercise to determine a likely floor in the share price. In this scenario, we assume a terminal value multiple of 6x 2030F EBITDA to match the assumed low growth trajectory of the business. This scenario leads to a fair value share price estimate of $18.81 per share.

In the base case, we assume that Appen can meet its promise of returning to an industry standard growth rate (28% p.a.). We assume that Appen’s top line grows at 28% in 2021F and then gradually declines (linearly) to a growth rate of 20% YoY in 2030F. This is a moderate growth scenario that tapers off, such that Appen is likely to have underperformed the industry growth rate overall across the forecast period (2020-2030F). We consider this a viable scenario, but, there is a fair chance that the business outperforms this scenario if Appen can match and maintain the industry growth rate.  In this scenario, we assume a terminal value multiple of 7x 2030F EBITDA to match the moderate growth trajectory of the business. This scenario leads to a fair value share price estimate of $36.71 per share.

In the bull case, we assume that Appen beats the projected industry growth rate (28% p.a.). We assume Appen’s top line grows at 28% in 2021F and then rises to 36% YoY growth in 2022F buoyed by a bullish post-COVID environment and maturation of new customer projects and/or additional growth from one or more new blue-chip clients (tech giants – Alibaba, Tencent etc.) in China. We then assume that the growth rate gradually declines (linearly) back to a growth rate of 28% by 2030F. This is a high growth scenario that beats the projected industry growth rate and then tapers off in line with the industry growth rate. We consider this a possible scenario. In this scenario, we assume a terminal value multiple of 8x 2030F EBITDA to match the high growth trajectory of the business. This scenario leads to a fair value share price estimate of $66.98 per share.

Finally, reverse engineering the model to arrive at the current share price (circa $25 per share) reveals insights into what the investor community currently collectively believe the growth rate of Appen will be going forward.  Using the same margin, WACC, and terminal value assumptions as per the above scenarios, reverse engineering the model reveals that the investor community are collectively pricing in a revenue CAGR of 19% across 2020-2030F for a share price of $25 per share, well below the expected industry growth rate (28% p.a.). This contrasts with the Appen track record of high growth; APX achieved a 54% revenue CAGR across 1H FY2015 to 1H FY2020.

Ultimately, using a blended valuation approach and applying a likelihood of 15% to the base case, 50% to the base case and 35% to the bull case, we arrive at a TEP Investments target price for Appen of $44.62 per share.

The law of entropy & investor psychology

However, unless you have a crystal ball (we do not even get one here at Oxford University, unfortunately), projecting the future growth rate of this high-quality business is a little bit like throwing darts at a dartboard. We can refine our throw with practice, industry insights and objective research. Yet, the law of entropy states that all things in this universe trend towards disorder – there is no guarantee where our dart will land.

Moreover, left unchecked, disorder increases over time. Energy disperses, and systems dissolve into chaos. The more disordered something is, the more entropic we consider it. The Greek root of the word translates to “a turning towards transformation” — with that transformation being chaos.

Entropy is fundamentally a probabilistic idea: For every possible “usefully ordered” state of molecules, there are many, many more possible “disordered” states. Just as energy tends towards a less useful, more disordered state, so do businesses and organizations in general. Rearranging the molecules — or business systems and people — into an “ordered” state and maintaining it requires a continual injection of outside energy.

So, although we cannot hope to predict the world and future business outcomes with high levels of precision, can we still win on the roulette table of life that is the investing world?

I believe so.

There is something more predictable than the future ever could be and this is what the best investors manage to understand and take advantage of – human behaviour.  In short, we can predictably assume that human behaviour is unpredictable. Humans are not rational market participants. Humans are irrational market participants.

Humans emotions and investment markets will always continue to swing back and forth along the investment pendulum. Charlie Munger’s ‘Psychology of Human Misjudgment’ speech given at the Harvard Law School in the mid-1990s is in my eyes the finest investment speech ever given. The speech does not directly touch on investments but that tells you something – the most enduring advantages are psychological.

We do not need to predict the future of Appen with certainty. What we do need to know is that investors are currently viewing Appen with a bearish view of its future growth trajectory – this has been highlighted via the DCF analysis above. The pendulum of investing has swung out towards one of its two endpoints – it is much closer to fear than it is to greed. We also know that a pendulum rarely spends considerable amounts of time at either end of its arc and the further it swings out to an extreme the higher the likelihood that the pendulum will reverse towards the other direction.

With a more positive growth outlook from Appen at the February 2021 full year FY2020 update or with a more positive update at the H1FY2021 update in August 2021, analysts and investors will begin to normalise their view of long term Appen growth rates towards the base case and potentially to towards the bull case. Fair value of this company swings extraordinarily with relatively small shifts in the long-term growth rate.

With the earnings downgrade, the market pendulum has swung to the bears – but just like the inescapable tug of gravity, the reversal of the ‘sentiment pendulum’ is inevitable.

It is almost impossible to predict the future of Appen’s business performance flawlessly, but it is almost a certainty that investor sentiment will swing once more… and with it… the share price.

T.E.P.

“We can all observe that stock prices, set in an auction market, are more volatile than business values. Several studies and casual observation reveal that individual prices oscillate widely around a central price year in year out, and for no apparent reason. Certainly, business values do not do this. Over time, this offers the prospect that any business, indeed all businesses, will be meaningfully mispriced.” – Nicholas Sleep