Measuring Engagement of Social Media websites in a Web 2.0 world

Right after lunch, and back in to the thick of it. Peter O’Neill knows his stuff, starting his presentation on web analytics with credentials including Yes, the title’s buzzword heavy - and this is a large room with plenty of filled chairs.

It’s clear that nobody likes the term Web 2.0; but as a marketting buzzword it’s inevitable, essential and easy to sell. For Peter, content created directly for the online scope falls in to this category. Another hot buzzword is Engagement, and this is where Web Analytics shines. Are people interested? Are they coming back? How long do they stay? It seems obvious that if you can measure these indicators of interest, then you can measure customer satisfaction, which leads to loyalty and thus to revenue.

All of these assumptions are based on the age-old idea that loyal customers are good customers. Which metrics can tell you this? Surprisingly, the answer is not clear cut - despite the size of the online advertising industry built on these principles.

The main interactions a user has include viewing pages, making purchases, downloading a file, registering. Some pages are more important than others - who is looking at your Contact Us page? Who is writing comments on your products, or sharing things with friends? Who’s throwing sheep at each other, and should you measure that as a statistic?

Sheep thrown per day per person is a simple but effective demonstrator of Facebook’s engagement, but does not suit every website. Which metrics do suit depend heavily on your business objectives. For personal blogs, the writer is often interest in who wants to contact them for work, say. Other websites are sales driven, whether that’s through ad-revenue where unique visitors rule, or sponsorships where clickthroughs and relevancy are highly important. For some websites the focus is niche and a small number of loyal visitors is desirable.

Peter pulls up the example of STA Travel where what’s important are which actions are most likely to lead to a purchase. These may include people who’ve downloaded the tools, subscribed to the newsletter or read the travel tips - these people are already engaged and worth focusing on. The suggestion is that comparing compelling site features with acquisition rates is key to measuring the success of a feature, but if you flip that around, it shows you which features lead you to the best success at achieving your aims (and therefore where to spend your budget). Web Analytics can facilitate data driven design - and that’s more valuable than a designer telling you how pretty the site is.

On to another, familiar example: Bebo. I expected to hear the time-old story of how Bebo changed their focus to UK Teenagers, despite a generic social network starting point, due to following their demographic data. But the current hot topic seems to be ‘The Gap Year’, a sponsorship driven reality TV tie-in, sponsors including everyone from Canon to Colgate, and keeping such sponsors requires demonstrating a return on investment. But it’s a delicate balancing act - you can’t push the products too hard, or people will fall-back; but too little, and it’s difficult to show how the money is coming in versus regular revenue on other parts of the site.

I’ve worked for a number of companies who develop websites and analytics is often an afterthought, especially at agencies who roll out a large number of sites for a large number of customers. Referencing “Web Analytics Demystified” by Eric Peterson, Peter shows us a controversial and unweildy formula for measuring engagement - and it seems to serve only to confuse. Peter agrees, and this is an on-going topic of discussion. Nevertheless, the metrics used by the formula are paramount, and every website trying to make money would do well to put some thought into the capture and reporting of these.

Peter offers us 5 simple steps to metric-success. First, define your business objectives. Then define which actions on the site can lead to achieving them. Ensure these actions are tagged - that is, log them - and make sure you do that right the first time! Track their performance over time, as data is not and never will be 100% accurate. Finally, track the correlation between product features, traffic sources and that mystical paragon alluded to in the topic - engagement levels.

There are some things which contribute to engagement offline that are difficult to measure - imagine your web-advert going viral; no web logs to look at if it’s being re-posted to YouTube. You can, in this case, do the same thing that companies have been doing for years, and resort to regular offline market research. But whilst online, measure as much as you can, all the time, right from the word go - Google Analytics, Omniture, or any (every?) other way you can. Other useful tools in a Web 2.0 world are the obvious Feedburner, and less obvious Tweetburner and Facebook Lexicon - and of course good old Google News.

What do I take away from this? Tracking your engagement is hard to do right, unreliable and easy to do wrong, very, very important, and, with special relevance to Social Media, it’s important to remember that there are so many new ways to interact with a site. Make sure your design decisions are data-driven, because your purse-strings will demand it!

Tags: , , , , ,

This entry was posted on Saturday, October 4th, 2008 at 1:14 pm and is filed under Sessions. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Leave a Reply