It’s not enough to throw up a website and sit back and wait for traffic. You need to analyze your traffic, identify user behavior on your site, perform competitor analysis and find out who links to you. SEO isn’t even enough. Web data mining concepts help you store user data and create reports that focus on traffic analysis. This article is an overview of data mining concepts and how you can use data to increase website sales and improve online marketing ROI.
Web data mining is the process of logging small pieces of information from server logs, competitor websites and aggregators. Separately, this data doesn’t tell you much. Combined, however, this data can tell you a lot about user behavior and competitor habits. Web data mining helps you form the best web page layout, upsell customers, target specific traffic, include content that competes with other sites and perform link analysis. Web data mining requires large amounts of storage space, but the cloud gives you as much space as you need to mine terabytes of data.
What You Need for Data Mining?
To properly data mine, you need more advanced capabilities than simple data entry personnel. You need programs, bots, an enterprise database solution and the report queries to give you the right data. The resources you need also depend on the type of mining you want to do. There are four basic types of web data mining:
- Web Server: This type grabs data about website traffic. It logs the IP address, the page the user lands on and the time the user spends on the site.
- Application: This type of mining identifies specific user behavior within your application. For instance, you probably want to log the amount of times a user clicks a link, the pages the user navigates to and the products the user browses the most.
- Content Analysis: Content is king when you want your website to rank in search engines. What content sells well, what ranks in the search engines and what phrases are you targeting? Combined, you take these three components and identify the areas where your site competes well and where it fails. You can also use this type of mining on competitor sites to identify the phrases and products other websites target.
- Link Analysis: Links are essential for website traffic, advertising and search engine rank. There are several aggregator sites such as Majestic SEO and Ahrefs that already gather link data. However, you can also crawl websites with your own programs and gather link information from competitors.
Your Web Server Data
Your web server data is probably the easiest to log of the four mining types. You control the server and logs, so you have access to any data you want. You can turn logging off and on, so you also control the type of data you store. Web server data mining is the best for finding out user demographics.
You probably wonder what type of data you should store. The important information depends on your business and what you’re targeting. For instance, if you sell globally or even nationally, you might want to keep track of search engine traffic location. From what countries do you get most of your traffic? Which states bring in the most traffic and revenue? What time of day does your website get the most traffic? What days of the week are the most popular? What browsers do most of your users use? What are the browser’s language settings? How much mobile traffic do you get? You can then use this information to customize your user interface. You can also use the data to figure out if you should create a section that focuses on a certain language. For instance, if you get a lot of traffic from Germany, you should create a section with German content.
The amount of time the user spends on your site is also a good piece of information. Are users bouncing from your site without browsing further? Bounce rate is a serious issue for ecommerce stores, because it means that potential customers are leaving your site and possibly choosing a competitor. Bounce rate incorporates the time on the site and the user’s engagement. If the user opens your page from a search engine and immediately leaves, it’s a bounce. If the user finds interesting material and clicks a link on your site, you’ve officially engaged the user. This type of user behavior pattern is how you customize and tweak navigation menus and content links for the best user engagement.
Application Analysis and User Experience
What happens when a user opens your eCommerce store? What products do users browse? What items sell the best? What areas of your site do users prefer? All of these items are mined for better user experience. For instance, suppose you have a link on your main page that’s labeled “Widgets.” You also have a link a bit down the page that’s labeled “Red Widgets.” You notice that more people prefer red widgets over general widgets. You use this data to move the red widgets link up the page, so it’s more prominent to potential customers. This type of page layout customization based on data analysis will increase your sales potential.
Application analysis helps you identify usability for your site. Each user event is used to identify areas of your site that are popular with your website visitors. Popular areas can be promoted more often, and you use this data to strengthen weak areas of your website.
Third-party tools such as Google Analytics help you track links that users click. You can also manually log clicks from within your application. Log the links users click and the relevant landing page. You can use web server analysis with application analysis to form reports that identify user behavior patterns. Combined, this data helps you create a website layout that increases sales.
Content Analysis and Search Competition
What phrases do you and competitors target? It’s a simple question but not easy to identify the answer. The content you create targets specific search phrases. The problem is identifying the phrases that potential customers type into search engines. Google stopped passing search phrases to webmasters, so it’s even more difficult than it was before. You must be able to put yourself in the searcher’s shoes and figure out what he would type to find your website.
You can use keyword tools such as Adwords and WordTracker to come up with similar phrases from certain words. However, it’s up to you to create the content and figure out which phrases are best for site rank and searchers. You can also read your competitor’s site to identify any phrases that you’ve missed.
There are some keyword analysis tools that crawl a website and identify popular phrases within its content. These tools help you find common phrases and use those phrases to generate more ideas for your site. Some webmasters create their own programs to crawl and analyze a competitor’s content.
Backlink Analysis
Backlinks have gotten a lot of attention lately. Google has manual actions that it places on unscrupulous link-builders, and the Penguin algorithm can kill a website’s search engine rank due to low quality link patterns. Even with these issues, backlinks are still important factors for website analysis.
You can create your own crawlers for backlink analysis, but it’s easier just to use current third-party tools. These tools are expensive and cost hundreds of dollars each month. However, you get full link analysis including anchor text, the domains on which the links are located, and the number of links coming from a specific site.
Since backlinks are part of the ranking algorithm, you use this type of data to identify where your competitor advertises, any affiliate programs and any related competitor sites. You can also use these tools to find out who is linking to your own site. If you have affiliate programs, you can view affiliate sites to make sure they follow your guidelines.
Put all of these resources together, and you have massive web data mining information. You then use your database’s resources and queries to create reports. These reports are then used to find the best advertising, page layout and affiliates for your website. It sounds much simpler than it actually is. Web data mining is difficult and takes massive resources. You need the storage and the computing power. You also need to execute the right reports or you get the wrong information. However, having the data is the first step.