Analysing web traffic instruction
Best practice techniques for interpreting web traffic data in order to obtain evidence-based and actionable findings for improving websites. This instruction should be used whenever analysing the performance of a website or website section via traffic data to determine whether changes need to be made. Making changes to website structures and page layouts should be informed by the thorough, evidence-based approach provided here.
This instruction does not apply to:
- courseware, including scholarly work, student work and teaching and learning materials
- websites that have no relationship to RMIT (for example, personal or private sites).
- Google sites
Instruction steps and actions
Web analytics measure where a website’s visitors come from, where they go within a site and how they interact with it. It can capture:
- the precise geographic location of visitors
- the webpage they were on before visiting a site and the action they took to get there
- how long they stay
- what they click on and in, mouse over, submit and search for within a site
Details of the visitor’s connection type and internet service provider are also stored.
All of these details can tell a story of how well or poorly a website is performing for its target audience so that the website can be optimised and marketing campaigns adjusted. Analysing web traffic can identify potential usability issues, popular content and the impact of online and offline marketing. Stakeholders for analytics therefore include designers, developers, content authors and marketers among others.
Results from web analytics can be used to indicate specific problems as well as where further user research is required to troubleshoot performance. Conversely, traffic analysis can complement and quantify observations made during one-on-one user sessions. Analytics can answer what is happening on a website, while qualitative user research can answer why it is happening.
Accuracy of data is paramount to forming useful insights. Setting up and configuring analytics programs are covered in the Implementing Web Analytics Tracking Code Guideline.
Step 1: Engage stakeholders and confirm website goals
If the insights gleaned from analyzing web traffic are to result in action, the feedback loop needs to be complete. Stakeholders need to be part of the report planning and communication. Identify those people whose roles rely on the performance of the website to meet their organisational goals, and those who are responsible for supporting and improving the website. People may have different levels of interest in the website so consider using the RASCI (responsible, accountable, supporting, consulted, informed) framework to help prioritise people’s involvement and input.
Talk about the process of web traffic analysis and how it supports an approach of hypothesising about the impact of a change, experimenting with design, content, feature and marketing ideas, and monitoring results to support improvement. The goal of the reporting should be to provide reasons for making further changes to the site and other offline activities, and to quantify what effect a change might and does produce.
Document stakeholder names in your analysis plan.
Step 2: Prioritise website goals
The numbers that matter are those that reveal how the website is tracking against the organisation’s or project’s strategic and/or tactical goals. Confirm with stakeholders what strategic or tactical goals the website is supporting. Prioritise them to find those that will provide the most organisational value (e.g. reduce costs, increase leads).
Discuss also the website’s most important visitors in terms of website behaviour (e.g. those that buy the most, comment the most, log in the most, watch videos to the end, share the most).
Then identify what aspects of the site can be measured to reflect the performance against those goals and the behaviour of those prized visitors. These metrics are the key performance indicators (KPIs). One goal may require several KPIs to measure it.
Some may be annual metrics, others monthly, weekly, daily or even hourly depending on the goals with which they align.
List the priority website goals in your analysis plan.
Step 3: Identify and refine KPIs to measure
Fight for quality KPIs over quantity so that the analytics report stays effective and efficient. KPIs need to be:
- quantifiable - they can be measured objectively and compared over time
- actionable - they provide business value and can be responded to
- appropriate - if a change is made to the site does this help measure the impact.
Broad goal categories include:
- Increase sales
- Improve customer service
- Improve efficiency (optimise organisational support costs)
Types of measurement include:
- Quantitative (i.e. specific data to user experience)
- Qualitative (i.e. quality of the user experience)
- Engagement (e.g. level of engagement with content)
- Productivity (e.g. time spent completing a task decreases)
Examples of KPIs include:
- Task completion rate
- conversion rates - checkout conversion, order conversion, newsletter subscription, enquiry form, requests for content downloads, bookmark page
- Conversion rate for top keywords (marketing)
- average order value
- customer loyalty - ratio of new to previous visitors or customers
- search engine referrals, paid search term referrals, paid search term conversion rates
- cost per lead
- traffic concentration
- depth of visit
- customer satisfaction
- length of visit
- internal search terms
- social media participation - sharing, liking, commenting, mentions
- audio or video interaction
List the KPIs to be measured in your analysis plan. These can be presented as research questions (see appendix below).
