I applied online. The recruiter contacted me right away and scheduled a phone screening, which included an overview of your resume, a few basic math problems (weighted averages, break-even quantities/prices, etc.), and your salary expectations. After that came an online coding challenge through HackerRank, in which you are tested on your scripting skills in SQL, Python, etc. A new (at the time) online Data Challenge assessment followed, for which you are given a hypothetical business situation with some datasets, and tested on your working knowledge of open-source software and business analytical skills. I enjoyed the flexible aspect of the data challenge, and was invited for an on-site interview in Richmond, Virginia.
The in-person interview was very similar in structure to many others described here. Behavioral + Case Studies + Job Fit (discussing your Data Challenge submission). Throughout the interview, I felt that their online job descriptions did not reflect clearly the type of candidates they were looking for. The descriptions were ambiguous at times, and sounded very generic (SAS, SQL, data queries, procedure automation, reporting, etc.) The company appears to have updated these online posting descriptions since, but they are clearly looking for computer engineers with software development backgrounds, rather than conventional data analysts with database management and analytical skills. They told me that we were among the first to interview for "a new generation of data analysts", which sounded encouraging, but at the price of the company actively moving away from licensed software towards open-source technology. So if your strongest selling points are, for example, SAS, Stata, or MATLAB, do not apply, this is not the right company for your talents.
Overall, the interview experience was good. The interviewers explained very well the different aspects of the job, the company, and the direction it's headed. I felt that Capital One really wants to lead the technological developments in the banking field. This is a great vision, but unfortunately leaves very little opportunities for licensed technologies, their dependable processing capabilities, and those that have invested heavily in these skill sets.