Key Takeaways
Replace click-only analytics with five complementary views of the customer
Clicks describe behavior, not motive. Web Analytics 2.0 broadens measurement beyond page views and visits. It combines site activity, business outcomes, controlled tests, direct customer feedback, and information about competitors. Together, these sources address four practical questions: what happened, how much value it created, why people acted, and what is happening elsewhere.
A report pile is not insight. At one company, Kaushik stopped 200 routine reports. Nobody asked for them during the next three weeks. Their absence exposed the problem: they tracked activity but not results. The point is not to collect every possible data stream. Use several distinct sources because each reveals something the others miss, then connect findings to a decision that can improve the customer experience.
What's powerful here is the shift from measurement as surveillance to measurement as triangulation. A clickstream can show where users leave, while a survey may reveal they never intended to buy; a test can establish whether a design change helps. This resembles mixed-methods research, where different forms of evidence constrain each other's blind spots. The practical limit is coordination: more sources can create more contradictions and work. Multiplicity adds value only when a specific decision needs the extra perspective.
Tie every website to measurable revenue, savings, or customer success
Count value, not applause. The book argues that a site should be judged by outcomes such as sales, reduced operating costs, customer satisfaction, or loyalty, rather than traffic alone. A key performance indicator, or KPI, is a metric tied to an explicit objective. The objective differs by site: an online store may measure orders, while a support site may measure whether visitors solve their problem.
Include smaller wins. Macro conversions are a site's main result, such as a completed purchase. Micro conversions are valuable secondary actions, such as downloading a product guide, opening an account, or finding store directions. Kaushik estimated that four goals for his blog created $26,210 in economic value in one month, including actions that did not directly generate a sale. Assign values carefully, using finance or offline data when possible.
The strongest feature of outcome measurement is that it translates digital activity into the language used to allocate resources. It also broadens the definition of value: a successful support page may prevent a costly phone call, even without an online transaction. This is a useful antidote to vanity metrics. Still, an imputed value is an estimate, not cash in hand. Teams should label assumptions, validate them against observed behavior, and avoid letting a convenient dollar figure masquerade as certainty.
Segment people and behavior before trusting averages or percentages
Averages hide different journeys. A site-wide average can combine loyal customers, first-time researchers, job applicants, and people who arrived by mistake. Segmentation separates groups by behavior, source, or purpose so a metric can point toward action. Distributions often reveal more than averages: an average visit length might conceal that most visitors leave quickly while a smaller group spends much longer.
Check what metrics actually count. Bounce Rate is the share of visits with only one page view. Exit Rate counts visits that end on a particular page, whether or not the visitor succeeded. Conversion Rate also depends on its denominator: visits treat each session as a new chance, while unique visitors allow for consideration across multiple sessions. Daily unique visitor totals can double-count the same person across days. Define metrics, add raw counts and context, and inspect segments before deciding what is broken.
This is a lesson in measurement validity: a number is useful only if its operational definition matches the question. Segmenting also reduces Simpson's paradox, where a combined trend can reverse the pattern seen within groups. Yet segmentation can become a fishing expedition, producing apparent stories from tiny samples. The discipline is to choose segments tied to a business hypothesis, keep base counts visible, and resist interpreting every difference as a meaningful customer preference.
Fund analytical talent before buying another impressive analytics platform
Tools store data; people interpret it. The book's 10/90 Rule proposes that a successful analytics program should devote roughly 10 percent of its resources to tools and implementation, and 90 percent to skilled analysts. Even a free platform carries implementation, administration, and training costs. A powerful system can still produce little value if nobody can connect its reports to business decisions.
Choose fit over prestige. Before selecting a vendor, decide whether the organization needs broad reporting or deep analysis, assess technical and business capacity, and calculate total cost of ownership. A company once spent 15 months implementing a tool before discovering it was a poor fit, then spent more time replacing it. Pilot finalists on real tasks, test segmentation, support, data exports, and costs, and choose for needs over the next few years rather than imagined future perfection.
The 10/90 Rule usefully reframes analytics as a capability, not a software purchase. It aligns with the broader idea that technology yields returns only when paired with expertise and organizational routines. The exact ratio should not be treated as a universal budget law: costs vary with scale, regulation, and technical complexity. The durable question is whether the organization has enough skilled people to turn a measurement into a decision, an experiment, and a documented result.
