14 Comments
User's avatar
Laura Ferraz Baick's avatar

The part about cheaper models not fixing the math is so true!!! We thought GPT-4 → 3.5 was going to save us. Instead, we just shipped five new features that made every workflow 3x heavier. Customers loved it. Our P&L did not.

Chris Tottman's avatar

Hard earned experience 🥵 so thanks for sharing. People need a Moore's Law on capability 📈 and costs 📉

Zain Haseeb's avatar

"Metered margin business pretending to be SaaS" that's going to stick with me. Great read Chris!

The question I keep returning to: at what point does the metered margin reality force a fundamental rethink of AI product pricing models? Are we heading toward outcome-based pricing as the only sustainable structure?

Chris Tottman's avatar

I'm uncertain but at some point Moore's Law kicks in ie it's 10x better out of the box BUT also at the moment the usage is being financed by investors (ie burning unprecedented amounts of cash) - at some point (in years)- AI is much better & we start paying more or competiton and low defensibility means mass competition and lower prices. Like internet connectivity in the late 90s. Started with a 200k ARR leased line in 1995 and ended 2000 with that costing less than 5k ARR... No that's zero.. 😕

David Jackson's avatar

Chris, you say "Start by defining your unit of value. Not tokens. Not calls. Something the customer recognises.

Customers recognise results that matter to them.

Customer results are not just a key result to measure, it is also the best basis for aligning GTM and real alignment has real benefits on unit economics - including GM. Figure out the results that matter to key customer roles and that becomes the basis of what marketing communicates; what sales sell; what services enable and, most importantly what the product delivers.

A leadership scorecard that tracks both key unit economics and key customer results is part of a customer results aligned GTM approach.

Chris Tottman's avatar

Fantastic advice and wisdom David - thanks for sharing. I'm Publishing 3x per week for founders and operators. You'll see hundreds of assets in the feeds you can enjoy and repurpose. Thanks for coming by and commenting. ✨

John Brewton's avatar

Control shows up in design decisions long before it shows up in numbers.

Chris Tottman's avatar

💯 true John - thanks for sharing your wisdom as always ✨

Peter Stansbury's avatar

Great assessment of a high risk, it can feel like a gold rush, you can't afford to get left behind. But many people are going to be crushed in the stampede.

Chris Tottman's avatar

Washed away by the wave 🌊

Taisiya Kudashkina's avatar

You can't optimize your way out of structural dependency. Either you own the economics or you're permanently renting profitability from OpenAI/Anthropic and hoping they stay generous.

Chris Tottman's avatar

💯 true. Much like your business sales surviving on paid ads on Meta. At some point everyone is working for Meta 😢

Vvinit Khichaa's avatar

Such an important aspect of scaling an apps layer AI business - very well explained on how initial enthusiasm can lead to severe negative consequences down the line despite a well-running product!

Chris Tottman's avatar

Thanks for sharing your thoughts and wisdom Vvinit 🌞