Points Points Points!

Points Points Points!

June 2, 2024

Points points points

Some thoughts about points. Everyone loves them. They let you give away value to users without giving out....value. They're games, status, fun. Or they can break the trust users have with a brand. Airline miles, Starbucks rewards, even the buy 10 get one free punch card are all a form of points.

So what's the problem with them?

The problem with points is that you're giving something away to users that's hard to directly value for some user behavior you want to incentivize. If you were giving away cash, the relationship would be much clearer, but points are in the eye of the beholder, because the giver can always change the redeeming mechanism.

The problem arises in two circumstances:

  1. You're paying too much for the behavior you want, like giving $1,000 for someone to give you a like on Facebook.
  2. You're paying way too little for the kind of behavior you want, like paying $50 for someone to work on content for 20 hours.

The former problem causes you to waste resources. In this case, you're paying way too much money. The latter issue causes resentment and brand degradation. For example, you pay someone points, but at the end of the day, it's only worth $7, and that person put in 15 hours of work and thought they were going to be rewarded in one way, but they were rewarded in another.

The lack of transparency and the breakdowns in trust that ensues. It's doesn't necessarily come from bad motivations, rather it's is a misspricing because you're using an inefficient market mechanism because in crypto, points convert into tokens and you don't know in advance how much the token is going to be worth.

Of course, there's some leeway here. You could be off the mark by a little. That could be fine. If you gave someone points for 15 hours of work that turned into $1,500 when they were expecting $1,700, sure you missed by $200 but they're still overall happy with it. On the flip side you pay them $2,000, they're happy with it, and you only lost $300, but you still got quality work. You overpaid by a little, but it's still within the margins of error.

The problem occurs beyond that relatively narrow bound and in volatile crypto markets that becomes an issue. Toss in bots and points and yield farmers and you often get a tragedy of the commons - everyone is incentivized to farm even though that drives down the actual proper rewards.

Web2 points

What does the traditional points world have to teach us in web3? As mentioned at the top, points have been around forever in traditional retailing.

In the traditional world, points are used in two circumstances and almost always when the giver of points knows the dollar value of those points. The first is in customer retention, where the majority of points systems are used and the second is in customer acquisition. The reason points are used much more for customer retention than acquisition is because when you acquire a customer it's very hard to know in advance the value of a customer and the cost of acquiring fluctuates more than the cost of retaining.

Points in customer acquisition

Where do we see points used in customer acquisition? Cloud credits is the classic example, where you get "points" that have a dollar value but can only be spent on the cloud provider. This is AWS taking the customer acquisition cost, giving it to you in points that can only be translated into usage of their product. It's worthwhile for AWS because overtime they've built their CAC/LTV models. They know the profile of successful vs failed companies and they know close enough that if they give a startup $10,000 in AWS credits, on average, it'll be worth it because every, say, 15 customers there'll be one who pays me $1,000,000. because it's a startup that'll succeed. And the LTV to CAC ratio is going to be positive, but importantly - they know it in advance.

Most startups can't do that because most startups aren't AWS. They can't charge in the same currency that they're paying with. But mainly, they just don't have a good enough model of CAC to LTV, which is a must-have if you're going to give away points for customer acquisition. Imagine if Amazon gave away points but didn't know how it would translate into their bottom line. All of a sudden, they'd be giving away $5 million worth of credits, yet only bringing in $200,000 in revenue. That would be a problem, and then Amazon would change its 'point' system.

This is why customer acquisition is so hard with points - most companies just don't have a good model around it, and barring that, don't know how much to 'pay'. In crypto this becomes harder due to price volatility. One day you'll be overpaying, the next underpaying because your token has moved by 20%.

Points in customer retention

Points come greatly in handy for customer retention. There are lots of examples here - airline miles, Starbucks rewards or even your local coffee shop giving you a free coffee every 10 purchases. In this kind of case, the coffee shop knows that you're going to get a free $4 coffee for every $40 that you spend. Therefore, they know that they're going to lose that $4 in margin, but retain you as a customer. It's a simple equation and they can bake that into the price of a coffee.

The same is true in airlines. It's a little bit more complicated, but the principle holds true. Buy X amount of miles, you'll receive Y amount of miles for free or for some discounted price. The cost of your customer retention is just going to go into general gross margin. What makes dealing with customer retention so much easier is it's much easier to forecast the financial aspect of the points that are being given away. The points will only be given away if a customer actually purchases something. There's a direct financial impact and transaction either way. It's much easier to track and attribute versus measuring sign ups, downloads or any other acquisition metric. Also, often retaining an existing customer is cheaper than acquiring a new one, so the financial calculus makes more sense.

Points points points...should you use them?

Being able to financially forecast the impacts of points can't be discounted. It's the most important factor for utilizing any system with points. Without it, especially when you don't know or don't have as much control over the conversion rate (i.e your token price), you are going to wind up in that place where you're either giving away too much, which is bad for your business, or you're giving away too little, which is bad for your brand. This is where most crypto companies fail when they start dealing in points. Unfortunately, most crypto companies rely on points and tokens to bootstrap their network effect or simply to acquire new customers. In essence they're using the wrong mechanism for the task, which is why most points systems have failed.


Designing Tokenomics by Yosh Zlotogorski

Web3 Research

๐ŸŽงย Ethereum Audible podcast

๐Ÿ’Žย Ethereum Research

๐Ÿ›ย Governance research

๐Ÿ“šย Tokenomics course primer

๐Ÿ“‘ย Blog