←

Incentive Alignment ^{JUL 2023}

In product development, there’s a lot of talk about how to better "align incentives," but I’ve found that many folks seem to have differing understandings of what this means. What follows is my attempt at taxonomizing what I see as the three degrees of product incentive alignment:

From a high-level, I think there are three degrees of product incentive alignment:

Products that align their goals with what nets the most money

Products that align their goals with the goals of their customers

Products that align their goals with the goals of their user

First Degree

While you can make a lot of money and build an objectively successful business with a product that only has the first degree of alignment, this is often a regulation and ethical danger-zone. These products usually take advantage of addictive tendencies^[1], employ gambling mechanics^[2], exploit in-elastic demand^[3], or use regulatory capture to rent-seek^[4]. Another example is schemes like those used by financial advisors or funds that primarily get paid based on their assets under management (AUM) rather than a fixed fee or proportion of generated profit. The main problem with this type of product is that it usually does a poor job of solving real problems—or it creates as many as problems as it solves.

I think many products that claim to have third degree alignment actually have first degree alignment. A particularly inflammatory example is psychotherapy. Unlike a gym—which can loss-lead per-user revenue against the value of referrals—psychotherapy isn’t typically a viral-growth business. If a therapist were able to help you achieve some objective criteria associated with your recovery, it stands to reason that you’d no longer need their services, thereby churning off their product. For this reason, therapists are generally incentivized to not have their clients set objective success criteria, but rather encourage them to consider therapy to be a journey and not a destination (and therefore as something that is never meant to end). Thinking this way may be correct, but it also cynically fits as an emergent justification to support their business model. Ostensibly professional ethics (and regulated licensing with ethics agreements) negates this, as any upstanding honest therapist will want what is best for their client regardless of the effect it has on their business. Optimistically, this is probably true for most therapists—albeit with the forces of capitalism working again them—but it’s probably not true for some.

Second Degree

Most products have second-degree alignment, which is most common when a business model has a separate purchaser and user. In these cases, the two will almost always have some differing incentives. Consider Facebook, whose customers (advertisers) and users (you and me) have different goals^[5]. This also happens in enterprise contexts—consider your average analytics tool. Its users are often individual contributors (ICs) who need insights for making decisions on the ground. However, because of the need for data security, purchasing authority for such tools typically rests with the office of the CTO/CISO.^[6] When the customer is an executive and the user is an IC, you’ll often see products win by appealing first to their customer’s needs (e.g., with top-down views for reporting, auditing, and monitoring) and then only tangentially address user needs (who often aren’t in the room where these decisions are being made). As you’d expect, this tends to deprioritize good user-experiences, as executive customers aren’t typically knowledgeable enough about the workflows of ICs to make decisions on this vector.

Some of these products can achieve partial steps towards third degree alignment without reaching it fully. This is common when a product facilitates a two-sided market. For example, users might get some real value out of advertising when they are actually interested in the advertised products and find value in being led to them. I’d argue that this isn’t true 3rd degree alignment because the users and customers still have different objective functions, but it is closer. However, I suspect that many products that think they have this probably don’t.

Third Degree

The most straightforward path to third degree alignment is having your customer also be your user, as it is for truly^[7] consumer products and with enterprise products that serve an executive or purchasing authority as the target end user. This is the case for some products like Rippling or Ramp, which sell software that helps business owners run their businesses^[8].

It’s also possible, albeit more challenging, to achieve third degree alignment by innovating on the business model. For example, income share agreements (ISAs) better align the incentives of all parties when it comes to professional training programs. Before ISAs, an educational institution would get paid regardless of outcome. With ISAs, they get paid contingent on (or even in-proportion-to) the outcome they achieve for the students.

Other products can achieve third degree alignment by overcoming their myopic short-term incentives by taking a longer view that nets out to greater eventual revenue. One example of this is a freemium model like the one used by Dropbox or Figma, where products win their customers (companies) by first accumulating a critical mass of users at a free (or discounted) self-service tier (also known as "bottom-up" sales^[9]). High-end gyms, which aim to make their users happy enough that they might refer their family and friends, can also achieve a high degree of alignment^[10].

