Three Approaches to Evaluating Content Success in SiteCatalyst

If you’re cur­rently mea­sur­ing var­i­ous kinds of out­comes on your site (orders, leads, micro-conversions, etc.) with SiteCatalyst’s cus­tom events (or default ecom­merce met­rics), you will want to know which con­tent on your site is influ­enc­ing them (suc­cesses or fail­ures). Some con­tent may be pulling more than their fair share while other con­tent may just be get­ting in the way of con­ver­sions. The term, “con­tent”, can span every­thing from pages to site sec­tions (groups of pages) to page types (e.g., arti­cle page, cat­e­gory page, prod­uct page, etc.), etc., but for the pur­poses of this post I’m going to stick to the stan­dard HTML page when I say content.

There are three dif­fer­ent approaches you can use to deter­mine con­tent effec­tive­ness (or inef­fec­tive­ness). The right method for mea­sur­ing con­tent may depend on the type of out­come you’re mea­sur­ing as well as the dif­fi­culty of imple­ment­ing the approach for your par­tic­u­lar site. Let’s run through the three dif­fer­ent options.

1. Event Par­tic­i­pa­tion (Indirect)

First, one of the best ways to eval­u­ate con­tent per­for­mance in Site­Cat­a­lyst is to use the event par­tic­i­pa­tion approach. Event par­tic­i­pa­tion essen­tially looks at the down­stream, indi­rect influ­ence of pages on sub­se­quent con­ver­sion events hap­pen­ing dur­ing the same visit. In the exam­ple below, each page that pre­ceded the order (rep­re­sented by the shop­ping bag) receives full credit for the pur­chase. Any pages viewed after the order wouldn’t receive any credit for the purchase.

On its own, the par­tic­i­pa­tion met­ric will tell you two things: what per­cent­age of the total events the page influ­enced and how many total events the page influ­enced. While this infor­ma­tion is use­ful, it will be skewed towards high-traffic pages (e.g., home­page), pages where the event actu­ally occurs (e.g., con­fir­ma­tion page), or multi-step process pages that always pre­cede an event (e.g., billing/shipping page, lead form page, etc.).

In order to find the golden nuggets amidst the ever-growing pile of pages on your site, you will want to cre­ate a cal­cu­lated met­ric which divides the par­tic­i­pa­tion met­ric by vis­its. Now you will have the aver­age rate in which the page influ­ences a par­tic­u­lar out­come if it is seen dur­ing a visit. Cer­tain pages will always out­per­form other pages due to their pur­pose or posi­tion; how­ever, the par­tic­i­pa­tion rate puts the low-traffic and high-traffic pages on more equal footing.

In a use case exam­ple, you might have sev­eral case study pages on your site and you’re try­ing to gen­er­ate trial soft­ware down­loads. Using the cal­cu­lated met­ric (trial down­load par­tic­i­pa­tion / vis­its), you can com­pare the dif­fer­ent case study pages side-by-side to deter­mine which ones have a higher rate of influ­ence on trial down­loads. If you iden­tify clear win­ners and losers amongst the case study pages, three actions can be taken.

First, you can fea­ture high-performing case stud­ies in more promi­nent posi­tions (e.g., home­page) to give them more expo­sure, which may lead to more trial down­loads. Sec­ond, you can rework low-performing case stud­ies to improve their per­for­mance. Third, you can remove the weaker case stud­ies if they just aren’t res­onat­ing with site vis­i­tors, which will mean fewer dis­trac­tions from your top-performing case studies.

Note: If you are inter­ested in event par­tic­i­pa­tion, you will need to have your account man­ager or con­sul­tant enable event par­tic­i­pa­tion for spe­cific cus­tom events and traf­fic vari­ables. Once event par­tic­i­pa­tion is enabled, it is not retroac­tive — it only rolls for­ward from the moment it was enabled. If you have Dis­cover, you auto­mat­i­cally have par­tic­i­pa­tion for all of your cus­tom events.

