I find this to be a questionable conclusion. Is it true? Maybe, but the data as published does not support it. The survey doesn't correlate the quantity of gems spent with the cosmetic purchase category. Buyer numbers mean nothing compared to gem quantity. It must be considered that a large portion of the 51.1% low-end gem buyers (Below 800) might possibly overlap with the Services/Upgrades categories.
It was a fascinating read but the conclusion as stated suggests interpretation bias. The data could better back the conclusion if you show a direct correlation or if you separate one-time service and upgrade purchases that represent a very small portion of the overall gem usage.