October 6, 2009

Twitter does even less than you thought

First, for Twitter messages relating to some product or brand, more are about facts than feelings. So, rather than being the "pulse of the planet," it is turning more into a Yahoo! Answers service, just with faster response rates. It's doubtful that anyone would pay to use Twitter for this purpose since you can use Google, etc. right now, and those responses are already pretty quick. You'd have to bet on there being a large enough group of people who would want information right away, rather than take a minute or two using Google. Maybe, but I highly doubt it.

Ads won't work long-term. They hardly earn money in the first place, except for companies like Google who get a decent buck from advertisers who want their product to show up when someone does a relevant search. The difference is that whereas Google has developed an algorithm for returning relevant results, Twitter doesn't provide a list of results at all -- the users who answer the question do. These users would receive none of the ad revenue that Twitter made from search-based ads. Of course, most of the people on there are completely insane, so they might not notice. Or they may be so pathetic that they'd work for free just to obtain their own slice of internet immortality, like those prolific Amazon reviewers.

The only possibility would be to charge lots for ads like Google does for its search-based ads, and give some portion of that to the users who answer questions. That seems like an awful mess, though, because then every user has an incentive to respond -- no matter how vapidly -- just to say that they were part of the response that the question-poser was looking for, and whose eyeballs the advertiser wanted to target with the search-based ad.

Trying to figure out which users really were providing useful information would be a nightmare. They couldn't hire enough people to look through every response to every question, so they'd have to use something like a customer rating system. But again, to figure out who among all responders merited a cut of the ad revenue, the question-poser would have to rate every single responder. In reality, they would likely give a high rating to one of the good responders and not bother rating any of the others, even the good ones.

They'd have to turn to something like About.com with its experts who field questions. That way, the responder can build up a reputation and assuage some of the worry that an asker might have, and a customer rating system (to determine the cut of ad revenue) would be easy with just one person responding. Of course, this is just another variant on all previous forms of pay-for-info services. It would look nothing like what it does right now, where anyone can write about anything to anyone else. There's an obvious parallel between Twitter's likely path and that of Wikipedia, where participation must ultimately be fettered in order to ensure quality -- and certainly to get paid, if they wanted to really improve their site.

Related to this is the uselessness to marketers who want to know how the public perceives their product. It's bad enough that most people don't share their feelings, but even when they do, they are almost uniformly positive, and all across the internet. Of course, that means the marketers can't learn shit from Twitter's pulse -- they get a highly biased estimate from self-selected fanboys who, after opening up their new toy, immediately rush to their MacBook to send off a gushing Twitter message, typing with one hand.

It's an odd finding that those who hate what they've bought tend not to register their complaints online -- or in real life, as the WSJ article says that most word-of-mouth is positive too. But writing detailed negative reviews, or ranting on and on in person, only reminds you how badly suckered you were, and most people would rather not dwell on getting hosed big-time. Owning something makes you think more highly of it.

The only case where people could get useful feedback from customer reviews are where the rater has no stake in the thing being rated, which eliminates almost all reviews by people who own the thing. Take Hot or Not -- the people whose looks you're rating aren't related to you, aren't your friends or partners, and won't ever run into you. Thus, raters are perfectly clear-headed when judging looks and tend to give low or middling scores. (If memory serves, Hot or Not actually "corrected" for this tendency by artificially inflating scores in Lake Wobegon style. If you noticed all those fugly dogs who scored 6 or 7, that's why.)

So if you need to know how good-looking you are (or aren't), you can profit from internet ratings. If you're a marketer who wants to know what the broad public thinks, you'll have to go back to test panels or focus groups where the raters don't have a personal stake in the rating. Investing in Twitter will only hasten your descent into cluelessness.

1 comment:

  1. I didn't realize Hot or Not "corrected" scores but it all makes sense now. I always wondered why my ratings were so out of line. That is pretty bold of them to use percentiles as if the people who post on their site represent an accurate cross-section of the population.


You MUST enter a nickname with the "Name/URL" option if you're not signed in. We can't follow who is saying what if everyone is "Anonymous."