How do you measure social media ROI on Twitter? A ReTweetability Index?

I was asked a “quick question” about my Twitter use yesterday:

How many user responses do you get from your tweets?

It’s a fair question. As soon as I started thinking about how to answer it, I realized how many dimensions a proper answer should measure and compute that number.

What constitutes  a user response?

A @reply?

A RT?

A FWD using a (via)?

A HT (hat tip)?

A @mention?

There’s “influencer marketing ” metrics to be considered in there too, like whether users can be driven to comment or watch something elsewhere.

If you accounted for each metric on a given tweet, what measurements for the ROI of Twitter use could you generate?

If you’re measuring click traffic, you can see the traffic for @digiphile at bit.ly. That includes links that have been shared on three different Twitter accounts: @digipile, @epicureanist & @ITCompliance. The reach of the first account is dramatically greater, so results vary widely.

In aggregate, my qualitative answer to the original questions has to be:

“It depends.”

There are so many other variables: when I tweet, what I tweet, whether there is a link, if it’s directed to another user or if it includes an identifying source for a link.

Dan Zarrella, a social media and marketing scientist, has been at the forefront of retweet research focused on offering other quantitative models for measuring ROI.

His newly published Retweetability Index ranks Twitterers by his own formula:

[ Retweets Per Day / In(Tweets Per Day) ] / In(Followers)

I’m ranked at 1919 today with an index in the 7000s. I think that means I get a modest amount of user response. For more information, check out what Jennifer Grove ‘s post on Zarrella’s  ReTweetability Index at Mashable.

I’ll ask “How do you measure social media ROI on Twitter?” today and see what other people think.

In the future, I hope Dan and other researchers will create and share formulas or indices that include @replies, FWD, (via),  HTs, @mentions & influencer variables.