7 biggest mistakes in web analytics for online shops
Introduction
Anyone who wants to run an online shop profitably must keep a close eye not only on their market position, but above all on costs. Generating revenue alone is not enough to ensure a positive bottom line. In many cases, it is precisely those costs that are not immediately visible at first glance that place a long-term burden on profit.
With this article, we want to shift the focus of economic success toward the different types of costs faced by online shop operators. In e-commerce controlling, this area is commonly summarized under the term web analytics.
From our perspective, we have compiled the 7 biggest web analytics mistakes for you.
At the beginning, we revisit selected topics from classic controlling. We have often seen that important costs are not included in the profit-and-loss calculation at all—leading to flawed web analytics. However, analyzing and evaluating classic financial KPIs is the foundation of efficient web controlling and must therefore come first.
Other key questions we answer in this article include:
What exactly is web analytics and how is it defined?
How can it help me measure the revenue–cost relationship in my online shop more effectively?
How can I increase profit through efficient web analytics, or reduce the costs required to generate revenue?
Which areas should I pay particular attention to in web controlling in order to achieve these goals?
In addition to answering these questions, we illustrate each point with numerous case studies from our day-to-day work and discuss the options provided by the many tools now available on the market for analyzing and evaluating individual metrics.
1. Classic Controlling Is Not Used as the Starting Point for Web Controlling
As already mentioned, to describe the individual web analytics mistakes accurately, we first need to take a look at classic controlling. Measuring and interpreting conventional financial KPIs—such as revenue per order or profit per item—is the foundation of any web controlling. Without a precise overview of all costs in your online shop, web analytics becomes meaningless, and optimization goals for your shop are difficult to define.
At first, it comes down to a seemingly trivial question: “What am I actually earning from this?” As obvious as this may sound, we unfortunately see again and again that precise answers cannot be provided even here.
Using two practical examples, we want to illustrate the impact that insufficient consideration of conventional KPIs can have. After that, we summarize the most important KPIs in classic controlling once again.
Case Study 1: The Costs Behind Revenue Are Not Broken Down Accurately per Order
As a shop operator, you know how essential it is to view costs in relation to generated revenue. Nevertheless, we often see that many factors are not included in the calculation—or that cost components are not allocated down to the individual order or item sold.
The costs of successfully selling an item do not end with warehouse costs or expenses for your chosen logistics provider. In addition to further personnel and packaging costs, it is especially return-related costs—and the resulting additional personnel, packaging, and logistics costs—that are often simply forgotten.
A precise breakdown of all costs should therefore be performed for each order—ideally in an automated way. Once this has been done, items that produce many returns can be removed from the range, pricing can be adjusted, or other measures can be implemented.
Image: none
Figure 1 shows that in 2014, the return rate—especially for low-priced items—was over 20% for more than one third of surveyed online shops. A closer look at return costs in this segment is therefore essential.
Case Study 2: Revenue Increases, but Profit Does Not Increase Proportionally
A second example of insufficient cost consideration often arises in what is actually a positive development: increased revenue. But here too, higher costs can melt away profits, and the overall result may turn negative.
There are many different reasons for this, often reinforcing one another. If revenue increases relatively suddenly—for example due to additional seasonal sales—we often see with our customers that the higher volume of orders can no longer be processed at the usual quality level.
A vicious cycle emerges: the increased number of orders can only be handled with more picking staff, who are not initially available and must also be trained by your existing employees. Customer service inquiries rise, requiring additional staffing there as well. In that area, highly qualified staff is often needed, which involves greater recruiting effort than hiring for logistics. Especially in high-priced and technical product segments, customers expect extensive advice before purchase—which you must still deliver at the same quality level even with rising call volumes.
The problem: costs per order can rise dramatically. New customers with many questions may be lost—or, even worse, regular customers may no longer receive the service level they are used to.
The solution: implement classic web controlling that captures all costs and allocates them down to the individual item or order. Such classical accounting is the foundation for all subsequent web analytics.
