Many Web Analytics Metrics and Ways to Measure

Screen Shot 2015-01-29 at 4.06.21 PMOur third Analytics class focused on the various types of website traffic data points and ways to measure them, both qualitatively and quantitatively.

I learned that you can measure a vast quantity of web data points or metrics. I think it helps to try to remember them by categories, which include: conversions (rate, by source..), traffic ‘referral’ sources, geographic location, visits, visitors (new, return, unique…), time on page, time on site, bounce rate, exit rate, engagement and purchase habits (cart abandonment, days/visits to purchase…).

It was also worthwhile learning that analytics aren’t infallible and that loopholes exist. For example, a visitor may spend five minutes on a page but if they don’t go to a new page, their visit is tracked as a ‘bounce,’ which is inaccurately perceived as negative.

I think a good point to remember from the class is also that the impact or positive/negative attribute of a metric varies according to the type of site and its purpose.  For example, for a service business site, like Sofia’s, you want first time visitors who scan the site, exit on the contact page and follow through to enquire about purchasing a service.  Repeat visitors who never follow through offer negligible value.  In contrast, for a social network (like the like the caregivers one I’m proposing for my senior project), repeat visitors are critical for the site’s longevity; you also want them to come to the site, ‘engage’ (by posting or responding to a post) and leave, most likely without going to the contact page. Ideally, you want repeat visits from those who’ve signed on as members (i.e., converted) but even those who check the site a few times before committing are ok, particularly as the site evolves.

As for ways to measure web analytics, we learned about various methods. It sounds like one of the key ways to quantitatively measure analytics is using JavaScript tags. I’ve used this method to measure another blog by embedding the tag via a CloudFare workaround. Unfortunately, I haven’t blogged much since I set this up in Oct. 2013. However, my page views for Nov. 2013 were 1.47, which doesn’t sound as bad when I consider that page views are not as relevant on a blog because people usually visit the most recent post and bypass the archives. We also learned about various qualitative measurement methods, including heuristic evaluations, site visits, usability testing, surveys, web enabled research, experience testing and collecting competitive intelligence data (e.g. panel based measurement, similar to AC Nielsen for  TV).

The we did an in class exercise to clarify our understanding and from this I learned more about methods.  My takeaways from this assignment, (which I ended up presenting) were:

  • When trying to determine a website’s goal, always search for ‘how’ it is making revenue (unless it’s clearly run by a not-for-profit or clearly funded by another source).
  • Be very specific when citing KPIs or metrics. For example, don’t just say “measure return on investment” but clarify ‘what’ you are measuring (e.g. orders), against what (e.g. operational costs) and how frequently (e.g. weekly).
  • Heuristic Evaluations, such as Site Visits and Remote Testing, do not usually have a sufficient number of participants to be statistically significant, even if you are testing a metric that can be quantified.

KPIs and Metrics: The Signposts and Footsteps to Success

In our second Analytics class, Sofia introduced the value of Key Performance Indicators (KPIs) and Metrics to a project’s success.

My understanding from the class is that KPIs are like signposts. That is, they are: important measures that show how your project is progressing toward achieving its SMART goals with business implications and can inform course corrective actions, where required. We also learned that KPIs are specific to the project and as such, the number of KPIs varies but should be  enough ‘to do the job.’ This is a good way to assess if you have enough KPIs but still challenging, as I think it’s easy to be over zealous when you’re just starting.

An important KPI criteria that stood out to me is that they should deliver useful information in a timely manner, ideally in less than two months, unless you’re in a slower industry. For example, if you have an established, broad appeal, social network that promotes and sells premium services, a KPI might be your cost per order/sale and can be attained within a month.  In contrast, if you have a new social network in a niche market, (e.g. the social network for caregivers of senior citizen family member I’m considering for my senior project), it may take two to three months to attain this KPI.

Depending on the KPIs selected, I thought it noteworthy that part of the task may include defining key components, such as valuable exits or successful events, tailored to your project. For each example, you need to define ‘what’  a user needs to do before they leave the site, to assess whether the visit exit is valuable or the event successful.

Since KPIs have a major impact on the business, such as impacting revenue, costs or conversions, it makes sense to reference them when reporting up to internal executives or clients. To this end, Sofia discussed the importance of segmenting your selected KPIs into custom reports, possibly separate ones, for reporting to executives, as well as project team members.

I also found it interesting to learn about KPI’s various categories, including: Actionable Outcome KPIs; Calculated KPIs; Engagement KPIs; and Business KPIs. There are also Social Media KPIs and Conversion KPIs but many of these measures are ‘Metrics’ and not KPIs. Metrics, Sofia explained, are timely qualitative or quantitative data points that help inform your strategy but don’t directly impact the business. They remind me of footprints.

