Using NBA Metrics to Scout Superstar Startups

What makes NBA player (and back-to-back MVP) Giannis Antetokounmpo a bona fide superstar? And what would you look for in the numbers to spot the next Giannis? In fact, just as in tech companies, many sports franchises are trying to answer a basic question: what do the numbers today tell us about the possible outcomes of tomorrow — and what (or who) do we need to get to a winning outcome?

In basketball and tech in particular, a deeper understanding of efficiency — both in how to measure it and how to leverage that to build winning teams — and usage has changed the game in the last decade. Efficiency is essentially a snapshot view of how well a player or go-to-market team can perform, given some constraint such as cap space or advertising budget, while usage helps us understand how that efficiency will hold up over time. 

To mostly productive ends, NBA teams now use efficiency and usage as the metrics that dictate the way they use possessions and how they find superstar players and build their lineups around those superstars. Many tech companies are starting to use a similar framework when allocating capital against go-to-market (GTM) efforts, often measured as lifetime value (LTV) over customer acquisition costs (CAC). 

The emergence of the usage/efficiency frontier from basketball demonstrates how to use the concept in tech going forward. We believe a new paradigm for efficiency at the growth-stage will emerge — one that shifts the focus away from static LTV/CAC comparisons and toward a longitudinal view that considers the different paths to superstar efficiency at scale.

What is efficiency?

“How many points can Giannis score when given the reins on 25% of his team’s possessions?” “How much long-term revenue can a sales team generate when given a budget of $1M?” 

Though these questions derive from two very different scenarios, both are about understanding efficiency: how well can a player or sales and marketing team turn an opportunity into a positive outcome?

At its core, any efficiency metric is an expression of performance vs. a constraint. In basketball, this could be points per possession; in software GTM teams, it’s often LTV/CAC. Performance metrics express a desired outcome (such as points or customer lifetime value) divided by the opportunity cost (possessions or dollars invested). 

But these ratios, considered in a vacuum, are insufficient to make strategic decisions. Player A who scores six points on three possessions per game is very different from Player B who scores 30 points on 15 possessions, as Player B maintains their efficiency over three times as many possessions. Considering efficiency with the context of usage shows how the ratio holds up when the player is given more opportunities.   

In fact, when it comes thinking about how to win in the future, teams and companies really care about marginal efficiency: what is the efficiency of the next possession or next dollar invested? Said differently, if we allocate an extra possession every game to this player or an extra million dollars of investment to our sales and marketing teams, will the return be just as productive? Teams and companies are looking for repeatability, and the best indicator is the “sample size” (or denominator) of the efficiency metric today – also known as usage. This makes intuitive sense: we’re much more likely to trust a restaurant with 200 4-star reviews than one with a single 5-star review. 

The larger the denominator of an efficiency ratio (possessions used or dollars invested), the more likely the efficiency is to persist with an incremental allocation. Of course, when it comes to young players or newer companies, their efficiency metrics likely include smaller denominators. So how do you predict which high-efficiency players or startups will become the superstars of tomorrow? Let’s start by looking at how efficiency metrics became prominent in both the NBA and in go-to-market teams at high-growth companies.  

The role of efficiency metrics in the NBA

Analytics in basketball have evolved rapidly in the past couple decades. For many years, the discussion centered almost exclusively on performance or outcomes-based statistics — points, rebounds, and assists per game. These statistics often represented rough proxies for the relative value of players; crossing 20 points per game, for example, often implied a player had broken into the upper echelon. 

But, just like there’s more to a business than revenue, there’s much more nuance to basketball than points per game. At some point in the 2010s, these summary statistics started to give way to ones that added context to the pure outcomes and showed not only how much a player scored but also how efficiently they did so. 

This focus on efficiency has radically shifted how the game is played. Teams like the Houston Rockets have put into practice the realization that 3-pointers and shots around the rim are typically the most efficient — and have all but eschewed the mid-range jumper that was a hallmark of superstars like Kobe Bryant and Tracy McGrady in the early 2000s. 

Moving to a solely efficiency-based analytical framework is not a perfect fix to the limits of a “points-per-game” lens, however. Looking just at offensive efficiency in the 2019-2020 NBA season would place that season’s MVP Giannis Antetokounmpo behind less valuable players like Hassan Whiteside and teammate Khris Middleton. 

So how to properly account for the value of a player like Giannis? This is where usage has come in. In basketball terms, it’s the share of their team’s possessions a player bore that end with a given player shooting a field goal, shooting free throws, or turning the ball over. When we map efficiency against usage, our intuitive sense for the most valuable scorers begins to match the data.