We often experience consulting conversations where it gradually becomes clear that these key fundamentals are not consistently integrated into daily analysis. As a consulting partner, this then often forces an extensive “trace-back,” which costs more resources across all areas.
For this reason, we have compiled the most important classic controlling factors for your online shop once again. This list does not claim to be complete, and priorities may vary depending on your individual needs:
The Most Important KPIs in Classic Controlling
Considering delivery costs
What does it cost me to ship a single order including all costs (packaging, logistics, warehousing, etc.)?
How long does processing an order take—i.e., what personnel costs are incurred?
Considering the return rate
Costs for additional packaging, staff, and logistics
What does processing a single return cost me including all costs?
What is my inventory turnover rate?
Can I sell purchased goods immediately, or do I store them for extended periods on average?
Do I have many seasonal products that I can usually only sell at a loss after the season?
2. Web Analytics Is Not a Regular Part of Daily Operations
Now that we have laid the foundation using conventional controlling, we can start building the “web analytics house.” Continuing this metaphor briefly: this mistake is about the ongoing monitoring of that construction. Even the most extensive web analytics loses its effect if measured data is not interpreted regularly and integrated into daily work.
Permanently installed web analytics tools not only reduce your workload but can also inform you—if desired—via email with daily figures and reports for all defined KPIs. That way, you can start the day with controlling firmly in hand—right from your first cup of coffee.
Below is an overview of the key tools on the market for effective and sustainable web analytics:
Google Analytics
Google’s own tool is by far the most widely used analysis instrument among free tools. It stands out for its virtually endless range of metrics. By now, Google also offers measurement criteria that used to exist only in paid solutions. Among other things, the entire customer journey can now be analyzed (see mistake #5) and even the share of returns per product can be measured.
In addition, Google Analytics allows you to customize dashboards to your needs and send your defined KPIs to you daily via email.
Image: none
PIWIK
This open-source program has established itself largely due to its strong data security. This is achieved by storing sensitive log data on your own server. A customizable dashboard offers a high degree of personalization, which can be extended even further through the open source code.
These two tools account for the vast majority of users in free web analytics. In the paid segment, the following two tools have proven themselves:
Image: none
Econda
The German provider’s extensive toolbox is specifically tailored to e-commerce requirements. Econda’s major advantage over other tools lies in its use of both aggregated and raw data. This means you get up-to-date figures that are not condensed or extrapolated.
Another benefit is that even TV or radio spots can be measured. Of course, the Econda dashboard can also be tailored to your needs and can notify you by email about current developments in your shop. Predefined reports additionally simplify the selection of metrics.
Image: none
Adobe Analytics
The tool from the US software giant is integrated into its Marketing Cloud. This suite of web analytics and online marketing tools allows you to monitor all activities of your online shop—from tracking visitor flows to determining the optimal timing of Facebook posts.
However, the high costs quickly reduce the attractiveness of these benefits—so Adobe’s suite remains primarily relevant for the enterprise segment.
With all the tools available on the market, you should not lose sight of the core insight: the benefit of web analytics software is not an end in itself. Even the best tools are only supporting instruments for evaluating KPIs that you have defined beforehand.
For this reason, in our third point we will highlight the importance of clear target definitions for web analytics.
Image: none
3. There Is No Clear Goal Definition for Web Analytics, or KPIs Are Misinterpreted
For every web analyst, observing and interpreting key KPIs using the tools mentioned above is part of the daily routine. Unfortunately, we still often see that either the wrong numbers are interpreted (due to missing goal definition) or the right KPIs are interpreted incorrectly. We want to examine both issues more closely.
1) Missing goal definition
To optimize your online shop continuously and efficiently, the first question you should ask as a shop operator is:
What do I actually want to achieve with web controlling?
A real goal definition is still too rare—or is not consistently placed before analysis. As in many other contexts, the desired state must be defined before solving a problem. Only then can improvement measures be defined and implemented.
An example illustrates this: every shop operator would likely agree if we suggested optimizing their traffic. That sounds plausible at first—but on closer examination it is rarely the true goal. Higher traffic often means more bandwidth usage, potentially slower load times, and often higher advertising spend.