Ask visitors why they came, what failed, and whether they succeeded
Behavior needs explanation. Clickstream data records actions but cannot tell whether a visitor was pleased, confused, or simply finished. Voice of Customer, or VOC, means gathering direct customer feedback through methods such as surveys and usability tests. A brief exit survey can ask visitors their purpose, whether they completed it, and what prevented success if they did not. Those answers explain patterns that traffic reports alone leave ambiguous.
Watch tasks, not just opinions. A lab usability study observes people attempting realistic tasks on a site. The book describes studies commonly using 8 to 12 participants, with recurring problems often visible in small groups. Moderators should avoid coaching, and teams should prioritize observed failures, then measure whether fixes work. Small samples are good for finding usability issues, not for estimating how common they are across an entire customer base.
The combination of stated intent and observed behavior is especially valuable because people may not accurately explain their own actions. Usability research can reveal friction that logs cannot record, while surveys can show whether the task mattered in the first place. This complements behavioral science's distinction between what people say and what they do. The caution is representativeness: a handful of observed users can uncover a design flaw, but broader claims require broader evidence.
Use live experiments to settle arguments that dashboards cannot resolve
Turn opinions into testable claims. A/B testing compares different page versions, while multivariate testing varies several page elements at once. Both let real visitors encounter alternatives, with success defined in advance through outcomes such as purchases, task completion, or fewer bounces. Start with a clear hypothesis and a meaningful change, not cosmetic tinkering that cannot affect results.
Test beyond the page. In one example, a company tested offering a low-priced product for free to a small share of visitors. The free offer attracted customers and also shifted some purchases toward more expensive products. Controlled experiments can address harder questions too, such as whether branded paid-search ads add sales or merely divert clicks from free search results. Randomization helps isolate effects, but tests still need adequate traffic, sound implementation, and checks for unintended harm to other goals.
Experimentation replaces status-based debate with evidence about a defined intervention. Its deeper organizational benefit is procedural: an idea can be tried without granting its sponsor permanent authority over the design. This is closely related to causal inference, where randomized assignment helps separate an intervention's effect from competing explanations. Tests are not automatic truth machines, however. Poorly chosen metrics, low sample sizes, or many simultaneous experiments can produce confident but misleading winners.
Benchmark competitors for direction, not false precision
Your dashboard has no windshield. Internal analytics shows how a site performs; competitive intelligence, or CI, estimates how competitors and markets perform. CI can reveal traffic trends, search interest, audience profiles, referral sources, and keywords. It helps identify opportunities that a site's own reports cannot show, such as competitors gaining ground or a company's search share remaining small despite rising traffic.
Know how the estimate was made. CI providers may use browser toolbars, panels, internet service provider data, search logs, site-submitted data, or combinations of sources. Each method has limitations and possible bias. Their totals will rarely match a site's own analytics, which has direct access to its instrumentation. Treat differences as expected. Check methodology and sample composition, then compare trends over time rather than treating a competitor's estimated visit count as an exact fact.
Competitive data restores strategic context that a site's internal dashboard cannot supply. It can reveal that a rising metric is still losing ground against the market, an important correction to inward-looking optimization. The methodological warning is essential: an estimate with systematic bias can still support trend comparisons, but it cannot safely answer every question. Sound use begins by asking what population the data represents and whether changes in that population could explain the apparent movement.
Measure the long-tail search journey, not just last-click sales
Small queries add up. In one example, 13 high-volume search phrases brought about 5,128 visits, while 26,124 less common phrases brought about 35,534. Focusing only on the leading keywords would miss most search visits. Many less specific queries introduce people who are still exploring; they may return later through a brand search and convert then.
Give each stage a fair test. Last-click reporting awards the sale to the final recorded campaign, which can make an earlier discovery query look worthless. Group keywords by their role in the buying process and judge early queries by relevant signals such as low bounce or continued research, while expecting stronger conversion evidence from purchase-ready terms. The same caution applies to marketing attribution generally: a credit-allocation formula describes a rule, not proof that a channel caused the sale.