Tip: Automate where possible. Analysts with access to GA can set up automatically generated reports so that they communicate via email the KPIs of interest at specified time frames. Automate as much as possible to save time when preparing for your regular reports. Setting up and configuring analytics is covered in the Implementing Web Analytics Tracking Code document.
Step 4: Identify timescales
Pick a point in time to commence measurement to establish a baseline, or benchmark, from which future performance can be compared. Ideally this point in time should be a moment where it is business as usual to provide straightforward comparisons in the future, rather than an unusual seasonal (e.g. summer holidays) or cyclical (e.g. budget announcement) fluctuation.
Against each KPI, list the points in time that will be compared.
Step 5: Set your targets
Set targets for each KPI so that you know when success has been reached. This might be communicated as a numerical value or a percentage increase or decrease. Each target needs to follow the SMART (specific, measurable, achievable, realistic, timely) approach.
Against each KPI, list a target value (or measure of success)
Step 6: Conduct the analysis
Use GA to produce a data report for each KPI. Each report should take into account:
- Relevant pages
Analyse the data to measure success rates against the targets set. Identify instances of successes and failures. Try to identify the causes of failure. Take care not to draw immediate conclusions - discuss ambiguous findings with the Senior User Experience Analyst.
When you can’t work out what the data is saying
- Find different ways of looking at the data. Visual representations like pie charts, bar graphs, ranks or word clouds may better reveal the story.
- Look at trends over time and look for patterns or anomalies. Compare at different time scales (e.g. daily, quarterly).
- Consider A/B testing different designs, content or features to understand which performs better. Refer to the Usability Testing Instructions.
- Engage other user research tools such as usability testing to troubleshoot what the issues with the site are. Refer to the Usability Testing Instructions.
Step 7: Report actionable findings
Produce a report to share with stakeholders. The report should indicate how well the website is performing against business objectives and include recommendations for website improvements if required. Avoid using raw data in the report – instead use graphs relevant to the points being made.
Present the report to stakeholders and discuss the recommendations it contains.
Tip: Before finalising your report, consider asking WSIP to provide a review of your draft. Contact the Senior User Experience Analyst. Also, if serious user experience problems have been identified, WSIP will be able to provide support in relation to stakeholder management.
Appendix: Refining your research questions
The organisation goals and KPIs will determine the appropriate metrics for your reporting, but there is a range of behaviours and activities you may want to track depending on the maturity of your analytics initiative:
- Website metrics: visits, page views, top entry and exit pages, traffic sources, popular content, keyword drivers, conversions, capacity and uptime.
- Behaviour optimisation: funnel analysis, A/B testing, path analysis, landing pages, custom tracking (e.g. email campaigns).
- Integrated marketing: segmentation, search engine optimisation, KPI alerts, social monitoring.
Some important areas of analysis are as follows.
Homepage and landing page optimisation
Consider the bounce rates and page values of the home page and other key entry points or landing pages of the site. Visit pages with a high bounce rate to see if it is obvious what is wrong with them (e.g. broken links, insufficient or inappropriate content, weak calls to action). Visit pages with high page values and determine why they might be successful.
- What page elements are most interacted with?
- How many people visit this page and leave the site immediately? How does this compare to the site average? If they stay, where do they tend to go next?
- What search term did those that left immediately use? What search term did those that delved further use?
- What referral sites deliver the visitors most likely to stay versus those most likely to leave?
- Are the wrong visitors being lured to the site or is the page performing poorly?
- Does the high bounce rate mean people are not finding what they need or are able to conduct their task efficiently and move on?
- Where do segments from different sources click on a landing page? How can a landing page be customised for each traffic source for better results?
Search engine optimisation
Look not just at the search terms people are using to access the website but also where those people end up on the site. Use a word cloud generator such as Wordle to paste in the top 100, 000 keywords and see instantly what is hidden in the long tail.
- How many referrals come from organic search versus paid advertising placements?
