- the application of statistics and mathematical analysis to a field of study.
- a combining form with the meaning “the science of measuring,” that specified by the initial element: biometrics; econometrics.
It sounds like a science, but it is really an art.
Often times, when we are discussing metrics (or web analytics) we are talking about more than just numbers, raw measurement or statistics. Metrics involve the interpretation of of data.
Look at google analytics and you will see the type of data that most people think of when you say [web] “metrics:” Visits, Unique Visitors, Pageviews, Pages/Visit, Avg. Time on Site, Bounce Rate, New Visits, Location, Language, Network, Traffic Sources, Site Speed, Searches, Sales, etc.
This type of information can be helpful on its own, but it is largely one-dimensional. Raw statistics loose significance without context. Good metrics are defined in terms of strategy. What is our goal and what kind of specific statistics indicate success? A statistic like unique visits may be less important than net sales for a business like CustomInk.com that is a totally online operation. The opposite may be true if you are Coca-Cola and your website is more for branding purposes, not sales.
To add dimension to the numbers, metrics can also be constructed in the form of an equation or an aggregation of data. These analytics express valuable but subjective concepts such as loyalty, engagement, and virality. Take Facebook Insights relatively recent introduction of two new metrics: Weekly Total Reach and People Talking About This. Total reach refers to the number of unique individuals who saw any content related to your page. People Talking About This combines all likes, posts, check-ins, mentions, etc.
These simplified metrics reflect data that Facebook considers important. However, it is always important to dig deeper in order avoid “measurement inversion.” This is when metrics seem to emphasize what organizations find immediately measurable — even if those are low value — and tend to ignore high value measurements simply because they seem harder to measure (whether they are or not). For example, while Facebook is more interested in measuing the overall “conversation” surrounding a given page, an individual business may be more interested in investigating a particular element, such as check-ins, if that metrics relates to an ongoing promotion.
Today, metrics are evolving quickly. Batch metrics (collected daily, hourly, etc.) were once the standard. Now, many companies demand real-time metrics, especially when it comes to social media. Advertising metrics that drive much of the value online are constantly being tweaked in an attempt to more accurately reflect the true worth of a given ad. Code metrics that calculate how efficient a program or script is running can get very complicated but are essential for optimizing web performance.
For our purposes, we will most likely be dealing with metrics in the social graph. You can think of it as metrics 2.0 [or, even 3.0] . The key distinction between basic web analytics and metics in the social graph are relationships. How are things (both “individuals” and “objects”) related to one another ? These types of interrelationships can be conceptualized by sociograms and emphasize choices and preferences.
Social network analysis software (SNA software) facilitates both quantitative and qualitative analysis of social networks by describing features of a network, either through numerical or visual representation. We now have much more data than we know what to do with. Creatively identifying how to interpret the information is the tricky part. This is why we often see larger, more established internet companies buying up analytics start-ups who have an interesting twist on interpreting different types of data.
The first step for each of our companies is to identify a goal. Next, we must find out what data is available and investigate what types of relationships can indicate success.
And always remember, if you don’t measure it, you cannot optimize it.