Written by Nick Heer.

Canalys’ Methodology

Me, quoting Canalys’ only reference to their methodology on their website:

Each of our services benefits from the same strict methodology, with clear, global definitions and detailed, country-level data published on the same day – no preliminary data or forecasts.

If they’re not using preliminary data or forecasts, they must have direct industry connections (or are using estimates, which they don’t specify). Strategy Analytics pegged the iPad’s market share in Q2 at 28%, while Canalys claimed it was 43% — a significant difference. I don’t trust either report, but why such disparity?

Now I know. Pin-Chen Tang, an analyst with Canalys, shared some of their techniques via email:

The primary research technique used is to approach the product vendors, which then share their quarterly and annual shipment information with Canalys. Canalys requests detail regarding shipments across the various countries and product types. Vendors often provide this information to Canalys in electronic format and submit it via e-mail. […]

Despite the benefits, some vendors choose to operate a less than fully open policy. They may prefer to use a method based on the sharing of estimates, thus increasing the resources required by both Canalys and the vendor concerned. […]

A few vendors refuse to cooperate with market analysts. These are often, but not exclusively, privately owned companies unfamiliar with the duties of disclosure that are common in publicly quoted companies.

Some companies will reveal information to Canalys, but not to shareholders — why do I doubt that? I think the second and third types are much more frequently encountered (may I remind you, again, that the biggest non-Apple tablet brands don’t share their sales figures publicly). So, like Strategy Analytics, Canalys uses estimation:

All feedback provided to Canalys is kept in the strictest confidence. Canalys will not reveal the extent to which vendors have been cooperative. All results Canalys publishes are described as ‘estimates’ regardless of the actual data source. This process guarantees the impartiality of the results.

Golly. I wonder who that benefits.