My last name is Gross, not Shyamalan, which should have been clearly illustrated above with my “…” cliffhanger. My lack of ability to twist a plot and the realization that table settings alone make for a crappy dinner party has compelled me to give you all (all three of you) a two for one introductory post. Lucky you.
Imagine a world where web analytics segmentation tools allow you to not only specify what a user saw/did but when they did it. Sigh. Too long have we been stuck within the three dimensions of page, visit and visitor. Give me the fourth dimension. One day the vendors will hear my demands/pleas, but for those who can’t wait and are using HBX or any other tool that allows you to create campaigns there is in fact a work around.
Ok—here is the part where I give my boss, Gary Angel, credit for not only teaching me this technique but also already blogging on this technique. If you have already read his posts on time based analysis you’ll find some repeated content but I have written this to provide some new insights and ideas for using the analysis.
At the most basic level the work around consists of campaign(s) + segment= date parsing, in that order. As with all good analysis, however, a sizable portion of thinking and planning should be spent before anything is built into the web analytics tool. Obviously this method can be effective for measuring the effectiveness and longevity of your keyword, banner, email or other similar campaigns (just make sure your landing pages have unique identifying parameters for each campaign). Additionally, do not overlook building campaigns based on essential pages or links within your site. If you have an on site tool (not to be confused in this post with your web analytics tool), especially one that is several steps, consider doing a time segment analysis to see if it is stimulating the change you expected it to. Trust me, it will be more interesting and telling than a funnel.
Once you have decided the content you want to utilize for your time segmentation analysis its time to build your campaigns. Unfortunately campaigns can only be built starting with a future data, so the soonest you can begin collecting data is tomorrow. Campaign building is a quick, easy process but it is important to not overlook your date range. Building a single campaign and giving it a start date of July 1st and end date of July 31st gives no options when it comes to segmentation and analysis. With a segment spanning from May 1st through August 31st you could see what the people who came in on a campaign or saw a specific page did before, after or during July. You would, however, have no idea if visitors responded to the same campaign or viewed the same page before or after July or when in July the page view/campaign entry occurred. It is more time consuming to build multiple campaigns for the same site page or specific entry page URL parameter, but utilizing a weekly or even daily granularity for campaigns will pay off. With weekly campaigns you could build a segment to see what people did on your site who utilize a tool during the first week of July and not access that tool in the four weeks before or four weeks after.
When your campaigns are complete and your segments have been built the rest is up to you, but I recommend keeping it simple. Usually I will pull data from the month before tool access to establish a baseline then weekly data from the week before at the four (or more) weeks after the tool is used. Looking at a weekly granularity post on site tool use, campaign response or page view will give you details on the residual effect, diminishing returns, life span or acquisition cycle that you cannot see any other way. There are few, if any, instances where you will need to build KPIs unique to the time based segment, for the most part your most basic and essential KPIs should be telling. Be careful not to only look at metrics relevant to what you expect the campaign, tool or page to do, indirect benefits are always fun to pop into a PowerPoint for the higher ups.
Questions? Comments? Hate mail? Suggestions? Shoot me a message at jgross@semphonic.com.
Comments