The real objective is usually hidden behind this and often consists of generating leads or direct sales that should be increased.
This example is just one expression of defining a shop KPI or solving a problem. Below, we list what we consider the most important KPIs. Please note that a combination of multiple KPIs is typically required and that each one should be checked against your individual goal definition:
Quantity of visitor flows: the pure number of visitors to your shop—often referred to as traffic.
Quality of visitor flows: extends the view beyond quantity; includes conversion share relative to visitors, total visit time in the shop, or visit time per page.
Conversion rate: measures orders relative to total visitors. This alone shows that traffic by itself is not a solid metric for evaluating economic success.
Return on Investment (ROI): a classic controlling KPI describing profit relative to invested capital. In e-commerce it is often used to evaluate marketing activities. For example, landing page costs can be compared directly with the profit generated by that page—allowing you to assess marketing spend effectively.
Bounce rate: indicates which visitors leave a page without further interaction. Bounce rate can be determined for each URL and offers insight into acceptance of individual shop pages. There is no universal benchmark, as pages that collect information for checkout tend to have higher bounce rates than pure product overview pages.
Number of returning visitors: extremely important for assessing how well the shop resonates with regular customers. Shops selling trend goods or everyday items typically show higher returning visitor shares than shops selling high-priced heating systems or swimming pools. This KPI indicates overall satisfaction and whether your shop reaches a sufficient number of returning visitors (as defined by you).
Cart analysis and checkout abandonment rate: for many of our clients, this area offers huge optimization potential if interpreted correctly. KPIs in this category measure direct customer purchase behavior. If traffic is satisfactory but cart values seem low, cart analysis can trigger optimization of accessory products for specific items. We will address this in more detail in mistake #7.
2) Incorrect interpretation of KPIs
A second aspect is incorrect interpretation of important KPIs. Even when essential web controlling metrics have been defined, we often see imprecise reading of the data, leading to wrong conclusions for shop optimization. Another real-world example illustrates this.
Case Study 3: Misinterpreting the conversion rate
In many cases, conversion rate is treated as the ultimate measure of a shop’s economic success. After all, the ratio of orders to visits seems to provide a universal statement about quality, presentation, and pricing—of products and the shop overall.
However, we recommend questioning this supposed universality regularly and comparing the value within the correct industry context. For example, conversion rates in the grocery sector can be relatively high—often above 6%—due to a high share of repeat buyers purchasing the same products repeatedly. In contrast, conversion rates for high-priced goods such as heating or sanitary systems, often purchased only once, can fall below 1%.
This shows how important context and correct interpretation of the numbers output by tools really are. Unfortunately, we repeatedly see clients comparing conversion rates against the wrong industry benchmarks, drawing false conclusions, and optimizing areas that might actually be lower priority.
Similarly, a drop in conversion rate is often immediately interpreted as alarming. But the calculation basis is frequently forgotten: a lower conversion rate can also result from rising traffic—an actually positive effect. If sales do not increase at the same pace, conversion rate falls. In such a case, you should identify what caused the higher traffic. Are early marketing measures starting to work on specific channels—Google Shopping, price comparison portals, etc.? Once that’s clarified, you can analyze the (still missing) purchase completions.
This way you “kill two birds with one stone”: you identify which measures caused visitor growth (which marketing activities are working), and you see which sales channels increased conversions—or where conversions remain absent.
4. No or Insufficient Customer Profiles Are Created
Creating customer profiles allows qualitative assessment of users and goes beyond pure quantity. This area is also referred to as Web Analytics 3.0 and represents an interesting advancement of classic controlling and web analytics.
This part of web controlling is often neglected entirely because access to robust qualitative data can be quite complex. Simple user tracking is no longer sufficient if you want meaningful insights about customer motivations and expectations.
Solutions for collecting qualitative data are offered, for example, by the Hamburg-based company eTracker with its tools Visitor Voice and Page Feedback. These allow you to enter into direct dialogue with customers, learning more about their goals and wishes while also checking whether your defined target groups are being addressed effectively by your shop’s content.