- What kind of behaviour do you see from a visitor using the same keyword in organic search versus paid search?
- Which keywords are most effective in delivering qualified traffic?
- Are there any surprising or missing keywords?
- How many impressions does your website need on a search engine before it generates equivalent clicks from local, interstate or overseas visitors?
- What paid search terms need to be geo-targeted based on popularity in different locations?
Internal search optimisation
What people search for within the website clearly demonstrates their intent. Check how well their intent matches the content and features they encounter.
- From what page do people start searching?
- What search terms are the most commonly used?
- Where do people go within the site after searching for certain keywords?
- Are there any surprising or missing keywords?
- What is the conversion rate like for the top keywords?
- What keywords have no supporting content?
- How do keywords differ for people from different locations, sources and other segments?
- Consider the processes on the site that require a visitor to pass through multiple steps to complete their task (e.g. fill out a form.) It might be useful to think of macro conversions on the site (e.g. buy a textbook) and micro conversions (e.g. submit a question to a forum, register) to capture the range of desirable visitor behaviours. The key is to discover what are the top conversion obstacles. Use a free tool such as PadiTrack to look at conversions from visitors across sessions to capture those who start one day and finish another day.
- From where do most visitors access the first step?
- How many visitors start the process but don’t complete it?
- On which step so most people drop out?
- Which segments of visitors convert differently? Do people sourced from email campaigns drop out at a different point to people sourced from organic search? What about Victorians versus people from interstate? What about people who use certain keywords to get to the site?
Content and cross-reference optimisation
Some content will attract more visitors, have more clicks within them and deliver more visitors to your most important pages. Consider the following questions to identify what content this is.
- What content is the most visited? Bookmarked? Shared?
- Where are people clicking on a page? Which clicks are delivering the most value?
- Which calls to action attract the most clicks? Where you have duplicate calls to action, which position works better?
- Which pages do people most often view as part of the same session? Is there an easy way to move between these pages?
- How does different content perform over time? What are the seasonal or cyclical trends?
- What days of the week and times of the day are most popular? How does this relate to the publishing schedule?
- What pages are slow and impacting usability? Where are people spending a suspect amount of time?
- What are the shortest paths to a successful conversion?
- Which content is the most successful and results in the most value (whether its revenue, conversions or another business goal)?
- Which pages have the most potential to deliver more value?
Visitor clustering and segmentation
Its useful to isolate the behavior of visitors based on how they get to the site, the keywords they used, their time on the site and so on. Segmentation can also be performed based on demographics such as geographic location, gender, age group. Your aim is to identify whether the target audience is being attracted to the site as well as what different segments of the audience do once they arrive. This way you can determine what types of visitors are the most valuable and how to attract and retain more of them.
- How many times do people return to the site in a month? What segment do the frequent visitors belong to?
- What proportion of people visit twice in 24 hours? What segment do they belong to?
- What types of visitors are most likely to convert?
- Which countries, states and cities have the highest conversion rates?
- What keywords do people from different locations use?
Since mobile use is growing the focus on mobile behaviour is important. Recognising the different access and usage patterns can inform design, content and marketing strategy.
- What are the main mobile platforms visitors are using?
- Are mobile users most likely to come via paid search, organic search, directly or from social media?
- What day of the week or time of day are the different platforms being used?
- How does mobile performance compare to general site performance?
Marketing campaign analysis
Analysing marketing campaigns can help improve their effectiveness and reduce costs by targeting the right audience and most valuable visitors.
- Is there a balance of traffic from a variety of sources (e.g. search engines, direct traffic, referring sites, campaigns)? Is traffic from one type of source particularly strong or missing entirely? How does this play out over time?
- Is the balance of traffic from the various sources in line with the marketing spend per source?
- What is the Cost per Acquisition (CpA) per source (SEM, Display, Facebook advertising)?
- What proportion of direct traffic is returning visitors or new visitors driven by offline marketing activity?
- Which banners, emails, Facebook display campaigns, social media campaigns etc generate the most value?
- Which sites and campaigns produce the most revenue?
- Besides Google, what international search engines should be we advertise on based on traffic?
- Which campaigns are not working?