Multi-Touch Attribution Campaign Tracking with WebTrends

This article is a follow-up to the webinar

All web analytics platforms have some way of tracking marketing campaign performance usually out-of-the-box or with a little bit of set up. Generally they all do a pretty good job of this and provide key reports to make important business decisions about which campaigns to invest more money in, which to reduce spending on, and which to get rid of altogether. But often these decisions are made without insight into the whole picture. Why? The answer is simply because most campaign reports are set up in the industry standard way of attributing all conversions to the last or most recent campaign clicked. This is and has long been the industry standard, but it is time for a change as this method ignores the fact that people often go through multiple campaigns before converting.

So what other attribution options are there? And why wouldn’t I want to attribute conversion credit to the most recent campaign? – There are typically 3 options for campaign attribution:

  1. Last Touch (Most recent campaign)
  2. First Touch (Original campaign)
  3. Multi-touch (All campaign touches)

Technically there are two options for multi-touch attribution. One option is to give full credit to all campaign touches and the other option is to give partial credit to each touch. For example, if 3 different campaign touches resulted in a sale of $30 you could credit each touch with $10. But for the purposes of this article we will focus on the full credit option. As for the question “why wouldn’t I want to attribute conversion credit to the most recent campaign?” – this is not really the right question to ask. The better question to ask is, “Do I have the best possible insight into the performance of my marketing campaigns?” The answer to that question is almost always “no” if you are only analyzing a single attribution method. So rather than replacing industry standard last touch reports, adding first touch and multi-touch to your arsenal of reports is the best course of action.

Fortunately for WebTrends users, there has been a great method for gaining insight into all campaign touches for quite some time although a little work up front is necessary to gain the full power of this. If you are already doing basic campaign tracking within WebTrends then the visitor history table is already turned on and with minimal effort you can set up two new custom reports which report on the first touch campaign and all campaign touches respectively. To do this you need to make use of two features of the visitor history table and create two new custom dimensions, one based on WT.vr.fc (the fc stands for “first campaign”) and another based on WT.vr.ac (the ac stands for “all campaigns”). Once you have the dimensions set up you create custom reports using those dimensions and whichever metrics you want applied. To make things easier, copy the existing campaign ID report and just change the dimension to base the report on.

The “first touch” report ends up looking nearly identical to the existing campaign ID report but the rows of data will be different since the revenue and other conversion credit is applied to the first campaign that referred the conversion as opposed to the last.

Standard Campaign ID Report Sample
First Touch Campaign ID Sample

The “all touches” report is where you’ll notice more differences. You will see some or many (depending on the date range you have selected) rows of data that have multiple campaign IDs separated by semi colons. To view only the data that contains multiple campaign touches just filter the report by a semi colon.

Multi-Touch Campaign ID Report Sample

So what do you do with this information? What does it all mean?
Spending some time with this new data will likely reveal some patterns you never had insight into before. For example, you may notice certain campaigns appear to perform poorly according to your traditional last touch reports but the same campaign’s performance as a first touch is much better, or vice versa. Since the first touch report is so similar to the out of the box campaign ID report it is fairly straightforward. The only difference is that the first touch gets the credit. The all touch reports are more complicated though. What I find most useful about this report is the ability to determine a campaign’s total reach and compare it to its absolute reach.  Take for example campaign ID 32. In the above screenshots you will notice that this campaign ID has $63,441 attributed to it as a last touch campaign, $35,839 attributed to it as a first touch campaign, and $82,036 attributed to it when you search for it in the all touches report (See fig. 4 below). What this data is telling us in this particular case is that:

  • $63,441 in revenue was most recently referred by campaign 32
  • Only $35,839 in revenue was initially referred by campaign 32
  • But overall campaign 32 at least partially referred $82,036 in revenue

As you can see, there can be very significant differences in campaign performance depending on how you look at the data. Taking the easy way out and looking only at a single attribution method can lead to less than fully-informed decisions being made about your campaigns. What if you were relying solely on first-touch reports in this example? That could lead you to reduce your budget on campaign 32 when in reality it was performing much better than your first-touch report told you.

