How to achieve success with web analytics

challenges of current web analytics

The successful implementation and execution of web analytics programs is challenging, especially in larger organizations. Web analytics success depends strongly on clearly defined business goals and because many organizations lack a clearly defined web strategy with clearly defined business goals, web analytics programs often suffer. This is an ongoing issue at the business level.

Common issues within organizations respect to web analytics.

* Management doesn’t want people to spend time on web analytics
* Management tends to think: purchase, install and read reports (aka analytics is magic)
* The discipline is rarely a dedicated operation — it’s a part time gig in many orgs
* Many practitioners are still struggling to make the business case.

it’s very bullshit, but true..
Concrete vs. Magic with Analytics

The perception that web analytics work by magic is prolific and problematic. Combating this need not be difficult, but some internal education is usually necessary. The following list of tasks is a solid way to start a web analysis project:

1. Collect business requirements
2. Define metrics and methods of collection
3. Find or make data available (e.g., coding tags, systems integration, etc.)
4. Calculating metrics — development of models
5. Build reports and conduct analysis
6. Educate stakeholders as to how to use the resulting analysis

Wise Use of Web Traffic Reports
Traffic Data Needs Context

Traffic reporting is not always useful, nor is it always wise to distribute such reports. When used well, traffic reporting provides snapshot views of important website activity. These reports serve to answer the question “how is the site doing” at a glance.

But traffic data has a context and that context is the previously defined web business goals. So for such reports to be useful one needs to provide as much context as possible: What are the goals? Which higher numbers are better? Which lower numbers are better? What does the vocabulary mean? Etc.
Big Numbers are Not Always Good

High traffic numbers are not necessarily a sign of success. If you’re running a portal and have 1.25 pages per visitor session, this could be a sign of success — users might be finding exactly what they need, quickly. But if you’re running a publishing site, then seeing 4 page views per visit would be a tremendous success — users are discovering interesting content and continuing to read and interact with the website.

Context is essential. Distributing traffic data without supporting contextual information can be worse than meaningless, it can distract from core business goals. Many traffic reports require deeper analysis to understand the implications.
Traffic Reporting is Not Analysis

For web analysis professionals, it’s important to differentiate (in the mind of the report consumers) the difference between web traffic reporting and web analytics reporting. And it’s important also to have good reasons, e.g., business goals, for distributing traffic reports.

Web analytics analysis typically goes far beyond traffic reporting and provides answers only available by drilling down into traffic reports integrating more sophisticated business data. For example, with a “Top 10 Search Terms (internal)” report, one very often wants to know what happened after the search was executed — …was the content found? …was there an exit event? …what happened next?
Analytics Dashboards Are Over Rated

Following from the discussion of web traffic reporting, Phil took a few swings at the use of web analytics dashboards. Like traffic reports, the point was made that dashboards often fail to be useful in terms of making business decisions. Dashboards tend to suffer from the following problems:

* They don’t often contain a proper explanation of use
* Context tends not to be explained well
* Reports get “thrown over the wall” without discussion or follow-up
* Baselines and goals may not be present
* They often are more about traffic than business activities

So what’s the answer? Identifying the stakeholders, understanding both their business interests and the way they consume information, and then delivering specialized reporting and/or data that fits them. One size fits all reporting can be self-defeating.

Doing better web analytics

By focusing on business goals, patterns and needs for information consumption, and deeper analysis, web analytics projects can deliver enormous value. Phil provided some guidance for how to succeed.
7 Business Questions Analytics Can Address

Keeping a focus on business questions is the foundation of a successful web analytics process. Some examples of these are:

1. What gets funded?
2. Which content or functional areas to deploy/fund human resources?
3. What projects to prioritize?
4. When is it time to redesign?
5. When to change content?
6. What navigation elements are working?
7. What search terms to purchase?

5 Questions for Web Analysts

When web analysts strive to refine their practice, Phil proposes that they ask these questions of themselves:

1. Who should be seeing raw data versus the resulting analysis?
2. When should we be analyzing versus reporting versus performing calculations on data?
3. How should we be presenting data and analysis to the different stakeholders?
4. What what data do we need to analyze?
5. Why are we analyzing this data?
6. Are we analyzing the key financial metrics (money talks at all levels)?

Explorative Web Analysis is a Must Have

Explorative web analysis is the process of understanding how people interact with a website or application. Our presenter today asserts that this level of analysis is key to the business value of web analytics. This form of analysis, according to Phil, involves drilling down on data sets to a level of detail that goes beyond standard traffic reports.

He states that typical analytics tools — such as Omniture, WebTrends, Unica, Nedstat and Google Analytics — may be adequate for this, but there are also often cases where one will need to access and manipulate raw data. This type of analysis involves activities like:

* Answering many of the typical questions raised by traffic reports
* Segmenting visitor activity to better understand the performance of content and the results of marketing campaigns
* Painting more rich pictures of how users are interacting with applications, content and navigation

Typical Explorative Analysis Exercises
General - All Types of Sites

Some broadly applicable explorative analysis operations include:

* Homepage analysis (e.g., content placement optimization)
* Internal search analysis (e.g., assess search usage and usability, identify ways to monetize search)
* Funnel and workflow analysis (e.g., identify fall out points and recommend process changes)
* Landing page analysis (e.g., analyze effectiveness of entry points with regards to moving visitors to and through key action funnels)

For Publishers

Online publishers have specialized needs. They are focused on driving visitors into specific content hotspots and optimizing the monetization of their content. Phil highlighted a few exploration examples for publishing contexts:

* Ad real-estate analysis — identifying the best locations for ads to be placed on a page
* Functional analysis — categorizing key site areas by function and create baseline measurements of each functional area
* SEO analysis — examine site entrances and site path behaviors segmented by key phrases.