From my perspective, it may seem like a good "temporary solution" for vendors, but it creates doubts and friction points for users and organizations. I see some risks in that approach.
#1 Are we assuming that everyone in the company will spend the same amount of credits on the same task? Will there be a personal credit counter for each user or a company-wide counter? There will be advanced users with low consumption and rookies who will need time to learn and optimize the process.
#2 Will there be a clear estimate of the credit cost for each task beforehand? Before starting, the user needs to know how much of their credit they will use. For example, a 20-slide deck presentation. Will all iterations be included until the task is finished, or will each additional iteration consume credits as needed until the task is completed?
The credit approach is a deterministic view in scenarios where outcomes are not yet clear.
Very interesting read as usual. I guess the original flat rates were brilliant for introducing the new tech and getting us hooked. Token pricing ramps margin but comes with user friction, but most are hooked now and will swallow it.
Especially when it comes to work tooling, variable unpredictable pricing models don’t fit with a lot of expenses plans and general budgeting. So I’m intrigued as to where this goes.
I do think anchoring credits to outcomes is solid. I’ve had scenarios with Replit where it’s run up pretty big bills, but having been caught in a doom cycle of fails. That’s frustrating. I’m not entirely clear on how their credits trigger in fairness but if they were charged when an app hits a capability milestone (outcome) and actually works, then it’s much easier to justify as a user
Great work, and your timing is always impeccable. My question is about measurement. For a few of our customers, we transitioned from a traditional seat-based model to a value- or outcome-based model. Then sometimes the measurement of the outcome becomes a sticking point. Do you see 100% trust in the seller reporting the "credits" used, or do you see third parties involved? Without naming names, one product I tried for prospecting, I couldn't figure out how the credits were calculated, so the lack of transparency forced me to move off the solution. Would love your thoughts.
I think there is a future role for value auditors that will be able to answer this questions in a fair and transparent way for both parties. To do this they will need to have (and agree on) a formal value model. That is not the hardest part though. There are agents emerging that can generate and manage value models. A bigger issue is attribution. "Success has many parent." This is also solvable, using techniques from causal machine learning, and these solutions are already used in healthcare. They are not widely known or understood in business.
@kyle, this is an _excellent_ analysis and summary. Great work!
Thanks, Paul!
From my perspective, it may seem like a good "temporary solution" for vendors, but it creates doubts and friction points for users and organizations. I see some risks in that approach.
#1 Are we assuming that everyone in the company will spend the same amount of credits on the same task? Will there be a personal credit counter for each user or a company-wide counter? There will be advanced users with low consumption and rookies who will need time to learn and optimize the process.
#2 Will there be a clear estimate of the credit cost for each task beforehand? Before starting, the user needs to know how much of their credit they will use. For example, a 20-slide deck presentation. Will all iterations be included until the task is finished, or will each additional iteration consume credits as needed until the task is completed?
The credit approach is a deterministic view in scenarios where outcomes are not yet clear.
Very interesting read as usual. I guess the original flat rates were brilliant for introducing the new tech and getting us hooked. Token pricing ramps margin but comes with user friction, but most are hooked now and will swallow it.
Especially when it comes to work tooling, variable unpredictable pricing models don’t fit with a lot of expenses plans and general budgeting. So I’m intrigued as to where this goes.
I do think anchoring credits to outcomes is solid. I’ve had scenarios with Replit where it’s run up pretty big bills, but having been caught in a doom cycle of fails. That’s frustrating. I’m not entirely clear on how their credits trigger in fairness but if they were charged when an app hits a capability milestone (outcome) and actually works, then it’s much easier to justify as a user
Great work, and your timing is always impeccable. My question is about measurement. For a few of our customers, we transitioned from a traditional seat-based model to a value- or outcome-based model. Then sometimes the measurement of the outcome becomes a sticking point. Do you see 100% trust in the seller reporting the "credits" used, or do you see third parties involved? Without naming names, one product I tried for prospecting, I couldn't figure out how the credits were calculated, so the lack of transparency forced me to move off the solution. Would love your thoughts.
I think there is a future role for value auditors that will be able to answer this questions in a fair and transparent way for both parties. To do this they will need to have (and agree on) a formal value model. That is not the hardest part though. There are agents emerging that can generate and manage value models. A bigger issue is attribution. "Success has many parent." This is also solvable, using techniques from causal machine learning, and these solutions are already used in healthcare. They are not widely known or understood in business.