Major research firms publish extensive tracking of AI spending, adoption rates, and funding activity. A data set that would meaningfully improve how the market's real health is understood, and that doesn't currently exist in any systematic, widely cited form, is a tracked rate of AI project abandonment - how many initiated enterprise AI projects are quietly discontinued before reaching production, and why.
Why this gap matters
Spending and adoption figures capture initiatives while they're active, but say nothing about how many initiatives were started and later abandoned without ever delivering value - a number that would materially change how the reported spending and adoption figures should be interpreted if it were tracked and disclosed systematically.
Why no one currently tracks this well
Project abandonment is diffuse, happens quietly without a public announcement, and isn't something individual companies have strong incentive to disclose, since it reflects an unsuccessful initiative rather than a positive development worth publicizing. Building a reliable data set would require survey methodology specifically designed to capture failures, not just active projects, which is more sensitive and harder to get honest responses to than standard adoption surveys.
What this data would reveal if it existed
An accurate abandonment rate would give a much more complete picture of AI adoption's real success rate, and tracking it by industry, company size, and use case would likely reveal patterns about where AI initiatives most commonly fail before delivering value - genuinely useful information that's currently missing from the standard market research picture.
Why we think this is worth building
Until this data set exists, the current picture of AI market health, built entirely from active spending and adoption metrics, is systematically incomplete in a way that likely overstates how successful AI initiatives are on average.