The long-tail example captures a structural feature of search: many individually rare expressions can collectively represent substantial demand. It also exposes a weakness in last-click accounting, which confuses proximity to purchase with causal contribution. Multi-touch paths offer context, but no arbitrary credit split establishes what would have happened without a campaign. Incremental tests, such as changing spend across comparable markets, provide stronger evidence when the budget decision matters.
Treat online and offline actions as one customer journey
Customers cross channel boundaries. Someone may research a product online, visit a store, and purchase by phone. A web report that counts only online orders misses part of the site's contribution. The book calls this nonline marketing: customers move between digital and physical channels rather than following separate online and offline paths.
Create practical links between channels. Vanity URLs and unique coupon codes can connect offline advertising to site visits or store purchases. Store-locator use, phone calls, surveys, and geographically controlled campaigns can reveal other parts of the journey. The same measurement problem appears in social media, mobile apps, and video: activity may happen outside a conventional webpage. Event tracking can record meaningful interactions, while feed subscribers and video completion can capture outcomes page views miss. No single tactic provides a complete account.
This view anticipates customer-journey analysis: channels are touchpoints in a process, not independent departments. It also makes measurement a design problem, since codes, call tracking, and shared identifiers must be planned before campaigns run. Privacy and consent matter whenever identity is linked across systems. Correlated online and offline movement is useful evidence, but not automatically causation; a well-designed geographic or randomized test can strengthen the claim.
Turn analytics into decisions with a short, evolving action dashboard
A dashboard should prescribe a next step. The action dashboard centers on a few important measures and adds context: trends, interpretation, recommended actions, and the likely impact of inaction. It identifies who owns the business result and who analyzes it. The aim is to replace a dense report that executives must decipher with a concise account that helps them decide.
Move before data becomes obsolete. Web data is imperfect, and the book favors consistent, usable measurement over waiting for impossible perfection. Its implementation path starts with basic tagging and tool settings, then adds campaign tracking, revenue detail, and richer media events as needs justify the effort. A data-driven culture grows through visible results, direct work with business teams, and investment in analysts. Metrics and dashboards should change when the business changes.
The action dashboard treats analysis as a bridge between evidence and organizational behavior, not as a final display. Naming owners and stating the cost of delay can make accountability concrete. This is also a governance mechanism: what an organization measures signals what it rewards. That power deserves care, since narrow targets can invite gaming or neglect of unmeasured harms. A compact dashboard works best when supported by deeper analysis and regular review, not when it becomes a new ritual of oversimplification.
Analysis
Web Analytics 2.0 is best understood as a management argument disguised as a measurement manual. Its central claim is that analytics fails less often because of insufficient data than because organizations confuse data collection with learning. The proposed remedy is an evidence portfolio: behavioral traces, outcome measures, direct customer testimony, experiments, and competitive context. Each source has a different inferential role. Clickstream describes observed activity; surveys expose stated purpose; usability studies reveal friction; experiments estimate causal effects; competitive data supplies an external reference. The sources are not interchangeable, and their disagreements can be informative.
The book's strongest practical insight is its insistence that measurement definitions shape behavior. A team rewarded for visits will seek visits; one judged by task completion or margin per visitor has different incentives. This anticipates a central problem in organizational economics: metrics do not merely describe performance, they alter it. The remedy is not to abandon KPIs, but to connect a small set of measures to actual goals, inspect side effects, and revise definitions as incentives distort behavior.
Its limitation is historical as well as methodological. Published in 2009, the technical examples depend on vendors and practices that have since changed. Cookies, browser tracking, mobile platforms, and social distribution are now governed by different technical and privacy conditions. The durable lesson is therefore not to copy the named tools, but to preserve the underlying discipline: ask how data was generated, who is missing, and what decisions it can support.
The book is also appropriately skeptical of universal attribution formulas, though its proposed alternatives remain difficult to operationalize. A multi-touch credit model can describe a chosen accounting convention, but it cannot by itself identify incremental impact. Randomized experiments and well-designed market comparisons offer stronger causal evidence, while carrying their own constraints around scale, spillovers, and execution. The mature reader should treat analytics not as a machine that emits certainty, but as a system for reducing uncertainty enough to act, learn, and correct course.
