Tegus and the Research Stack
Plus! Apple and Ads; MetaBonds!; Capex; Distribution; Expensive Executive Assistants
In this issue:
Tegus and the Research Stack
Apple and Ads
Expensive Executive Assistants
Tegus and the Research Stack
The other day, Tegus, a provider of industry expert calls, acquired Canalyst, a firm that produces financial models. I'm an avid Tegus user, and (disclosure!) they're a frequent Diff advertiser, so I got a few reader questions on the deal. It's a fun one, and a great example of how labor-intensive software companies selling to businesses can think about strategy.
The business of expert networks has existed for a long, long time. Depending on how you count, it might be one of the oldest forms of financial research: if you're trying to understand what a company does, or what's happening to it right now, track down someone who knows and ask them. Phil Fisher was doing this in the 1930s and writing about it in the 1950s, and Joseph de la Vega writes about looking for rumors and insights on the Dutch East India Company.
The modern expert network business starts with GLG (written up in The Diff last year), which originally produced detailed industry guides but which pivoted to arranging expert calls when customers asked to speak to the people who had authored those guides. The GLG model is, roughly, that an investor will pay $1,000 or so for a 45-minute call, the person on the other line will get around $300 of that, and the company in the middle pockets a margin for 1) running a compliance check to make sure the expert isn't in possession of information they shouldn't be sharing about specific companies, and 2) having such a wide network of experts.
Getting together such a network isn't a trivial problem, but it's also not an impossible one, and often the same people will be in multiple expert networks. Since the networks all lightly anonymize their users, it's entirely possible to end up talking to the same person through multiple connections. And while some calls are unique, many of them will fit into a few basic templates, where the exact questions are somewhat repetitive. (A good category of alpha-producing calls is to find some market that it's hard to get visibility into, and find someone who can tell you how things look in different regions and categories once a quarter. If it's a commoditized market with big public and private companies, this can be a great read into overall industry pricing. And it's likely after a while that there won't be much variance in the questions getting asked.)
When there's a product that's vulnerable to commoditization, a good strategic move can be to completely commoditize it and then move on to complements. This is explicitly what Tegus is trying to do with expert calls. They price them at roughly breakeven, and share the transcripts with other subscribers. This basically inverts the original value proposition of expert networks: instead of a source of unique, proprietary insights, they're a source of insights that everyone has access to. This is part of the story of technological innovations in finance generally, though; there was a time when "I'm going to use a calculator to figure out exactly what stock options are worth instead of using half a dozen rules of thumb" was a quirky, nerdy choice.
And it creates an interesting competitive dynamic. The nice thing about being the first to commoditize a product is that, if it catches on, it becomes mandatory for a large number of customers instead of nice-to-have for a subset of specialists. Once there are mid-quarter updates on how ad budgets are shifting between Snap and TikTok, or on what's happening in cloud spending, it's hard to justify not subscribing.1
Qualitative research is a hard problem to solve compared to quantitative research, because it's harder to determine when it's solved enough to be a viable, competitive product. Bloomberg started out as a great source for the fixed income markets, and since everything is implicitly benchmarked to bonds, they could naturally expand into other areas. And they started out with raw data; adding news was a natural addition to that. Newer companies that harvest SEC filings and other company releases can also assume some level of comprehensiveness, either by comparing the numbers they track to sell-side models or just by incorporating every single quantitative measure a company ever releases.
But with an expert network it's harder to do this, because the space of potential experts and potential topics is so vast. Building a completely open platform means either a) dramatically over-building the expert side, or b) accepting that most of the time, there won't be a match between who clients want to talk to and who is available to talk to them. Tegus solved this early by focusing exclusively on software, which has a few nice advantages:
The software industry changes fast, so there's demand for frequent updates even in established categories.
The industry has structurally high turnover, so the population of former employees of a given public company is high relative to the size of those companies.
Many industries are downstream from software; if you're collecting enough experts on e-commerce payments, you'll naturally have experts on e-commerce itself, which makes it easier to find people who know logistics, advertising, and manufacturing. And investors also tend to extend their views across the supply chain; if you care about ads, it's irresponsible not to invest some time looking into how big advertisers are doing.
Software turned out to be in a roaring bull market, in public and private companies, for most of its existence.
But the meta problem Tegus is trying to solve is not (just) that the expert network business had immense redundancy, with the same few dozen funds making the same calls to the same experts at high prices. The more general problem is that fundamental investment research is partly a business of generating clever insights into the market but can sometimes be mostly a well-paid form of data entry. Assuming analysts really are worth six, seven, or eight figures, anything that reduces the amount of time they spend Googling niche topics, ctrl-F-ing through SEC filings, or copying numbers from press releases into Excel is a good deal.
Hence the Canalyst acquisition. Reducing the gap between the research and modeling process makes both of them a bit faster. There's still a distraction tax from toggling back and forth between a financial model, a company's SEC filing, and an expert call that contextualizes what the company is saying. Right now, the primary source of truth is some combination of an Excel model and what the analyst explains goes into that model, but this is not necessarily the way things will stay. Any time a product has many use cases, and a lot of value is built on it, there's room to peel off most of the value creation by either integrating something with the core product that makes it significantly more valuable or by displacing it entirely. It's hard to compete with Excel's biggest intangible asset, the collective muscle memory of everyone who went through an investment banking program. But it's possible to make Excel on par with another platform that's purpose-built for the research process.
As with many other categories of business software, the real competition is extra monitors and alt-tabbing. The more use cases that can be encapsulated in one product and tightly integrated together, the stickier the product gets. It also makes customer research easier, because the set of tasks for which users leave the core product gets more sharply defined as they leave it less.