Technically, this can be implemented through a questionnaire shown when visitors close certain pages—for example, the order confirmation page. Data privacy receives the highest attention. Assuming user consent, scientifically grounded market research methods can collect reliable data. The evaluation is anonymized but still yields dependable insights for optimizing your online shop. With several thousand visitors per day, the likelihood of statistically meaningful data is relatively high.
Case Study 4: Combining classic web analytics with integrated market research
A strong example of such implementation can be found in the clothing shops Outfittery and Modomoto. They show that combining classic web analytics with integrated market research can produce entirely new e-commerce sales strategies.
The customer no longer has to search—they are offered a tailored solution. You no longer wait for the customer to say what they want; instead, you show them what they (apparently) need.
Customer profiling and behavior pattern analysis can of course be expanded in many ways. eTracker is only one solution among many. For example, Econda also offers customer-centered analyses via its Centricity toolbox, allowing you to generate groupable behavioral patterns.
Customer profile creation—more specifically, behavior pattern analysis—should be examined even more precisely when many different online marketing channels are used. This is summarized under the term customer journey.
5. The Customer Journey Is Considered Only Insufficiently
In our experience, most online shop operators focus on the analysis of the final contact before purchase completion and, based on this, derive budgets for entire online marketing campaigns aimed at improving product visibility.
We advise everyone to track the customer’s entire journey—from information intent to purchase intent. If you want to deploy your ad budget as efficiently as possible, you must first identify which paths customers take on their personal customer journey until conversion. Essentially, you need to find out which touchpoints were used in which phase of the information and purchasing process.
Possible touchpoints include:
Review platforms
Price comparison engines
Blogs
Social media channels
Organic search results pages (SERPs)
Google AdWords campaigns
To evaluate this information and purchasing process precisely, marketing experts often use the AIDA model, which describes phases before the direct purchase. The idea is that different information is needed at each stage—to capture attention, build interest, create desire, and ultimately trigger action (ideally a lead or conversion).
Ideally, each phase has its own touchpoint, served by the optimal marketing instrument. But to determine this, you must track and analyze all customer touchpoints. Only then can landing pages or placements in price portals deliver the most economically effective results.
Case Study 5: Example customer journey for “buy heating system online”
Let’s assume you operate a shop selling heating and sanitary systems. You specialize in heating systems costing several thousand euros on average. It’s obvious that such an investment is rarely an impulse purchase. We repeatedly find that many channels are used for information gathering and price comparisons.
As an example: the first access (information phase) might occur via entering the search phrase “buy heating system online” and browsing organic SERPs or Google AdWords ads. In a second comparison phase—after some time—the search term might be refined, for example: “buy gas condensing boiler online,” and products are compared. Then intensive price research may follow through price portals, such as searching for “Viessmann Vitodens 300-W.” Only the fourth step might lead to a direct visit to your shop.
If you only track direct shop visits, you lose valuable insights into the success of SEO, paid search advertising, or product placements in price portals. Viewing the entire customer journey enables reliable statements about the effectiveness of each marketing measure.
6. Conversion Costs for Marketing Campaigns Are Not Considered
So far, we have seen how extensive web analytics is and how many areas it can cover. In addition to evaluating classic KPIs, effectively measuring marketing activities has become essential—especially against the backdrop of growing cross-channel and omni-channel selling.
Here we want to focus explicitly on measuring the costs of marketing measures—also known as conversion costs.
The mistake often lies in insufficient consideration of campaign costs. Similar to the financial KPIs discussed in mistake #1, costs of marketing activities must be related to achieved sales. This includes the cost of Google AdWords ads, banner advertising on other portals, and similar measures.
Here too, it is advisable to break costs down per conversion to get a clear overview of “advertising-intensive” products.
Case Study 6: Costs of price portals — product feed optimization
Another example is the cost of price portals. Here, an item-level profitability measurement is recommended—relating portal costs to achieved revenue for each product in your assortment.
