Most people understand that testing is important, but as Conrad Saam of Avvo tells WebProNews, properly testing is the only way to get successful results. According to him, one of the biggest mistakes that SEOs make is that they rely too heavily on averages.
Saam says that everyone within a business needs to be on the same level with statistics. If they are not, smart decisions cannot be made.
He advises SEOs to conduct a Binary Test. In medical terminology, this would mean that a patient either lives or dies. In other words, there is no middle ground between the two. Saam says online calculators are very helpful since they provide users with the required sample size for testing. To utilize these calculators, he suggests doing a Google search for an A/B calculator for confidence intervals.
The second test Saam recommends is a T-Test. This test consists of an array of data with some variability associated with it. In SEO, he says this test could be a ranking improvement of 1-10 or 1-100. A T-Test gives users the statistical confidence level based on two different collections of data, which determines if “A” is better than “B.” Avvo uses Excel to conduct this test.
Although some people believe that more tests are needed, Saam doesn’t necessarily agree. For example, he says what happens if you survey 20 people and find that the 10 men surveyed are 4′ taller than the 10 women surveyed? If this were true, most people would feel comfortable saying that men are taller than women. However, if the survey found that the men were 4″ taller than the 10 women, then people would be less comfortable saying men are taller than women.
“Depending on that variability on the data, you need different levels of sample size to have confidence in your decision,” says Saam.
Is your testing helping you make smart decisions?