Review Summary
Web Analytics 2.0 is highly praised for its comprehensive coverage of web analytics, offering valuable insights for both beginners and experts. Readers appreciate Kaushik's focus on actionable insights, clear explanations, and practical examples. The book is considered a must-read for digital marketers, providing strategies for promoting analytics in various organizations. While some find certain sections outdated or too technical, many readers highlight the book's enduring relevance and its ability to change perspectives on data analysis. Critics note its occasional repetitiveness and self-promotion, but overall, it remains a highly recommended resource for understanding web analytics.
People Also Read
Glossary
Web Analytics 2.0
multi-source digital analysisAn approach to online measurement that combines quantitative and qualitative data from a website, its customers, and competitors. It uses behavioral data to show what happened, outcome analysis to assess value, experiments and customer feedback to investigate why, and competitive information to reveal what else is happening in the market. Its purpose is continual improvement of customer experience and business results.
Multiplicity
specialized tools, combinedThe practice of using multiple tools and data sources for distinct analytical jobs instead of forcing every question into one platform or a single source of truth. A clickstream tool, survey, experiment platform, and competitive data source may each provide separate evidence. Their value comes from combining insights thoughtfully, not from expecting their numbers to match.
10/90 Rule
invest chiefly in analystsA resource-allocation principle for analytics programs. Roughly 10 percent of investment goes toward tools and implementation, while about 90 percent goes toward skilled people who interpret evidence and recommend action. The figures are a guiding ratio rather than a universal accounting standard. The central claim is that software alone cannot convert a flood of metrics into business insight.
HiPPO
highest-paid person's opinionAn acronym for Highest Paid Person's Opinion, used for a senior decision maker's preference when it is treated as a substitute for evidence about customers. The term describes a recurring organizational risk: a powerful person's design or marketing judgment can prevail without being tested. Experiments and customer feedback provide ways to evaluate such ideas against observed behavior.
Macro and Micro Conversions
primary and secondary outcomesA Macro Conversion is the website's main desired result, such as a completed purchase or a successful support task. Micro Conversions are other valuable actions that serve visitors or the organization, such as registering, downloading information, subscribing, or locating a store. Measuring both reveals that a website may create value for several visitor purposes, not only its most obvious transaction.
Voice of Customer
direct customer feedbackQualitative evidence gathered directly from customers through methods such as surveys, interviews, and usability studies. It helps explain motivations, unmet needs, confusion, and satisfaction that clickstream data cannot establish by itself. Feedback can be collected at the page level or across the overall site experience. It should focus on questions the organization can act on.
Nonline marketing
connected cross-channel marketingA view of marketing in which customers move fluidly between online and offline channels while researching, evaluating, and purchasing. A person might see an advertisement, research on a website, visit a store, and complete a purchase by phone. Measurement therefore tries to connect activity across touchpoints instead of assigning each channel an isolated role.
Brand Evangelists Index
customer delight scoreA customer-rating measure designed to reward strong positive experiences and penalize negative ones, rather than treating mere satisfaction as excellent performance. In the book's example, the score subtracts dissatisfied and very dissatisfied responses from very satisfied and extremely satisfied responses, divides by total responses, and multiplies by 100. It is intended to set a higher performance standard.
Latent Visitor Behavior
post-acquisition customer behaviorVisitor actions observed after an initial campaign response, signup, or purchase. Examples include whether an acquired visitor returns, how soon the next visit occurs, how frequently the visitor engages, or whether another purchase follows. Measuring this behavior can reveal whether a campaign created durable value rather than a one-time response.
FAQ
What's Web Analytics 2.0 about?
- Data-Driven Decisions: The book emphasizes the importance of using data to make informed decisions in online marketing and web analytics, shifting from faith-based to data-driven decision-making.
- Expanded Framework: Avinash Kaushik introduces Web Analytics 2.0, which includes qualitative data, competitive intelligence, and customer-centric strategies.
- Practical Solutions: It provides actionable insights and strategies for analyzing web data, measuring success, and improving online performance.
Why should I read Web Analytics 2.0?
- Expert Insights: Avinash Kaushik is a recognized thought leader in web analytics, offering valuable guidance for marketers and analysts.