We discussed social media measures, such as the number of people accessing a site via a social network, and while important, they’re rarely KPIs. The reason is visits via social links just offer more ‘opportunity’ for people to consider a service/product/offer.  This opportunity may prompt some to sign-up or make a purchase. Since the second action directly impacts the business, its associated measures (e.g. orders per social acquisitions) would be the KPI. The exception might be a campaign where a key goal is to have an advertiser’s hashtag mentioned 100 times a week. In this case, hashtag mentions might be a KPI.

If I pursue a social network for seniors’ caregivers as my senior project, one KPI might be: engagements that include a successful event, defined as ‘joining’ the network,’ compared to overall site visits. Metrics could include specific pages visited and abandonments during registration. Sofia also suggested you should move from macro to micro insights to figure out why users are behaving a specific way online. In following this, these metrics for my project could be used to learn which pages are compelling or need improvement and how well the registration process is working.

The Heart of Analytic Success: SMART Goals and the Trinity Strategy


For our first analytics class, Sofia provided an overview of what analytics is, its value and how every successful project must start with SMART goals. Sofia defined analytics as the use of data to gain insights and make better decisions. I agree with this but would add that analytics helps you report up to decision-makers on how investment in your initiative has impacted behaviour and user experience to advance business goals, as reflected in outcomes. Analytics also helps you gain executive (or client) buy-in for future projects that build on a prior campaign’s analytics or results.

I think analytics is becoming increasingly important as we continuously have more volume and variety of information or ‘big data.’ Sofia defined big data as “a collection of data from traditional and digital sources inside and outside a company,” which companies are increasingly looking at for ongoing discovery and analysis to inform their decisions. This data includes many things we can measure and analyze from digital sources, such as how many people access a website through specific social media platforms and from what countries, as well as traditional sources, such as how much money an event raises or how many people attend it. What I found particularly new was learning the difference between ‘structured’ data, which is quantitative, such as how many visitors access your site via Twitter, and ‘unstructured’ data, which is more qualitative, such as comments posted on your company’s facebook page.

Before you can attain analytics, you need to set goals for a website or digital property. Sofia explained that you summarize each goal in a sentence that includes Specific, Measurable, Attainable, Realistic and Time-bound (SMART) attributes. This is a slight variation from PR campaigns (which I’m familiar with), where goals are broad and objectives have SMART attributes. In PR, I’m also more used to setting goals and objectives for campaigns with a definitive end point.  In this first class, we did an exercise to identify a website’s goals. The feedback from this exercise gave me the impression that SMART goals can be for an overall website or company (such as the Toronto Star) versus a set campaign. If that’s the case: how can you make these goals time-bound?  I think you might set them for an initial period goal, such as six or three months, and then reassess but it would be good to know for certain.

Sofia told us the grandfather of analytics is Avinash Kaushik, who discusses his theories extensively in his blog: Occam’s Razor.  The blog is named after a 14th-century English logician and one of the alternate translations of his principle is: plurality should not be posited without necessity.  I think this suits analytics because it provides a methodology for specifically identifying what big data an organization needs to measure and why (necessity), versus trying to measure all the data it can access (plurality).

We also learned about Kaushik’s Trinity Strategy. This is a strategic approach to extract insights and metrics from a website/other platform, based on the users’ behaviour and experience, as well as the overall outcomes.  It’s imperative that these insights and metrics can be ‘acted on,’ that is used to make decisions that alter the organization’s approach or to design future initiatives.

For example, let’s say you implement a promotional campaign to sell featured books highlighted on a page in your ecommerce site. You also promote it through direct mail flyers and social media. Your goal may be to sell 50 copies of each featured book within one month. The overall process begins with the clickstream data, which is the data collected through the site. I can include: who is accessing the book page, from where, via what devices or user path, etc. From this data, you can measure and assess users’ behaviour in response to the promotion, such as:

  • How many came to the site via each social media platform?
  • How many were driven there by the flyer?
  • Which book image or caption attracted users when they first landed on the page or made no impact for the duration (as measured by a heat map)

You then measure outcomes, such how much revenue was generated through specific online book sales. The third element is the experience, which tells why the users behaved the way they did. My understanding is this data is often accessed through additional steps, such user testing to measure effectiveness of user paths, experimenting  with site changes (e.g. trying book purchasing buttons in different positions) and customer/user surveys to assess how they feel about the site. These three elements, combined with competitive intelligence, help you uncover insights about what is attracting users to the site and getting them to buy the books, as well as what changes might improve outcomes. For example, in assessing this data, you may want to adjust the page layout, if users consistently miss a specific book displayed.