Web analytics can be a very confusing topic for small business owners. Much of the information on the web is written from a specialist’s point of view, and meant for other specialists. Some content is written specifically for decision makers inside organizations with complex operations, and these viewpoints don’t always fit with the needs of owner-managers and their teams.
Leveraging web analytics will almost always enable discovery of valuable insights that aid in an organization’s decision-making capabilities. But, leveraging web analytics won’t always add enough value to justify the expenditure in resources that a dedicated program requires. That’s an important distinction—if what you’re doing isn’t adding more strategic or immediately quantifiable value than it consumes, it’s not fiscally prudent to continue. The aim of this article is to give you a high-level view of web analytics, and introduce some of the key concepts to aid in making informed decisions.
Demystifying Web Analytics
Web analytics is a form of performance measurement that seeks to identify patterns of behaviour to which you can assign economic value. For instance, if Visitor X finds your web property and takes Action Y, the business will likely realize Revenue of Z. That’s the heart of web analytics, though the implementation is more complex in practice. Over time, this measurement and analysis helps to improve the return on what you allocate to your digital marketing activities.
Business Analytics vs Web Analytics vs Digital Measurement
These terms are commonly treated as interchangeable, but there are distinct differences between them that are important to note.
Business Analytics refers globally to any planned measurement of business performance that’s intended to guide making iterative improvements. Both web analytics and digital measurement can be considered business analytics.
Web Analytics refers specifically to the planned measurement of user behaviour on web properties in order to guide making iterative improvements.
Digital Measurement encompasses a more holistic look at user behaviours across multiple channels and platforms. It seeks to look at the performance of digital marketing as a system over looking solely at user behaviour on a website. Industry professionals are beginning to adopt this mindset as digital marketing becomes more nuanced and integrated with ‘offline’ activities.
The shift to digital measurement marks an important move to quantify user behaviours in new technologies, beyond traditional websites. Implementing digital measurement software in self-serve kiosks, point of sale systems, and emerging human-computer interaction devices has become increasingly common. Since 2013, Google has freely offered a software development kit that can be used to deploy its analytics software on any internet connected device (the “Measurement Protocol”).
Commonly Used Terms
Attribution
Assigning the source of a conversion or sale to a specific user behaviour or marketing channel. Historical attribution data is valuable when budgeting for a new fiscal year, as it gives some insight into what advertising activities generated revenue.
Conversion
A specific action taken by a user that indicates purchase intent (or an eCommerce sale). Conversions can be further classified as macroconversions (actions which are directly linked to revenue) and microconversions (actions which aren’t directly linked to revenue, but still hold business value).
Dimension
A dimension is a qualitative factor that adds detail to a metric by refining its focus. Examples of dimensions include location, mobile device, and a user’s browser. Using the location dimension, you could specify that you want the analytics software to determine the number of sessions originating in Montreal for a given date range, for example.
Filter
Filters are customizable sets of conditions that are used to include or exclude data en masse. For instance, if you wanted to look at data specific to users in your city, you could create a filter to include this location only. When using filters to manipulate data, be sure to keep backups of your unfiltered data in case you need to look back at it in the future.
Medium and Source
Medium and source together describe how a user entered your web property (medium is used to categorize source). For instance, an example of a medium is organic search (a search engine) with the corresponding source being yahoo.ca.
Metric
A metric is any quantifiable data point that can be measured. Time on site, sessions, and pageviews are all examples of metrics.
Pageview
A pageview is counted when an entire page (including all of its resources) is loaded. This is slightly different from ‘hits’, which is a count of all resources requested from the server. Loading each page requires multiple hits.
Session
A session is a single visit to a web property by a single user, and typically times out after a predetermined period of inactivity (where the user doesn’t interact with the property). Sessions are also sometimes called visits or interactions.
Tag
A tag is a snippet of data appended to a link that adds context to potentially ambiguous inbound traffic sources. For instance, if your business has a Twitter account that is active and an ad campaign running on Twitter, you would want to know if traffic (and conversions) from twitter.com should be attributed to the ad campaign or to your business-as-usual activities there. Adding a tag links in the ad campaign would make attribution more precise in this case.
Creating a Digital Measurement Plan
Many businesses don’t create a measurement plan before diving into web analytics. This contributes to data fatigue and causes teams to get derailed before they start to see value from their efforts. There’s so much data readily available that it’s easy to lose momentum and focus.
Fight off data fatigue by creating a plan before you begin. A useful digital measurement plan includes at least three components: business objectives,key performance indicators, and measurable behaviours (metrics). It’s important that the person creating the digital measurement plan understand all three, or solicit help from stakeholders who can fill in the gaps.
A well-prepared digital measurement plan can serve as a document to facilitate buy-in across stakeholder groups, and can even serve as a springboard for preparing a more robust sales and marketing plan.
Step 1: List Business Objectives
What’s your Big Hairy Audacious Goal (BHAG)? If you have one, write it down—this is important context for what you want to achieve. Also list any specific business objectives that you’ve identified.
Step 2: Develop and/or List KPIs
What are the indicators of success that relate to your business objectives? List them alongside the specific objectives to which they relate.
Step 3: Identify Measurable Behaviours
For each KPI, list measurable behaviours (metrics) that can be used to gauge your success for that KPI. This step is often the most time-consuming, as it requires a fair bit of intuition and inference when historical data doesn’t exist.
Step 4: Identify Reporting Frequency
How often do decision makers need access to the insights provided by your web analytics team? Daily? Weekly? Monthly? Quarterly? This will vary depending on your business objectives. It’s important to note that ‘reporting’ in this context does not mean collecting data—that should be heavily automated to reduce costs. Reporting refers to the human analysis of the data which generates actionable insights.
Interpreting Your Data
In order to really get value from a web analytics program, your organization needs the capacity to produce a useful interpretation of the data collected and processed by the analytics software. Having data without also having meaning amounts to wasted time and money. Always seek to identify actionable insights from your efforts. If you can’t make improvements to the business by leveraging your data, all you have is fun facts and figures.
The web analytics chain looks like this at a high level:
Data Collection -> Data Processing -> Data Analysis -> Data Interpretation
The first two steps in the analytics chain are typically automated by software, with the two being left up to humans (for now). In your interpretation of the data, there are three questions that you should seek to answer:
What happened? What might this mean? What could and should we do about it, if anything?
You can approach answering these questions with either reactive or proactive thinking. Reactive thinking is useful in evaluating the performance of content marketing or advertising campaigns (looking back). Proactive thinking is useful in identifying trends that are meaningful to your business (looking forward). Proactive thinking requires more deductive reasoning than reactive thinking, and is also less concrete.
When to Hire a Specialist
Essentially, web analytics is no different than any other form of business analysis that looks at numbers in order to measure performance. Hiring a specialist in this discipline is akin to hiring an analyst in more traditional disciplines—payroll, inventory, logistics, and the likes. Making the decision to hire a specialist is simple on the surface—either when the time requirements or the knowledge requirements move beyond what your team has available.
But there’s another factor in this decision: is the opportunity valuable enough to justify the cost? This question is much more difficult, and it’s often worthwhile to hire an experienced consultant to help you evaluate the opportunities and costs associated with implementing a web analytics program for your business. Implementing a program doesn’t necessarily mean having this function in-house, as there are many reputable data marketing firms that offer full-service programs that can be customized to your business.
—
This article was originally published in the October 2015 issue of Direct Marketing Magazine.