- Количество слайдов: 25
Web Analytics Best Practices Kevin Murphy NASA ESDIS
Web Analytics in Context • Tools primarily developed for commerce use • Adapt commercial practices to improve our users experiences and increase usage • Government privacy and IT policy limit what technologies we can utilize
What is Web Analytics • The assessment of a variety of data to help create a generalized understand of the visitor online experience (Source: Web Analytics Demystified) – Monitoring and making the necessary improvements to make the visitor’s experience better • Quantitative: Web traffic, web based transactions • Qualitative: Surveys, Usability Studies, User submitted information, etc.
Components of a Successful Web Analytics Strategy • • Customer first mentality Define actionable business questions Design metric capturing methodology Implement and verify metric gathering Measuring performance Communicate actionable results Rinse repeat Business Questions Metric Capture Methodology Implement and Verify Measure Performance Communicate Results
Customer Centricity How is the website doing in terms of delivering for the customer? Target Question Primary purpose Why are you here? Task completion rates Were you able to download your data or access the content you wanted? Customer satisfaction How can we improve your experience? Did we provide enough information? Content and structure Did we meet or exceed your expectations? Are you reaching content within 3 -6 clicks? Outreach Impacts Trending A/B Testing Visit Duration Conversion Rate Bounce Rates Download Rate Outcome Metrics
Business Questions • Open-ended and at a high-level. • Likely require you to go outside your current systems and sources to look for data and guidance in order to measure successes. • Rarely include columns and rows into which data is plunked.
Examples • What are the top five problems users experience on our website? • What is the most influential content on our website? • What is the impact of the website on user services? • What are the most productive inbound traffic streams? Which sources are missing?
Metric Capturing Methodology • Methodology based on business questions – Lots of data to waste time on • Start simple and progress to complexity – Gain confidence in data and method • Create actionable insights and metrics • Actively define, refine and test your hypothesis Click Stream (Web Metrics) Segmentation Behavior Data Driven Decisions Experience Research Testing (A/B, Multivariate) Survey Outcomes Errors/Problems Downloads Content/Service
Methodology • Ensure you can tie data together Behavior User_id Session_id Test_value Site_id User_id Session_id Tracking_code Data Driven Decisions Experience Outcomes User_id Session_ID
Note on RIA Rich Internet Applications (RIA) • • • Technologies include ajax, flash, rss, blogs, adobe Requires planning upfront, during the design phase of the tool Identify core events and use them as key metrics – Science data download – Invoke online subsetter – Task completion rate (funnel reports) • Tracking Methods – – • Page tagging using Java. Script Event tagging Custom hooks Custom logs for integration between EMS HTMLDB and Net. Insight Are you currently using RSS? Feedburner – Metrics • Publishing content more people are interested in (subscribers) • What type of content causes people to come back and consume content – Number of times links that are published via RSS draw people back.
Avoid the Data Quality Trap • Web metric data are dirty • People do crazy things on the web • There will never be reconciliation between different tools • Lack of cookies limit our ability to track. – the policy could change. comments? – Proxies, multiple computers • So how do you make decisions, believe in the data?
Decisions and Data Quality • Start simple • Goals are important • Trending provides one context • Testing will tell you if you are doing the right thing • Communicate what you know
Implementing and Verifying Metrics Click Stream Segmentation Key Metrics • Sometimes you need more than one metric collection method – Linking between Net. Insight and HTMLDB is available – ACSI/Custom survey Un Behavior e nc lue l inf ma r To opti vio ha Be – Page tag implementation guide, contact EMS, consult online resources de Us rstan Ex pe ers d rie nc e • Plan and understand Data Driven Decisions Experience Research Customer Satisfaction A/B Testing Heuristic Evaluations Outcomes Orders Conversion Rates Problem Resolution Nuances of Outcome
Implementation Best Practices • Tag all pages • Create specific error tags (Not Found, Technical, ect…) • Make sure tags go last • Just before