Nearly 27 years after it began life measuring the
audiences of online media, Comscore this morning issued its first-ever report on the next new
media disruptor -- AI -- and its initial finding may make the established digital "interactive' media green with envy, mainly because generative AI has proven to actually be interactive.
How interactive? Well, Comscore benchmarks the Big 3 generative AI platforms effective with March 2026 estimates with multiple prompts per conversation -- ranging from an average of 4.6 for
Google's Gemini to 4.9 for OpenAI's ChatGPT to 7.1 for Microsoft's Copilot.
The Comscore report doesn't explain those prompt multiples, but thinking about it makes sense to me, because in my
experience Gemini already is the most intuitive, since it has years of knowledge about users that have been using Google products for the longest, while ChatGPT is only a few years old.
advertisement
advertisement
As for
Copilot's prompt delta, that makes sense too, and it's all in the name -- Copilot is framed as a collaboration tool more than the others, which are premised more on being answer engines.
How
this will evolve over time is anybody's guess, but at least we now have a benchmark thanks to Comscore, which also provided us with share of prompt volume data to go along with it.
For an
industry -- you know, advertising -- that has long thought in terms of audience share, the shift to prompt share may take some getting used to, but I think the real value is in gauging the level of
interactivity among the major chatbots.
Of course, you could look at it another way -- in the inverse -- which is that the most optimal chatbot is the one that fulfills your information need
with the least amount of prompting. And for me, at least so far, it's been Gemini. And that's because I just assume it gets me.
That said, the Comscore data shows Gemini gaining the greatest
share of prompt volume among the Big 3, year-over-year, but I think that has more to do with increased usage than the underlying need to prompt.