A big part of the Tegus bet is that it's an answer to the classic question for any investment research provider: If you're so smart, why are you charging an annual subscription instead of a performance fee? The Tegus answer is that they aren't selling alpha. They're selling beta. And beta is a defensible business. Blackrock trades at 5.6x annual sales, compared to Man Group at 2.2x and Sculptor Capital Management at 2.1x. The money from selling beta is higher-margin and more durable—providing low-cost beta has better economies of scale, margins don't get eaten by specific employees who contribute a lot and can demand high compensation in return, and beta, unlike alpha, never really goes away.
Thanks to Tegus co-CEO Michael Elnick for taking some time to talk to me about this one. And Diff readers can sign up for a Tegus trial here.
Apple and Ads
Apple is hiring to build a demand-side platform, looking for people who have experience with "advertising related products for audiences in the hundreds of millions." The ad itself emphasizes Apple's talk about privacy—given how closely the company is covered, the ad was probably written with the assumption that it would have to appeal to tech blogs as well as the employees Apple is trying to reach. For now, they say this would focus on ads "in Apple Services," so it's not necessarily the case that they're going to start running ads in third-party apps after weakening that exact ecosystem through tighter privacy policies. On the other hand, it wouldn't be the first time the company built a closed ecosystem and then opened it up just enough to capture lots of economic upside, without losing control. That's the story of the app store itself.
Facebook parent Meta Platforms has raised $10bn in bonds. The company doesn't need the money; they have $40.5bn in cash and marketable securities as of the end of last quarter, and generate $58.5bn in annual operating cash flow against $22.6bn in capital expenditures. But they're also trading close to all-time relative to sales and EBITDA. So they may view this as an opportune time to adjust their capital structure; given the high equity comp in tech, it's likely frustrating for the company to feel like it's giving away extra cheap stock to retain employees, and levering up a bit is a good way to make that economically equivalent to higher cash compensation while still retaining the flexibility that equity offers.
The economy's mixed signals proceed apace: capital spending is rising at a faster pace than buybacks ($, WSJ). It's possible that the current economic configuration is more of a redistribution away from companies than a decline in total output: if big companies have reached the point where it's hard to pass on cost pressures, especially for wages, but it's also risky to rely on overseas supply chains, then we could see a combination of higher GDP growth, lower profits, and lower returns to shareholders relative to profits as they duplicate some parts of their global supply chains domestically and bid for comparatively scarce US-based labor.
Of course, the more pessimistic reading is that reshoring is a purely downside-driven concern, and that one reason companies are collectively allocating more towards capital expenditures than towards buybacks is that even with lower stock prices, they don't view their stocks as cheap. In that model, there's been an increase in the risk premium associated with sprawling supply chains and just-in-time inventory management, and the cure for it is to accept a lower return on capital by having the physical capital somewhere safe.
Coinbase has signed a deal with BlackRock giving clients access to crypto through Blackrock's Aladdin system ($, FT). The economics here are unknown, and probably tilt towards BlackRock given its size and relative outperformance. But it's an interesting milestone for crypto: retail customers tend to allocate their assets pretty informally, but BlackRock's institutional customers are very systematic. And a systematic strategy will generally rebalance into an asset class when it underperforms and rebalance out when it recovers. So the net impact of this is less about Coinbase itself, and more that it constrains the volatility of crypto, and means that it's a component of more big institutions' portfolios and thus gets higher investor mindshare and a bit more safety from regulation.
Expensive Executive Assistants
The WSJ has a fun piece on the world of high-priced executive assistants, who can make up to $400k annually ($). One detail that helps explain their pricing power: the recruiters who find them also charge premium fees. The variance in EA skill is probably a lot lower than for other domains—which is true of any job where the goal is more to avoid missteps than to come up with and execute a plan. But the value of avoiding missteps scales with the value of whoever the EA is working with. In that sense, their compensation might be too low; for a sufficiently busy and high-value person, a few administrative mistakes might hurt their productivity by a single-digit percent, and the implied value for avoiding those mistakes can easily be in the seven figures.
One reason that doesn't happen, as the article details, is that the EA title doesn't always apply to EA-like jobs. A chief of staff has the same sorts of responsibilities, but can earn even more. Another reason is that in some jobs, especially the ones that involve rapidly absorbing and synthesizing information to make decisions, everyone is performing EA-like functions. At a hedge fund that runs a main portfolio with industry-specific sub-portfolios, for example, a major job for those sub-managers is to ensure that the manager of the core book is aware of any notable trend that could affect either that industry or companies adjacent to it. And at a company like Apple, part of the role of anyone who reports to Tim Cook is to keep Cook informed about what’s going on in the supply chain, what moves competitors are making, what the government is likely to get upset about next, etc. Fundamentally, a big part of their value-add is ensuring that their boss receives and responds to relevant emails and IMs, and has a well-managed calendar that matches emerging priorities. It’s not most of their job, but it’s a big chunk, and if you take the share of their time spent on executive assistant-style actions and consider their annual comp, you end up with seven- and maybe eight-figure executive assistants.
I saw this when I was in the alternative data business. Some customers wanted data because they thought it would contribute to their investment process. Others would explicitly say they didn't trust the data but hated seeing stocks move every time our reports came out and not being informed about exactly why. With any strategy that a) produces positive but variable returns, and b) is getting more widely used, there's a point where the biggest alpha opportunity is in betting against it. I've even seen this in newsletters; for one investing Substack I read, a year ago you could read the issue when it arrived, trade at the end, and generally make money by the end of the day; a few months ago, you'd have to trade when you saw the headline and then read the rest of the story to see what you thought of the pitch; and now you probably need to use your broker's API or you're not going to be fast enough.