Multi-Touch Report Filtered by Campaign ID 32

Ok, so all that is well and good but manually analyzing campaign IDs one at a time is a lot of work! Yes it certainly is using the methods I just provided as examples. But there is a much better way to approach this. Taking things a step further we can export each of these reports and combine them together in Excel using the campaign IDs as our key values. What we want to end up with is something like the following which will allow us to analyze first, last, and multi-touch all within a single interface.

Multi-Touch Reporting in Excel Sample

In part two of this article I’ll show you how to set this all up in WebTrends. But for now, follow the steps discussed in this article to get these super handy reports in place so you’ll be ready for the next part.

4 thoughts on “Multi-Touch Attribution Campaign Tracking with WebTrends

  1. I should add that WT.vr.ac has two big flaws:

    1. Within a given row, the campaign names appear in no sensible order — definitely not chronological or consistent in any way. The order isn’t even dependent on the alphabetical order of the underlying campaign GUID.

    2. It’s entirely possible to see the same exact combination of campaigns listed more than once, but differently-ordered each time. Crazy. Requires more work to sort it out, yes.

    If WebTrends would just fix WT.vr.ac so that chronology is retained … wow, we advanced users could really take off with it!

    Al, since you apparently have WT 9, perhaps you can confirm whether WT.vr.ac still has those two flaws? If they have been fixed, I will personally fly out to Portland and take that engineer to dinner!

    • Great points Rocky. It would be great if the WT.vr.ac values were in chronological order so we could see a history of campaign touches in the order which they occurred. However, partly because of this flaw and also to simplify analysis, I prefer to export all the rows from the report I create from the WT.vr.ac parameter and extract the total “revenue contribution” of each campaign. So ultimately what I end up with is something like shown in figure 5. This lets me compare first click and last click side by side with “all clicks” or total “revenue contribution” which I find very helpful when analyzing campaign performance and especially helpful when first or last click reports suggest to consider lowering spend or axing a particular campaign altogether. The all touches report really helps me decide what kind of value the campaign really has.

      I plan on getting into this more on my follow-up post.

      Great question about the later versions of WebTrends. I am not sure if they have fixed the flaws of WT.vr.ac in 9.x or even 10.x but I definitely want to test this out now. I’ll be sure to post my results when I do but I don’t yet have access to version 10 (should be coming in a couple weeks though supposedly so my fingers are crossed).

      By the way, I remember reading your article http://www.webtrendsoutsider.com/2008/cool-custom-report-scruffy-campaign-attribution/ a while back and it was very helpful when I first started getting into multi-touch attribution in WebTrends. In fact, your article was how I first heard of WT.vr.ac so thank you! 🙂 Prior to that I had only created similar reports in Omniture.

      • So unfortunately it looks like the random order of the visitor history values still holds true with both versions 9 and 10. At first I had high hope as during my initial tests they were appearing in sequential order. However this was only true for multiple campaign visits during the same day. As my testing continued and my multiple campaign visits spread across multiple days, that is when the values became out of sequence. So I guess we still have to accept the limitations of this variable. Nice to see it is still alive and kicking even with version 10 though.

  2. Great topic. I’ve been meaning to get into this because of all the discussion happening right now over on the WebTrends user forum (forums.webtrends.com) but you beat me to it.

    What is fascinating to me is that apparently the WT.vr.ac parameter is still around. What version of WT are you using? (I am still back in the 8’s) This parameter has been omitted from WT’s documentation for years – see http://www.webtrendsoutsider.com/2008/cool-custom-report-scruffy-campaign-attribution/ which talks about its deprecated state back in 2008.

    I am hoping that WebTrends just never bothers to kill off this VH parameter. Since it sometimes seems to take forever for them to fix bugs or simple usability issues, it would be ultra-ironic if they remembered about WT.vr.ac’s deprecation and actually rushed to get it out of them. It is truly a diamond in the rough, although I haven’t figured out how to exploit it to the max. I’m looking forward to your next post.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s