I am far more interested in what percentage fail because of a breakdown in the founder relationship.
It's easy to start a company and get funded, when you have a great team. It's very difficult to maintain that relationship for a decade or more.
Yes it will be cool when we can see stats like "co-founder matching has created X startups, and you're Y% more likely to succeed." Those stats take time to reach significance. For now, it's OK for it to ride on vision + early indicators like "# of startups founded".
In my experience this happens far more often than issues with tech, but less frequently than running out of cash due to failure in product market fit for pure software plays.
Hardware companies tend to be rough for the founders, unless they are extremely savvy at raising funds.
The longer a start-up remains in the start-up phase the bigger the chance for co-founder conflicts due to differences in expectations and changing life circumstances.