**The most counterintuitive finding is that spending more on AI buys you nothing measurable. Time spent using it is what moves the outcomes.** I computed pairwise correlations after parsing the text-typed numeric columns. Each coefficient covers the 271 to 288 rows (of 300) where both values parsed. **Monthly AI cost is essentially uncorrelated with every outcome:** - Productivity score: r ≈ 0.013 - Satisfaction score: r ≈ 0.009 - Accuracy rating: r ≈ 0.008 - Time saved per week: r ≈ 0.055 - Monthly income: r ≈ -0.05, so higher earners don't pay more either. **Usage hours show the opposite pattern:** - Daily AI usage hours vs. weekly time saved: r ≈ 0.76 - Daily AI usage hours vs. productivity: r ≈ 0.64 - Daily AI usage hours vs. satisfaction: r ≈ 0.40 - Time saved vs. productivity: r ≈ 0.76 - Productivity vs. satisfaction: r ≈ 0.65 A paid tier is usually assumed to deliver better results. Here the people who report the biggest gains are the ones who use AI the most hours, regardless of what they pay. A free user and a heavy spender with the same usage hours look the same on productivity and satisfaction. Age (r ≈ -0.06) and reported accuracy also have almost no link to productivity or time saved. Accuracy tracks satisfaction only moderately (r ≈ 0.53). **Caveats:** These are correlations in observational survey data, so they don't prove cause. The cost column may also hide a split between free users (zero cost) and paid users, which I didn't examine. Still, a correlation of about 0.01 means there is no linear cost–outcome relationship in these 300 respondents. This argues against using subscription spend as a proxy for value. Usage depth is the better signal for rollout and tool-adoption decisions. The figures come from the two correlation queries already run (step-0 and step-1 above).