If configured according to your targets, a defined minimum margin per product is not undercut. This means you do not automatically sell your entire assortment via one or all portals—instead, you identify the most economical price portal for each product.
A suitable tool for monitoring price portals is Channel Pilot, which has established a strong reputation. Based on defined criteria, each product is checked daily for profitability in price portals. This not only reduces workload but can also lower conversion spending per marketing channel effectively.
Case Study 7: Customer Lifetime Value — the decisive factor in e-commerce
Customer lifetime value (CLV) measures how much value a customer generates over the long term—i.e., how much money you can earn from one customer over the entire business relationship.
A common mistake is ignoring CLV when planning acquisition spend for new customers. Large mail-order businesses have integrated this KPI into their web analytics for some time and increasingly accept the risk of deliberately taking losses on early conversions—sometimes only turning a profit from the third purchase onward. The goal is to increase the share of repeat customers who generate regular orders and create a stable revenue source.
For accurate measurement of acquisition costs per new customer, it is important to subtract repeat purchases—meaning you need to exclude existing customers from the calculation. This necessarily requires that you know your share of returning customers.
You can then relate this value to the profits generated by repeat customers and determine how high acquisition spend is allowed to be. This value can, of course, also be integrated into cost calculations per order or customer.
7. Cart and Checkout Abandonment Are Not Analyzed
This last point moves away from costs and marketing channel measurement and focuses on a process that is an essential daily part of your online shop: analyzing cart and checkout abandonment—the checkout process.
Surprisingly, this process often receives the least attention. Perhaps because it is “at the end” of a purchase journey not only for customers, but also in the thinking of many shop operators.
Yet the importance of checkout speaks for itself. At the moment cart and checkout abandonment happens, the customer has already found you in the “online shop jungle” and your products have already sparked interest. Every abandonment hurts because a potential buyer drops out—someone who, apparently, was already ready to purchase from you.
What happened?
Checkout is like a bottleneck that every customer must pass through as smoothly as possible. After successful cart filling, this is where the highest drop-off rates occur.
Before optimizing checkout, however, analyzing where and why users drop out during the order process is crucial. For example, if abandonment is evenly distributed across all checkout steps, one reason may be that there are simply too many steps to complete. The main goal would then be to streamline checkout.
But let’s take one step back: even when this area is analyzed at all, many analyses fail to differentiate between cart abandonment and checkout abandonment.
The main difference lies in the stage of purchase completion:
Cart abandonment occurs when items are added to the cart but the customer leaves suddenly. Reasons are often individual: the total amount may be too high, the customer may still be comparing alternatives, or they may already expect information such as shipping costs or payment methods that are only shown later.
Checkout abandonment occurs when the cart is filled and the buyer starts the checkout process but then cancels. Here we repeatedly see technical factors playing a major role—especially limited payment methods. For example, if invoice payment is not available, trust requirements increase and customers are often less willing to add additional items. Another factor can be the amount and type of data required, such as date of birth or phone number.
This even creates a conflict between web analytics and checkout. While the average customer age can be valuable for customer profiling and birthday newsletters, the willingness to provide a date of birth is often extremely low.
As you can see, analyzing cart and checkout abandonment is highly multifaceted. When you want to identify reasons for individual drop-offs, effective tools are helpful. Google, for example, offers a funnel visualization showing at which points what percentage of potential buyers drop out.
Image: none
Conclusion
You can therefore also place processes that actually represent the final stage of a visit and order journey at the center of your attention when optimizing conversions.
Web analytics helps you identify the pitfalls behind your shop’s abandonment rates precisely and then implement initial optimizations efficiently.
In addition, you gain the ability to measure the economic success of your marketing measures accurately and develop targeted strategies for promoting your products.
Despite all the advantages of tool-supported web analytics, you should not ignore the calculations behind it. Always ask yourself what is being measured with which figures—and align your interpretations accordingly.
We hope we have provided a small insight into the world of numbers and metrics. Perhaps you recognized yourself in one or two of the case studies.
Sources
Image credit: Adobe Stock / tadamichi