Some recently published research concludes that 71% of tweets are ignored. By 'ignored' the researchers mean tweets that do not get a reply or retweet. However, this figure does not allow for tweets that are read but do not initiate any response. Not every tweet requires a reply or retweet to make it worthwhile.

Working out the percentage of tweets that are read poses several problems. Reading a tweet is a passive action that leaves no evidence. The only way to find out if somebody has read a tweet is to physically watch them or ask them. I have come to the conclusion that we may never have a definitive answer. There are just too many variables. A library could tell you how many books it leant out last month, but could it tell you how many sentences from those books were actually read and comprehended?

There are many factors that could affect any statistics:

• Read but ignore - A percentage of users who read a tweet that asks that they reply if they have read it will just ignore it.
• Tweet fatigue - Twitter is all about the present, memes come and go and become stale very quickly. If I wanted to do a thorough test of how many of my tweets are read and kept on sending out a tweet saying "Did you read this tweet? Pls reply" my followers would soon tire of it and the reply rate would go down, thus effecting the figures.
• Time of day - A tweet sent at lunchtime is more likely to be read than one sent in the middle of the night. This does, however, depend on what time zone most of your followers are in.
• Number of followers - A tweet sent by someone with thousands of followers is more likely to be read than one sent by a user with a few dozen.
• Account activity - A tweet from a popular interactive account is, obviously, more likely to be read than one from an account that only tweets once a month.
• Spam - I thought of creating a hashtag that people could use to report the number of responses that they got to a tweet, but soon realised that it would be subject to spam along the lines of "I got 110% replies to my free laptop offer."
• Wording - The tweet itself could have an effect on the results. Subtle differences in wording can make a big difference online. Tweets with good 'call to action' marketing text would be more likely to get a response.
• What does 'read' mean - There is a distinction between 'seen' and 'read' tweets. You could, for example, try to work out the number of tweets read by monitoring the twitter streams consumed by twitter clients like TweetDeck. However, just because somebody has TweetDeck sitting open on a PC all day does not mean they are reading all those tweets. They may glance at many, but only home in on a select few. The human eye is very good at scanning.

It turns out that all tweets are equal, but some are more equal than others. To calculate a realistic percentage of tweets read figure would probably involve a lot of work for a dedicated research company or academics: Working out how many tweets are tweeted hour by hour. Selecting twitter users at random. Watching how often they go on twitter. Measuring how many tweets they read in their timeline. Crunching the numbers.

If it can't be easily done in a scientific way, let's try an ad-hoc straw poll. So, I tried a single tweet, early in the morning, asking people to reply if they had read it. I got replies, thirteen in all. So my tweet was read and it was read by 0.41% of my followers. I then tweeted and asked others to post a tweet with the same text and see what response they got. Here is the result:

Of course, these figures themselves may be skewed as you could argue that they are the percentage of tweets read that are posted by proactive users who replied to me. And you'd probably be right. Lies, damn lies and statistics.