Taking the Lumps Out of Monthly Data

Dealer site sessions

No sooner were the books closed on June 2014 when the excuses started coming out about last June being more favorable and not strictly comparable to this June, as well as comparisons of June to May. Holy cow, any month this year is not strictly comparable to the same month in the previous year. Comparisons from month to month are even worse; May had 31 days and five of those were Saturdays. Both sales and shopping activity are inconsistent across the days of the week, yet leaders need a way to compare performance without the added noise caused by calendar differences.

At Dealer eProcess, our business is dealer websites and other digital marketing, so we’ve created a tool for dealers that takes the lumps out of monthly data and allows for clean comparisons from month to month and for comparing to the same month in the previous year. Leadership demands the minimization of excuses and maximization of accountability. Incentivized pay plans further accent the importance of accurate performance gauges.

Below, we can see the variation in site sessions by day of the week. It’s important to have a sufficient sample size when performing this kind of analysis and using the results to build new tools. This chart included over 1,000,000 sessions (formally called visits) to dealer sites for each day of the week. Google Analytics is the industry standard and the only tool used in this data collection.

Dealer site sessions

The differences can be substantial. For total sessions, a site receiving 5,000 total sessions this October will need to improve performance 5.1% in order to achieve 5,000 sessions again in November. Conversely, if a dealer’s site received 5,000 sessions in June and the same number again in July, there needs to be some investigation into performance. Everything else being equal, July total site sessions should be 4.4% higher compared to June. Last year the difference for total sessions between June and July was even greater, 5.0%.

Separate tools are required for sessions when performing analysis based on the user’s device (mobile, desktop, and tablet). Similar tools can be created for vehicle sales, service revenue, and other critical variables that are inconsistent across the days of the week. Obtainment of sufficient quantity and reliability is the challenge. Manufacturers, larger dealer groups, and associations have advantages in these areas. With care, credible tools can be developed using aggregated data for use by individual dealers.

Some tools will require updating more frequently than others. The increased use of mobile devices is not simply replacing desktop sessions to dealer sites. Many dealers are experiencing increased sessions with a disproportionately high percentage of them coming on Saturdays. Although the shift is not dramatic, the near universal accessibility to the internet is changing when some shoppers visit dealer sites.

The alternative to weighting monthly data is the use of 28-day data. The chosen place to start with a 28-day rotation may be different for one group than another. Standards do not yet exist in this area. This is why dashboards from Dealer eProcess and many other vendors include the ability to perform customized date filters across all data sets.

Holidays are a problem for 28-day months, and more of a problem in some countries than in others. As an example, with just a few exceptions, major U.S. holidays are always split into a different 28 day period, just as they are split into different months. New Year’s and Christmas are the most notable exception. Veterans Day, Columbus Day, and Martin Luther King’s Birthday can cause a bit of confusion, although these holidays are less uniformly observed than President’s Day, Memorial Day, Fourth of July, and Labor Day. Thanksgiving is not a problem in the U.S., provided the 28-day period is always started on a Sunday or Monday, as is customary.

For most dealers, groups, and manufacturers, the use of weighted data through simple, reliable tools, like the one associated with this article, will prove to be the easiest way to go. However, one method or another must surely be utilized. Leaders can either gauge performance based on data that is controlled for calendar variations or allow time to be consumed by off-the-cuff explanations and excuses. Fortunately, the era of big data is beginning to yield some benefits in the form of easy-to-use tools.