Avoiding Paralysis by Analysis: Making Good Decisions with Your Data
Analyzing web data takes equal parts of one’s analytical and creative mind. The internet is a large place that is filled, not just with users, but with robots, crawlers, and many programs that can affect data quality. As analysts our goal is to use data to help us make decisions while staying mindful to all the noise in our data. Here are some helpful tips that I use when working with any client’s website data. Using these tips to find actionable insights in a sea of data can make you a better decision maker by giving you proof instead of using assumptions.
View data in segments, not aggregates
Aggregated data is of little use because it provides no context. Bucketing all your visitors into one pool gives nearly as much insight into your site’s performance – as measuring the number of cloudy days in the month. The first step on the road to meaningful website analytics is segmenting your data into smaller chunks that you can compare over time and identify points of concern.
If your web traffic falls suddenly in January by a large percent, it is important to identify which visitors stopped coming to your site. If organic and paid search traffic stayed level but direct traffic dropped, then you know where to put your energy. However, if they all fell equally, you may need more information to understand this trend. Additionally, you may notice that every January there is a dip in traffic. Comparing traffic in January 2014 to that in January 2013 may provide insight next year to boost CPC spending and keep revenue constant.
Creating custom reports can be helpful. They help you recognize what are the most important metrics for your site and weed out the noise from the hundreds of other lists. In addition, they reduce reporting errors by giving you a consistent report month over month.
Remember quality over quantity
From more clicks and page views to more visitors and impressions, too many site owners get caught up thinking that more is better. However, this is not the case. Quantitative metrics like impressions are analytics 1.0. Having an understanding of the quality of your clicks or what are your better-converting pages can save you a lot of money and stop you from chasing the White Rabbit into the Internet Wonderland. Bounce rate, exit rate, number of conversions, or time of site can help answer many questions that begin with ”why”.
One target of quantitative data that is easy to understand and take action on is your bounce rate by page. If you notice 10 people visited a page that has a 100% bounce rate, it behooves you to investigate that page, as it could contain broken elements or have a slow load time. If you are using this as a landing page, you would be wasting money.
Stay mindful of the middle of the list; whether it be pages, keywords, or referrals, the top ten seldom change. Keeping an eye on the middle twenty pages can help you understand where marginal improvement can be made with less effort. Most top referrals come from Google, and getting more referrals from this search engine could entail an extensive SEO effort. But if you see that referrals from Pinterest convert more often, than posting more pins might be a cheaper and less time-intensive way to earn more conversions.
Keep your eye on the prize
Goals, key performance indicators, or conversions all measure the same thing. Different tools may differ on their exact definition, but you can still use them to guide your site’s efforts. Having a goal gives you a way to test what does work or what does not work for your business.
E-commerce sites are easy to set up, but what if you do not sell anything online? Ask yourself why you have a website: do you need to generate leads or distribute information? Newsletter signups can be a conversion. Are you a blog? Get people to sign up for your RSS feed. Remember the reason you had for starting a website and find some way to measure it.
As you dive deeper into goal conversions it may be useful to start A/B testing. There are great posts online to help you get started and understand how to know if your test is conclusive or significant. A/B testing can be misleading if you do not understand the math behind your study.
Caveat Emptor: Analyst Beware
“There are three kinds of lies: lies, damned lies, and statistics.” – Benjamin Disraeli
Always question your data. Take percentages, for example. They are excellent tools to highlight changes in data; however they can be highly deceptive. A 200% change could be an increase from 1 visitor to 3 visitors or 10,000 visitors to 30,000 visitors.
Numbers can, and will, at times lie to you. The job of savvy data analysts is to distil as much truth as they can from data. Segmentation will help but your best defense is to find patterns in data. If you see that this month ten times the visitors came to your website, find out as much as you can about those visitors. Do not just accept that visitor count or assume it was because your SEO efforts finally kicked in, but find out what was the motivation of those new visitors. Be clear you understand bounce rate over exit rate before you start making assumptions about the quality of a page. And above all, do not overanalyze. With enough data you can tell any story you want, but too much data can lead you astray.
- Matthew Chess, Marketing Intelligence Analyst
The post Avoiding Paralysis by Analysis: Making Good Decisions with Your Data appeared first on Search Optics Blog.