The above article gives great insight into the benefits of (proper) Personas. However, I believe the author draws better insights regarding the shortcomings of analytics and A/B testing. Really good discussion on the ongoing debate of qualitative and quantitative intelligence gathering. Personally, I cannot stress enough the importance of the triangulation of data. This is the using qualitative data (eg: surveys, interviews, usability testing) and supporting these findings with the scale and validity of statistics. Additionally, the converse, explaining quantified phenomena with behavioral observations.
Certain metrics like Page Views, Time on Site, Pages Per Visit are still important metrics, but in the rapidly changing and competitive web, they require context to be useful. Our intelligence shouldn’t be restricted to standards established in a time before Google Analytics, Chartbeat, and Omniture. There are no excuses to not look beyond metrics defined decades ago by advertisers and marketers. We must always dig deeper into metrics that really indicate success and establish KPIs. The article below gives a great case to look beyond the face value of data:
Additionally, this HBR article gives some real cases of how mating quantitative methods and qualitative methods can make all the difference: