A clear and easily readable overview of web analytics from a strategical and technical perspective. This review pulls on key themes in the book and summarises key points as soundbites based on personal opinion, knowledge and experience.
The shift to the neglected cousin of marketing activity – ROOI (Return of Online Investment) made possible by innovative and effective use of web analytics. Not just as a number cruncher but a behaviour indicator and futurist.
The importance of a customer centric culture that is underpinned by, not just a data driven culture, but an analytic culture – the importance of accountability should not be pushed aside for a pretty video viral campaign. Another mentality to bring into the workspace is of continual development / optimisation not silo’d projects to just make a step, this should be ingrained.
With web analytics there is an element of re-education and empowerment and it is the duty of the analyst to drive this within the organisation – key tool is use of KPIs and Dashboards (people like something tangible and interpretable), but most importantly the web needs to fit into the bigger picture. Web analytics need to be put in context of the overall marketing activity, tying together online and offline activity.
Analysis vs. reporting – the latter being the half-way action of a large majority of companies which although pretty and shows the rest of the business the web is doing something it doesn’t change / drive anything. This require Analytics Interventions – a step process that moves into a road to recovery:
- Intervention Step 1: Admitting the Problem – Perceived Gap (Inadequate analytic tools / Inability to track a site)
- Intervention Step 2: Admit that you are the problem
- Intervention Step 3: Agree that this is a corporate problem
- Recovery Issue 1: Lack of established processes and methodologies
- Recovery Issue 2: Failure to establish proper KPIs and Metrics
- Recovery Issue 3: Data Inaccuracy (inaccurate cookies (repeat customers), internal visitors, caching servers, incomplete or incorrect tagging, inaccurate filters within the tracking tool
- Recovery Issue 4: Data Overload
- Recovery Issue 5: Inability to Monetize the impact of changes
- Recovery Issue 6: Inability to prioritise opportunities
- Recovery Issue 7: Limited Access to Data
- Recovery Issue 8: Inadequate Data Integration
- Recovery Issue 9: Starting too Big
- Recovery Issue 10: Failure to tie goals to KPIs
- Recovery Issue 11: No plan for acting on Insight
- Recovery Issue 12: Lack of committed individual and executive support
Preparing to be Data Driven
The four steps of web analytics (ZAAZ’s cyclical methodology)
- Defining Business Metrics (aka KPIs) – not just figures, need business metrics that are in context of the overall business strategy – understand desired behaviours and how to monetize this.
- Reports (KPIs and contributing metrics)
- Analysis (gaining actionable insight)
- Optimisation and Action
- Results and Starting Again
Defining Site Goals, KPIs and Key Metrics
The Conversion Funnel – 1) Awareness; 2) Interest; 3) Consideration; 4) Purchase.
Understanding KPIs – KPIs need to be tied to the organisations goals, measured over time and is agreed on by the organisation as a whole. Importance of the process behind the KPIs to evaluate, action and drive them. Normally experessed numerically the key to KPIs is that they flow directly from the business goals (high to low / top to bottom flow and personalisation level). Importantly it is not just about the final goal by the drivers that motivate users. Consideration needs to be given to the type of site i.e. lead generation vs customer service when defining KPIs. For a rounded view of the channel need to look at KPIs in terms of:
- Competitive Measurement
Establish targets with the bigger picture in mind, taking into consideration seasonal and daytime fluctuations. To help determine reasonable targets you can look at standard deviation, goal fulfilment, competitive audit (benchmarking).
Monetizing Site Behaviours
Assigning values to site behaviours – speaking the language of the stakeholders not falling into the web techie trap and therefore often getting overlooked as part of the whole business. The language of Money.
Monetization models are a key tool to level all aspects of the site to determine best business drivers, there are two types: additional revenue (direct and indirect) and reduced costs. Consider the use of a pro-forma to identify fiscal value of e-investments.
Getting the Right Data
Different data types can provide a more rounded understanding – focus here is on behavioural (what), attitudinal (why), competitive and transactional data types. Behavioural data is a good starting point to leverage other types of data and can identify areas of promise to hone in on. Attitudinal is typically through surveys and focus groups, both behavioural and attitudinal when used together provide greater insight than one in isolation.
Competitive data there are three main different types: 3rd Party Private Networks (external companies that collate data on internet usage, often provide tools to clients); User Centric Networks (i.e. Nielsen/Net Ratings – ideal for advertising); Network Centric (Hitwise) (work with anonymous data on log files of internet users via agreements with ISP). When looking at competitor data it is not about scorecard, you are looking for insight – this requires a plan and a clear understanding of what data types to use and why.