The relationship between usage rate and efficiency is an inverse correlation for the most part, and this makes sense — a player that only shoots when wide open is likely to have an extremely high efficiency, but those who are tasked with finding a shot on 20%+ of their team’s possessions are going to find themselves trying to make the most of difficult positions. And, when asked to take on an extra possession, the 20%+ usage players are likely to do so at similar efficiencies to their average — whereas the player who only dunks or shoots while wide-open may struggle to replicate their efficiency when forced into shooting in unfamiliar situations.

Notice we said “for the most part”: around the 30% marker, a small batch of players buck this trend of decreased efficiency with increased usage. These are the players that match our intuition for the most valuable scorers in the game — even when tasked with more than a third of their team’s possessions, these players often score at more than 115 points per 100 possessions. For reference, the 2016-2017 Golden State Warriors, considered one of the greatest offenses in the history of the NBA, averaged 116 points per 100 possessions as a team. If you have a player like Steph Curry achieving that efficiency or more over 30% of your possessions, it is much easier to outfit your team with players around him that can score at sky-high efficiencies in smaller, easier roles. This combination for the Warriors was the secret to their all-time offense.

Efficiency for high-growth companies

For investors and startups, the parallels to growth-stage companies are instructive. The simplest metric to evaluate a business is revenue. After all, the shorthand for most growth-stage valuations is as a multiple of revenue (though it’s also more complicated than that). Still, it matters not only how much revenue is generated, but also how efficiently it is done.

A common metric to express efficiency is LTV/CAC, which asks how well a company makes use of its investable dollars to generate net profit, adding helpful nuance to pure revenue or growth numbers. For example, it can highlight scenarios in which an outsized portion of revenue has been driven by heavy investments in outbound sales for an enterprise SaaS company or through paid customer acquisition for a consumer-facing one. But, in largely the same way that sorting by efficiency provides a misleading list of the most valuable NBA scorers, relying solely on the lens of LTV/CAC snapshots can miss the forest for the trees. 

For growth-stage companies, the additional layer of context is scale — how big is the business that achieves a given LTV/CAC? Like in basketball, high efficiency paired with scale is a hallmark of a valuable tech business. In this case, scale means how much investment in sales and marketing comprises the denominator for the LTV/CAC ratio. When we are evaluating a growth investment, and the company plans to 5-10x in size over the course of our investment, we care deeply about this scale overlay.  

LTV/CAC on its own is most useful to investors and startups as a means to predict and optimize capital efficiency at steady-state: it matters less where exactly a company’s LTV/CAC is today, and more what it implies about the company’s ability to maintain capital efficiency as it scales. 

Most times, investors and operators use a static LTV/CAC to inform their future view. This can be a helpful (if blunt) proxy. Like in basketball, the relationship between LTV/CAC and scale tends to be an inverse correlation — often the early adopter customers are the cheapest to acquire, especially if you have strong product-market fit. With an assumption around the decay of efficiency vs. scale, we can use LTV/CAC at a given scale today to forecast the way it is likely to evolve. For example, an LTV / CAC of >5x at $50M of CAC spend may imply an efficiency of closer to 3x at $100M of spend.

And, when considered as a time series, LTV/CAC becomes even more instructive. With the benefit of seeing a company’s efficiency evolve over time, we can more precisely gauge the likelihood that it will end up in the high-efficiency, high-scale “superstar” quadrant.

Spotting superstar players and companies

Consider the case of a high-efficiency startup in its hypergrowth days. The static LTV/CAC may jump off the page at 5x or even 10x+, an exciting surface-level signal for investors to build conviction and for operators to justify pouring “fuel on the fire” to their GTM efforts. But, the progression of this metric is most important. To use an NBA analogy, consider two illustrative paths for the “high-efficiency startup”: the James Harden path and the Pascal Siakam path. 

Both players looked like potential superstars at the same early juncture in their careers. Their respective third seasons in the NBA both registered at 120+ points per 100 possessions, which is typically a top 15% mark in the league, and well above the efficiency (y-axis) threshold for the “superstar” quadrant. But, the question of efficiency at scale was very much unanswered at this point, as these marks were achieved at sub-”superstar” usage levels, closer to 20%. 

In some sense, this highlights the value of additional “game film” for players (and companies). Following these seasons for each of the two players, their respective teams made very different long-term decisions: the Oklahoma City Thunder traded James Harden to the Houston Rockets, and the Toronto Raptors signed Pascal Siakam to a maximum contract extension at 4 years / $130M.

