Street Pricing Podcast: AI, Guesswork, and Who Stays Accountable

11 hours ago

4 min read

Pricing I/O has published a collection of insights from pricing experts on Marcos Rivera's Street Pricing Podcast about what AI is doing to SaaS pricing, including two points from FutureUP.

Many thanks to Marcos and the Pricing I/O team — for the invitation to the podcast and for including FutureUP in the selection.

Three things from that conversation are worth expanding on here:

1. AI can find the price you were too scared to test yourself

This story comes from my first startup, which used the same AI model that powers FutureUP today.

Our customer sold mobile applications across many countries, at a single flat price everywhere. Richer markets and poorer markets, the same number. They suspected they were leaving money on the table in some places and losing customers in others, but they had no way to know which was which, and no experience with AI to find out.

We modeled it. What came back was surprising: prices that should roughly double in some countries and be discounted by 80% in others. Against a flat price, that is not a tweak but a radical change.

So we ran a real experiment for three months — one group at the original flat price, another at the differentiated prices. Revenue for those products almost doubled. Conversion rates tripled.

AI unveiled a hidden opportunity that nobody had been willing to propose on instinct alone, and the experiment proved it right.

2. It's okay to be an AI expert and love guesswork

Guesswork usually gets a bad name in pricing. It is what you do before you have data. So it surprises people when I say I like it — and that it is where all my AI discussions start.

Before any modeling begins, we want the human instinct in the room. Where do you feel prices should go? Which segments do you suspect are underpriced? What have you always wanted to try and never dared?

Those hunches are not noise. They are a prior. You can structure them scientifically, and often you should — but even unstructured, they give you direction. Without them, you are testing a thousand possibilities, most of which were never worth testing.

So guesswork is not the opposite of rigor. It is the input to it. What matters is what happens next: whether the guess gets a number attached and a validation or an experiment run against it, or whether it simply becomes the decision.

Guessing is fine. Guessing without evidence at all is not.

3. AI can recommend the price, but a person still has to decide it

This is a frequent question: "Will AI take over pricing?" The answer is always no, and for good reason.

Take ERP systems, for instance. They have handled accounting for decades. Do companies still employ accountants and CFOs? They do. And when the numbers matter, who does the board trust — the ERP system or the CFO?

AI is like a fast Formula 1 car. Give the best car in the world to the worst driver in the world, and then wonder why you are losing the race. The car is not the problem. There is no driver.

Underneath the analogies is a point business people tend to skip: accountability. People are accountable. AI is not. If a price is wrong or great, you don't look at the AI system, but at the person who signed it off.

This is why human-AI collaboration is not important — it is mandatory. Not as a nice sentiment about keeping humans in the loop, but because someone has to own the decision.

What changes with human-AI collaboration is not who decides. It is what they decide with: evidence and a number, instead of intuition alone.

https://www.youtube.com/watch?v=KZnMU4VyoRM

The rest of the conversation

The episode ranged much wider, touching important topics such as:

What happens to seat-based pricing when AI handles the conversations and customers simply need fewer seats.
How token-based pricing relates to the value a customer actually receives.
The arrival of a real cost of goods sold in software — something SaaS has largely been able to ignore until now.
Outcome-based pricing as the next step beyond value-based pricing, and the risk that comes with standing behind a result.
What replacing a person with AI does to trust, reputation, and the customer on the other end of the call.
Whether SaaS business models survive this shift, as they survived the web and mobile before it.

An intriguing set of views, and refreshingly open about how much is still unsettled. None of us knows exactly how this plays out yet. What matters is what you do with an open question: attach a number to it and test it, rather than leaving it at instinct alone.