Trade-offs

Normatively, I think we build better tools when products pursue higher degrees of alignment. It may not always yield the best business possible, but it does seem to correlate well with "healthy" products that truly solve the problems they aim to. Therefore, in a capitalist system where corporations aim to maximize shareholder value, it’s likely foolish to pursue a user-oriented mission without achieving third-degree alignment. This is because, eventually, you’ll certainly compromise on what is best for the user to ensure you net more revenue or please your customers more.

Is it always possible or preferable to achieve higher alignment? I think the answer is usually yes, with a few possible exceptions. One such exception may be when having less alignment allows for greater access. A more "aligned" Twitter or Facebook would likely require that a person pay so that they both have "skin in the game" and so that the product goals line up with their own. However, these more aligned social networks would also, by definition, exclude people who cannot afford to pay. With social platforms living or dying based on their network effects, making such a trade-off could very-well be terminal, as it was for the payment-gated Twitter clone App.net. There is also an argument to made that the social networks would be healthier for both their users and the world in general if their incentives were more aligned. But a devil’s advocate could argue that pay-gating Twitter would diminish its value as a public good, and that allowing people to stay informed and speak truth to power should be free-to-access. Unfortunately, this is often where moralistic ethics conflict with business priorities, and unregulated capitalism will usually win out.

It’s probably also easier to work your way to third degree alignment if you’re willing to stall the growth of your business. Eventually, any growth-oriented consumer product will want to look into bulk enterprise sales—possibly when they’ve completely tapped out the consumer market. Similarly, many well-aligned products will make small decisions overtime (usually at the hand of PMs trying to make names for themselves) that make more money in the short term at the expense of the long term (usually by employing mechanisms that take advantage of human behavior). I think this effect eventually kills products, or at least demotes their alignment in degree over time. For these reasons, it’s going to be much harder for a venture-backed business to build and sustain third-degree alignment^[11]. Ironically, so-called "lifestyle" businesses that don’t aim to achieve parabolic growth are more capable of making continuous decisions that go against their short-term incentives in pursuit of longer-term objectives.

I also think that sometimes alignment can be a double-edged sword, like when incentives become proportional to the amount of money paid or when that proportion becomes itself misaligned with the cost of service. This is especially true when one party can amortize their earning across iterative games. However, this is usually quite rare or at least not obviously a bad thing. It’s also observable with revenue models centered around tipping or take rates. With tipping, like in an American restaurant, there is validity to the idea that better service should be rewarded, and that cost and quality of service is more-or-less proportional to the subtotal. However, tipping in proportion to the bill can also be poorly aligned (e.g. the service rendered for delivering a $100 bottle of wine is the same as it is for delivering a $1000 bottle of wine, but a 20% tip would be 1000% more for the latter). With take rates, a similar principle applies. The service Apple provides with its App Store is nearly the same for a high grossing app as it is for a hobbyist app, but the take rate makes it much more expensive for the higher grossing apps.

Purchasing for Alignment

For consumers making their own purchasing decisions, better aligned products should generally be "obvious" and win. However with any free or discounted product, it may be helpful to ask yourself "how does this make money?" Some businesses make money in less obvious ways like by selling user or order-flow data, which could be something a user morally disagrees with if they knew about it. Other companies are propped-up by venture investments and may become less aligned overtime. Some consumer products also employ practices like planned obstinance which may allow them to win certain markets by undercutting competitors with short-term oriented, price-conscious consumers, but also demotes them to only have first-degree alignment.

In a B2B context, I think it’s important for businesses to think critically about the way they purchase products as to improve the likelihood that the software is well aligned with its users. In theory, more aligned software should make users happier and more productive, and therefore be good for business. The best way of achieving this is probably by allowing users to make their own purchasing decisions as much as possible. But this isn’t always feasible, because it’s often necessary to standardize products across functions for synergetic reasons. In these cases, and more generally, the least you can do is make sure that the end-users are in the room and hopefully consulted when decisions are being made.