The engineer speaking: “What you cannot measure does not exist.”
It’s hard enough when someone else is telling you this, imagine how hard it is for me, having my marketeer brain argue with my engineer brain (half of each, don’t worry)! However, the marketeer halfbrain is doomed, once SoMe exit the fog of experimental pilot projects, a metric for success is needed, if nothing else to justify the investment that will be pouring into it.
Most recently my friend Oleg asked the question that most CMOs are starting to deal with, i.e. is there a unique metric for Social Media success?
I have already discussed a neat idea by Sociagility, whose success will ultimately depend on how widely it is accepted, but while it is a good attempt at capturing the “softer” (and elusive) aspect of who’s really influencing whom, I am more interested in a measurement model that I can apply to a standard lead generation problem.
The interaction model
Let’s look at the very simple process described in this scheme:
*retweets, comments, linkbacks
We are concerned with the overall yield of the sequence, i.e. how many “actions” (such as retweets, comments or likes) our post will generate, but in looking at this picture it is clear that the overall average yield is the product of the average yields for each step.
Of course, the model needed to represent your project could be much more complex, but what is essential is breaking it down to bits where you can measure both input and output easily, therefore making yield calculation easy.
In the simple diagram above we have also included a trivial first step, i.e. tweeting the post; as we can automate the process of generating a tweet every time we post, and can set its yield = 1 and ignore it for the rest of our considerations. These are however separate actions, so your model may be different.
ηBC, or the art of Writing Interesting Tweets
Anybody who has spent some time on Twitter knows that people do not follow every link that is thrown their way; quite the opposite, users normally focus and act on a fraction of what the receive on their streams, regardless of how carefully selected is the list of people contributing to it and quite likely regardless of how large is the number of people they follow (according to Hubspot as of january 2010, the average user follows about 170 people).
To improve this conversion rate you must carefully edit your tweets, and given that most autoweeting features of blogging platforms pick the title as the body copy of the tweet, a simple way to do this is to think of the title of your post as you would do of the tweet itself.
Among the most common tricks:
- use hashtags, especially if you’re chipping in to an ongoing debate; this way your contribution will also be seen by people who do not follow you, but follow the hashtag
- strong, provocative titles are more likely to attract attention than bland ones
- very short titles are a waste of real estate – the platform will make sure you don’t go overboard, but remember you’re trying to get people to read more
- don’t forget your 140 characters will be appended with the short link, typically 12-long, so think in 130 chars chunk
ηBC, or the art of Writing Timely Tweets
When to tweet is not indifferent. Some people tend to forget that, unlike a post which will always be there whenever a user checks out your blog, a tweet is ephemeral and stays on any of your followers’ stream only for the time it takes to be followed by twenty others and be pushed out of the screen, regardless if the user is logged on to Twitter or not.
Hubspot’ Dan Zarrella has this excellent presentation where he explains how the best results are achieved around 10am and 5pm – but of course, bear in mind the timezone of your intended audience if it’s different from the one you’re tweeting from.
Finally, it may be worth considering a Twitter scheduler like Twaitter; repeating tweets is usually not an issue because most people who follow you will only see one or two tweets per day, but careful not to go overboard lest you trigger unfollows. By trial and error I have found between four and six times a day are the best compromise for a campaign, and you might want to actually break down your campaign into two or three similar but not identical tweets repeated at 12-hours intervals.
ηCD, or the art of Stimulating Reactions
How do you start a discussion? Boy, I wish I knew that. Asking questions is a common way, and so is posting data and asking for comments. However I’d like to put comments in perspective.
Commenting to a post is something you do normally when you are in strong disagreement – there was a time when the only way to reply, expand or add to a post was a comment under the post itself; otherwise (and much more commonly) you could go to a public forum and start or participate in a much more peer-to-peer fashion to any of the discussions there.
I, like many people, always feel a bit of a guest when I comment on other people’ blogs, and I think this has a lot to do with the general reluctance to react this way. Of course A-list bloggers are an exception: anything Robert Scoble posts will elicit tens when not hudreds of comments, but when you read them, quite a few smell of linkbaiting for traffic. Which is reasonable: if you feel like contributing, you obviously want for your contribution to be seen by as many people as possible and – unlike you’re Scoble – your blog will always attract less eyeballs than, say, Twitter or Facebook which is why I bundle all form of actions into the same bucket.
In general, I have found that asking your readers to react without any incentive requires a very controversial topic or statement, which is normally a bt difficult for a brand; on the other hand, I have been surprised at the effect even a little incentive has on response rates; book giveaways, draft and lotteries all drive response rates through the roof and while one should not abuse them, their effective cost is minuscule compared to traditional advertising.
Incentive-driven participation also has the advantage of allowing us to ask users for opting in to our reach out activities, so while 1,000 comments would flatter our ego, 100 opted-in profiles may be worth more in lead generation mode.
There are so many measurement tools out there that I won’t even attempt to list them: Sysomos, Radian6, Meltwater just to name a few have all comprehensive tracking dashboards and coupled with Facebook and Google Analytics as well as YouTube Insights, they provide you with more data than you need; as usual, the problem is not so much getting more data, but interpreting it, so make sure you don’t bite more than you can chew.
One side word for the tools that allow you to profile influencers.
Hootsuite gives static information in a nice and compact form about any tweep you follow or that appears in your stream; if you want to get a little deeper and back in time, I recommend Tweetstats which also analyzes the time pattern of the tweep in question, allowing you a smart scheduling for example to get noticed and followed by one of these influencers.
My current favourite however is Simply Measured, which seems to have a keen eye on conversion rates across different Social Media platforms.