2. On-Page (Direct)

While event par­tic­i­pa­tion mea­sures the indi­rect influ­ence of a page or other con­tent on suc­cess events, you may want to know which spe­cific page has directly influ­enced an action or out­come. For exam­ple, you may want to know which pages are gen­er­at­ing video views or prod­uct ratings/reviews. In these sce­nar­ios, you don’t want any other pages to get equal credit for the action other than the page that the vis­i­tor was on when he or she insti­gated the action such as click­ing a link or but­ton. In the exam­ple below, the vis­i­tor passed through three pages before finally trig­ger­ing a video view on the fourth page (Page D), which is the only page that receives any credit for the video view.

In order to mea­sure the direct influ­ence of pages, you need to cap­ture the page name in an eVar on every page load. When­ever a cus­tom event fires, it will be attrib­uted to the last page that was loaded. If you would like to know at what point in the visit the vis­i­tor trig­gered the event, you will want to set a counter eVar that increases by one unit on every page load. This vari­able will give you a his­togram of the num­ber of page views that occur before the cus­tom event hap­pens. You can then man­u­ally cal­cu­late the aver­age num­ber of page views before some­one trig­gers the action in question.

In terms of direct influ­ence analy­sis, you’ll want to cre­ate a cal­cu­lated met­ric of the intended action divided by vis­its so you can com­pare the event con­ver­sion rates of var­i­ous pages. In addi­tion to com­par­ing indi­vid­ual pages against each other, you might also con­sider exam­in­ing dif­fer­ent groups of pages to spot action­able pat­terns or trends. For exam­ple, you might find that peo­ple are view­ing videos more fre­quently for a par­tic­u­lar prod­uct cat­e­gory (e.g., video games) or on a new page tem­plate that was just intro­duced. Based on these insights, you might con­sider cre­at­ing videos for other prod­uct pages within the same cat­e­gory that don’t yet have videos or adopt­ing the new page tem­plate more broadly on the site to increase over­all video views.

3. Pre­vi­ous Page (Origination)

In a related but slightly dif­fer­ent sce­nario, you may want to know which page orig­i­nated a spe­cific action such as an inter­nal search, a lead form start, or the use of a store loca­tor. Some­times it can be dif­fi­cult or impos­si­ble to track cer­tain actions with on-click trig­gers so the cus­tom event is set on the sub­se­quent page where the action actu­ally occurs. In the fol­low­ing exam­ple, inter­nal searches would be trig­gered on the search results page (Page D). How­ever, it’s more use­ful to know which page (Page C) the vis­i­tor was on when he or she ini­ti­ated the search query.

In these “look-back” sce­nar­ios, you will need to use the get­Pre­vi­ous­Value plug-in, which main­tains or per­sists the page name value through to the sub­se­quent page. Using this spe­cial plug-in, you can place the pre­vi­ous page name (or site sec­tion, page type, etc.) into an eVar vari­able. If you’re ana­lyz­ing con­tent per­for­mance, you’re going to cre­ate the same cal­cu­lated met­ric (action met­ric / vis­its) so that you can iden­tify low– and high-performing pages.

If you were to use the pre­vi­ous page approach for store loca­tor usage or lead form starts, pages with high actions per visit would be viewed favor­ably as the con­tent is dri­ving poten­tial offline sales through the store loca­tor or lead form. In the case of inter­nal searches, you will be look­ing for pages that have high inter­nal searches per visit, which indi­cates that the pages are fail­ing to present the right con­tent. Vis­i­tors are not find­ing what they expected and are being forced to search for what they really wanted. In the exam­ple below, the brack­e­tol­ogy page appears to be hav­ing some issues as it is expe­ri­enc­ing a higher than aver­age searches per visit.  Tip: With some addi­tional tag­ging that is rel­a­tively sim­ple, you will be able to tie search terms to spe­cific pages. You will then be able to see what types of key­words are being searched for, which will give you hints as to what infor­ma­tion may be inef­fec­tive or miss­ing from the page.

As you can see, there are many ways to ana­lyze the influ­ence of con­tent on suc­cess events, and dif­fer­ent approaches can yield dif­fer­ent insights. Hope­fully, shar­ing these dif­fer­ent approaches has gen­er­ated some new ideas for how your com­pany can bet­ter under­stand which con­tent is and isn’t con­tribut­ing to your online suc­cess. Good luck!