A version of this post originally appeared on TVbytheNumbers on Oct. 15.

I praised NBC earlier in the year for the experiment with TAMi — the total audience measurement index that seeks to measure the total viewing of shows on various platforms and not just television. While I do appreciate the attempt, I am no longer finding it praise-worthy and find it mostly useless because the comparisons and numbers used are apples-to-pears or apples-to-grapes style comparisons at best, and apples-to-orange comparisons at worst.

We think it’s fantastic that NBC is measuring and reporting numbers involving newer technologies, and we applaud them for that. But we wish they’d have reported based on total engagement so the numbers and relative comparisons would’ve actually been useful.

The standard Nielsen audience average numbers reported by us and others are engagement based. It’s measuring total minutes of viewing and dividing it by the duration of a show. If there are 600 million minutes of viewing for a one hour show, that show will show an average of 10 million viewers. Engagement-based measurement compared to engagement-based measurement is the only really worthwhile way to go.

Unfortunately TAMi doesn’t use this, not even for television, where it uses the “total audience” Nielsen measurement. These are people who watched at least 6 minutes of a show, but not the whole show and there is no way to garner total engagement from such numbers.

Similarly, the Internet streaming counts are bogus. For one, if a single episode is broken up into six streams, it counts as six streams, even if it was only one episode. Further, streams are counted whether they are completely viewed or not. There’s no way to derive the total engagement from that (total minutes of viewing divided by episode length) and so there’s no basis to compare them to anything useful.

Because it’s not truly an engagement based measurement, it’s of little real value. Except for this: Finally we have numbers from iTunes. And guess what? The numbers are tiny — even when they were giving away shows. Oh sure, there were more than 10 times as many downloads for the free episodes NBC gave away on iTunes (which leads us to believe downloads on Amazon Unbox are likely in the hundreds, if not the tens, since they are included as well).

The most downloaded episode surprisingly (at least to me) was the first (and free) episode of Knight Rider. Heroes is really the most downloaded show and after the freebie (which got 58,886 downloads) the second and third episode dropped to less than 20,000 downloads. So please excuse us if the next time we hear someone chime in about a show being No. X on iTunes and acting like it makes some significant difference and we tell them to STFU.

Again, because the data isn’t engagement-based focusing on total minutes, either for television or the streaming data, it’s mostly useless, but for including the download numbers we thank NBC profusely! The bottom line is, TV is still king, though streaming is growing. But, unless and until they provide minutes of streaming, it will be difficult to get a true idea of what portion of viewing of shows happens via sanctioned streaming channels.

You can download the full slide show (PDF) with data for all shows, but note some data for certain shows is missing. For example there is no download or mobile data for Chuck, even though it is available on iTunes.

Robert Seidman co-edits the blog TVbytheNumbers.com which focuses on television metrics.