Thinking out Loud
Once in a while, it’s hard not to stop and take stock of the things going on around our industry that don’t add up. Or if they do, they make me wonder. These are random thoughts, not all connected, and certainly in no particular order – starting with tokens…
- The more things change the more they stay the same. Are token charges really nothing more than cell phone minutes of yesteryear or the so-called connect-time changes legacy players like Dialog and CompuServe thrust upon us? Enterprise buyers dislike transactional pricing of any kind. At some point they have to disappear into the fabric of a solution so price predictability and budgets align.
In the words of one google search:
Token cost in AI is a usage-based pricing model where users pay for the number of "tokens"—units of text, images, or data—processed by a model, rather than a flat subscription fee. Tokens are roughly 3/4 of an English word, and costs apply to both input (prompts) and output (responses). [1, 2, 3]
Key Aspects of Token Costs
· Pricing Structure: Costs are typically measured per 1 million tokens, ranging from \(\$0.15\) to \(\$5.00\) for input and \(\$0.60\) to \(\$25.00\) for output. Output tokens usually cost 3–5x more than input tokens.
· Consumption-Based: Users pay for what they consume, making it essential for managing operational expenses in API-driven applications.
· Measurement: One token is approximately 4 characters or 0.75 words in English.
· Input vs. Output: Both the prompt you send (input) and the text the AI generates (output) are billed. [1, 2, 3, 4]
It’s nice to see proper footnotes and source in Google, and it’s always a miracle to get answers like this in half-a second in our modern age. Besides that, the notion of a token is as antiquated as Dialog. No enterprise on the planet is going to sustain one token being approximately 4 characters or .75 words in English. How do you even budget for that? Whoever fixes this first wins the race.
2. There needs to be a new book written called everything CEOs need to know about AI but were afraid to ask. CEO turnover is the highest it’s ever been. We seem to blame the brain drain on performance except that they are leaving high performing businesses at the same rate. Why is fresh thinking age related rather than results related. And the percentage of women CEOs has dropped again. Seems fresh thinking only comes from younger men. It isn’t clear that the next generation of CEOs will do any better. Who is making these decisions?
3. I am still wondering why with so much amazing information at our fingertips it’s still not easy for buyers to buy. Planning a quick trip to a new destination still requires Google searches, visits to Reddit, TripAdvisor, specialty sites that rate and rank places like Escapio, or Mr. & Mrs. Smith, and a review of links and articles at travel ‘magazines’ like Travel and Leisure, Conde Nast Traveler, and Skift.
And then we head offline to speak with a travel advisor after also going from all these aggregators to the actual website of the hotels themselves to make a final selection. Silicon Valley technocrats are trying to convince us our buying agents are going to do all this for us. And they will do a good part. It is still impossible to get some ‘bottom lines’ from anyone but a human.
4. Despite Questex finally getting their exit and a new market share top 3 leader in events in North America emerging with Apollo’s buy of it and Emerald, we stand by our words that events are not a moat.
They may be profitable and a good place for F2F in an otherwise digital world, but they do not scale, there is minimal loyalty (especially when looking at GenZ/Millennial patterns) and they take a boatload of repetitive work year over year over year.
I am happy for our industry colleagues who are getting valuations and good return on their effort after all these years. PE owners are not for the faint of heart and there is a special kind of CEO with a special place in my heart who can work with and for them successfully. They really are not about long-term sustainable value, so moats aren’t really on their minds. They say it to ensure the next six years are secure but as far as building companies that last? Let’s just say there is Another Way.
As for our making sense of the non-sense? We’re publishing our analysis about the Apollo transaction next week. Stay tuned.
We are embarking on original survey research in partnership with an industry expert looking at how buyers buy in an AI era and how and what they’re using the tools for. Don’t miss it when it comes out.
We’re hosting CEO Network meetings to give them confidential access to one another and we’re programming the Outsell CEO Summit for October 1, 2026 (mark your calendars!) to answer some of those hard questions.
As for tokens? I’ve said my piece. First company to solve for them wins the AI race.
Have a wonderful weekend.