Both the Thunder and Raptors would have been well-suited to see the “game film” from the next season — both Siakam and Harden took on substantially higher usage rates, closing in on the 30% usage threshold (x-axis) for the “superstar” quadrant, and saw a commensurate dip in efficiency. But the shape of these progressions was very different. Harden maintained a “superstar” 115+ efficiency, while Siakam dipped closer to a league-average of 108 in the 2019-2020 season. 

Of course, with this additional data point, GMs, coaches, and fans can see the difference in likely long-term progression between a player like James Harden in his fourth season in 2013 and Pascal Siakam in his fourth season in 2020. But without the benefit of hindsight, they may have still been able to predict this divergence. The relevant question is how productively the player would make use of additional possessions. 

We can often predict this future efficiency by studying a player’s scoring habits. In his third season in the NBA, James Harden was able to score efficiently in a variety of ways — his points that year were evenly distributed between shots near the basket, three-pointers, and free throws. Pascal Siakam in his third season, on the other hand, leaned much more heavily on one type of scoring, with more than 60% of his points coming from shots near the basket. When asked to up his usage the following year, James Harden could efficiently allocate these incremental possessions in a variety of ways that kept defenses guessing; Pascal Siakam’s more one-dimensional offensive skillset made maintaining his efficiency with the added workload much harder.

Simply put, the depth of a player’s offensive arsenal is often the best predictor of how efficiency will evolve with increased usage. For tech companies, a similar principle applies. A company with early traction in several products or different buyer types is likely to tolerate increased “usage” (investment) like James Harden. For example, a company that uses a variety of channels to acquire customers now (as opposed to relying on one channel like paid search ads) suggests that they can increase spend in the future over multiple channels without getting saturated. 

On the other hand, a company with a narrower remaining Serviceable Addressable Market (SAM) today that will rely upon new products or new buyer types for its next leg of growth may see efficiency evolve more like a Pascal Siakam. The implication here is not that these types are bad companies, just like Pascal Siakam is not a bad player; he remains well-above average in the NBA when considering the usage he was tasked with last season. Rather, it is that not all high-efficiency “startups” are built the same. In the tech ecosystem, there is a common rule of thumb for LTV/CAC that states a high LTV/CAC (say >3x) implies an under-investment in sales and marketing. This axiom could, however, be re-stated when considering usage as: how does one build a company so that the sales and marketing machine can operate comfortably in the “superstar” zone of 3x LTV/CAC for years to come? 

Consider the anonymized “SaaS Company A” below. Its trajectory is a poster child for the James Harden “glide path” to the superstar zone: the high LTV/CAC early-stage company that can tolerate increased investments in sales and marketing with limited dents to efficiency. In other words, a company that can dramatically expand their go-to-market efforts without seeing their LTV/CAC plummet. This is often propelled by a deep or growing SAM, either for a company aimed at a substantially large market, or one driven by founders and operators that have a knack for “finding market” as they refine their product or add new use cases.

Superstars and network effects

There is also the rare company that can fight the gravity of increased scale and actually improve efficiency over time. These companies, often powered by network effects, are best personified by reigning NBA MVP Giannis Antetokounmpo. His efficiency vs. usage progression over the past seven years has been nothing short of remarkable — defying the typical trajectory, he took on increased usage and added efficiency five years in a row.

A flywheel effect of sorts has driven this evolution. Antetokounmpo came into the league as a player known primarily for a rare combination of size and agility, but he has since added strength to supercharge his scoring around the rim, a highly-effective set of post-up moves and floaters, and even some three-point shooting. These additions would each be effective in a vacuum, but together they are devastating: his ferocious scoring around the rim leads to more space to shoot from the perimeter, and his newfound shooting stroke creates more opportunities for ultra-efficient drives to the basket.

This impact of this flywheel effect on Antetokounmpo’s path is not dissimilar to that of network effects on a tech company. Network effects fuel an ever-improving value proposition to customers, which results in easier sales and improved efficiency. 

For example, let’s look at another anonymized efficiency trajectory, for “SaaS Company B”. In 2018, at a ~4x LTV/CAC and ~$50M of S&M spend, this company seemed unlikely to glide into the “superstar zone.” But, powered by network effects, their GTM motion actually became more efficient as the company increased spending. More so than any individual LTV/CAC figure, evidence of this “Giannis Trajectory”, aka signs of the presence of network effects, can indicate to founders and investors alike that the company is poised to benefit from superstar efficiency over a long horizon.

 

With the right context around efficiency, teams, investors, and companies can better understand how to allocate and deploy resources both within a game and over the course of multiple seasons — as well as spot the rising superstars that can make or break their chance of winning. For the growth-stage companies that understand how to achieve and sustain superstar efficiency at scale, they will be in a better position to win markets and grow for a long time to come.