- Comprehensive Framework: The book offers a framework to understand web analytics complexities and leverage data effectively.
- Charitable Contribution: Purchasing the book supports charitable causes, with proceeds donated to organizations like The Smile Train and Ekal Vidyalaya.
What are the key takeaways of Web Analytics 2.0?
- Data-Driven Culture: The book stresses cultivating a data-driven culture where decisions are based on web analytics insights.
- Outcome Focus: Emphasizes measuring outcomes rather than just metrics like visits or page views to understand the true impact of online efforts.
- Multiplicity Strategy: Advocates using various data sources and tools for a holistic view of web performance.
How does Avinash Kaushik define web analytics in Web Analytics 2.0?
- Comprehensive Definition: Web analytics is defined as analyzing qualitative and quantitative data from your website and competition to improve the online experience and achieve desired outcomes.
- Beyond Clickstream Data: It includes understanding customer behavior, competitive intelligence, and qualitative insights.
- Customer Experience Focus: The goal is to enhance the online experience for customers, leading to better business outcomes.
What is the 10/90 Rule in Web Analytics 2.0?
- Resource Investment: For every $10 spent on analytics tools, $90 should be invested in hiring skilled analysts, highlighting the importance of human resources.
- Complexity of Analytics: Tools collect data, but insights require analysts to interpret the data effectively.
- Focus on Talent: Prioritizes hiring and training talented analysts to maximize analytics investments.
What is the difference between Macro and Micro Conversions in Web Analytics 2.0?
- Macro Conversions: These are primary goals like purchases or newsletter sign-ups, directly contributing to business objectives.
- Micro Conversions: Smaller actions like resource downloads or product page views, indicating user engagement and leading to macro conversions.
- Importance of Both: Tracking both provides a fuller picture of user behavior and helps optimize website performance.
How does Web Analytics 2.0 suggest creating a data-driven culture?
- Transforming Company Culture: Emphasizes exciting people about analytics and making data accessible to all organizational levels.
- Actionable Reports: Advises creating reports that drive action and align with business objectives, ensuring data relevance.
- Empowering Employees: Fostering a culture valuing data-driven decision-making empowers employees to use analytics in daily work.
What is the Brand Evangelists Index (BEI) in Web Analytics 2.0?
- New Metric Definition: BEI measures customer satisfaction and loyalty by focusing on very satisfied customers while penalizing negative ratings.
- Formula for Success: Defined as: “{ [ (Very Satisfied + Extremely Satisfied) – (Not Satisfied + Not At All Satisfied) ] ÷ Number of Responses } × 100.”
- Cultural Impact: Shifts focus from meeting expectations to striving for excellence, fostering a customer-centric culture.
What are some critical web metrics discussed in Web Analytics 2.0?
- Bounce Rate: Measures the percentage of visitors leaving after viewing one page, indicating landing page engagement.
- Conversion Rate: Percentage of visitors completing a desired action, key for assessing website effectiveness.
- Engagement Metrics: Includes time on site and pages per visit, assessing how effectively a website captures visitor interest.
How does Web Analytics 2.0 address the challenges of multitouch attribution?
- Understanding Multitouch Attribution: Recognizes various touchpoints a customer interacts with before converting, complicating attribution.
- Different Models: Discusses models like last-click and first-click credit, emphasizing a custom approach aligning with business goals.
- Practical Recommendations: Encourages focusing on actionable insights rather than complex models for effective multitouch attribution.
What role does competitive intelligence play in Web Analytics 2.0?
- Contextual Insights: Provides context for understanding a company's performance relative to competitors, crucial for strategic decisions.
- Benchmarking Performance: Encourages using competitive intelligence tools to benchmark key metrics, identifying improvement areas.
- Driving Action: Leveraging competitive insights helps present compelling cases to management, driving action and investment.
What are some advanced analytical techniques mentioned in Web Analytics 2.0?
- Segmentation: Stresses segmenting data to uncover deeper insights and understand different visitor behaviors.
- A/B Testing: Discusses experimentation and testing to optimize website performance and improve user experience.
- Multichannel Attribution: Introduces multitouch campaign attribution analysis to understand marketing channels' impact on conversions.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.