Other data types:
- Customer Interaction Data (call centre, customer and transaction data)
- Third Party Research (latest web trends and strategic thinking)
- Usability Benchmarking (usability focus group doing tasks on your own and your competitors website)
- Heuristic Evaluation and Expert Reviews (a more cost effective alternative to usability benchmarking in some cases)
- Community Sourced Data
Key to leveraging data is to do so together not one type in isolation, look for trends and commonalities – you are putting together a story not just facts. Being able to compare not just against industry or competitor standards but importantly against past performance is essential to identifying gaps, success, failure and ensuring that the online channel is indeed moving toward the business goals in allegiance with the rest of the business (relative indices can help here).
Customer engagement is a visitor based calculation – the simple rule is : the engagement of any visitor is a function of their lifetime of visits. By going back to the indices these can be leveraged across the business using a simple 8 step process:
- Define the key index
- Determine the frequency of reporting
- Cultivate a culture of analysis
- Make Budget Allocations
- Benchmark Marketing Initiatives
- Benchmark site-behavior types
- Prioritise resources
Analysing Site Performance
Most likely reasons the companies don’t succeed with analytics:
- They do reporting not analysis
- Lack of methodology to use the data
- Not focussing on the right metrics
- Poor tool implementation and configuration (skilled resource considerations in terms of implementation and ongoing use – need for training or / and recruitment)
- Accuracy Issues
- Poor communication of site goals
- Data Smog (too much data)
Analysis using the conversion funnel approach, look at encouraging the users to make the ‘right’ decision at each stage (click on, leave page, leave site), the funnel needs to be broken down to see a more granular process this defines the data that needs to be collected at each stage and determines how you look at the data – consider in terms of behavioural and attitudinal.
This basic approach can be used on lead generation as well as e-commerce sites, the main difference is that lead generations sites often have more than one way to get users to make contact with a sales rep unlike the shopping cart purchase process.
The search feature on the site is often a default behaviour for visitors – is incredibly valuable and keywords could indicate attitudes not just behaviours of on-site users. It is important to understand the effectiveness of this tool for user experience improvement and ultimately conversion uptake, and the search terms used for product preference, common problems, point of search, segmentation etc.
Other key areas for analysis – home page or/and campaign landing page effectiveness (bounce rate, source, where they go next), branding metrics (there are a number of key metrics i.e. direct visitor traffic, repeat buyers etc branding should not overshadow other metrics). Another approach is to segment traffic to identify behavioural differences (amazon.com the one-to-one market leader). Cross channel marketing is at the forefront of my thinking whenever I look at a campaign similarly in web analytics it would be foolish to report on a one channel basis – analysing drivers to offline conversion is a key metric that senior management will also appreciate. Delayed conversions (often hard to track – CRM systems and / or promo/origination codes) are a further consideration.
This is something I experienced in my project management roles and was often baffled that unquantifiable projects were going ahead of projects that when monetised would have a greater impact on the books – I place the majority of this to culture and the little nuances that make up the office community. However with an analytical mindset and by implementing monetised models projects are quantified and can be dynamically prioritised (enabled by granular measurements and easy to change technologies). Key features of this approach are:
- Prioritisation based on business impact
- Optimised resource planning
- Accelerating the release cycle
- Holding initiatives financially accountable
The problem is forecasting the potential impact, the first step is to assign a value to the desired site behaviour identify desired behaviour and impact on that behaviour through change – though there is no clear cut formula it can help provide a range you forecast too. When comparing projects risk, cost, likelihood all need to be taken into account not just impact – this can be done using a framework.
Moving from Analysis to Site Optimisation
This is where you put into action the work so far, key methodologies and testing tools:
- A/B Testing
- A/B/n Testing
- Multivariate Testing (looking at combinations not just different versions)
- Tools (most work with leading analytic software providers) include; Toch Clarity, Optimost, Site Spect, Google Website Optimiser.
Testing (although almost limitless) typically fall in the following test categories: price; promotion; layout; message; functionality; navigation. As discussed earlier tests need to be prioritised as well, again consider impact, time, risk etc – monetise testing. Testing can be optimised for segment performance so where one version is good for customer x it isn’t for customer y etc – this involves understanding your customer segments from the outset. Another consideration is time of day/week changes this can be automated. Keywords that I am getting from this are: granularity, accountability, realisation. The test plan:
- Identify Opportunities
- Develop a hypothesis
- Determine the test methodology
- Define success matrix
- Design your options
- Launch your test and analyse your results
- Reap the benefits
- Post test analysis (need baseline to understand impact of change)
- 7 Tips I Learned from Avinash Kaushik’s Web Analytics Course (digitaloperative.com)