If you agree that branded queries, similar to direct traffic, primarily occur as the result of a searcher learning of your brand through a previous marketing effort, communication, etc. then the default last non-direct attribution model within Google Analytics could make optimization of your non-branded channel a bit more complex. If branded search is truly the result of a previous marketing effort, such as a non-branded search campaign, then it’s important to properly align non-branded marketing expenses with the branded sales which were impacted by that spend. This means re-attributing those sales which were closed by branded that really should be credited to non-branded.
Google Analytics provides both custom channel groupings and multi-channel funnels reporting, which can be used to re-attribute the sales that were credited to branded but had previously been impacted by non-branded. However, these methods have some drawbacks as 1) neither functions with the Google Analytics API (in the way this analysis requires at least) and 2) these are fairly manual processes which will need to be repeated regularly. There is actually a much easier method for re-attributing these sales which will also function with the Google Analytics API, and thus, can become a fully automated component of your regular account analysis; the utm_nooverride=1 parameter.
How It Works
In short, the utm_nooverride=1 parameter tells Google Analytics not to overwrite the existing referral data for a visitor. So if a visitor previously came to the website via a non-branded search ad, then returns via a branded search ad, rather than attributing the new session to the branded ad, the utm_nooverride=1 parameter tells Google Analytics to instead attribute this session to the previous one – in this case the non-branded search ad. If this user converts during this session, the conversion will be attributed to the non-branded search ad – the previous source.
There are two steps to implementation. First, append the ‘?utm_nooverride=1’ parameter to the final ad URL on all branded ads. Do not append the tag to all ads – only branded URLs. Your URLs should appear as domain.com/?utm_nooverride=1. Next, enable the manual tagging override setting in Google Analytics (if you are using AdWords auto-tagging).
That’s it. While this change is not retroactive, you should now see the split of branded to non-branded sales begin to change. After a few days, you can use the multi-channel funnels reports to confirm that the change has worked, if you don’t see an immediate noticeable impact.
This change will not only re-attribute branded sales to non-branded search, but will also re-attribute them to email, social, referral, etc. if there is any previous referral data associated with those visitors. If there is significant overlap between branded search and other channels, then you may see the total sales attributed to your SEM channel decline, so be prepared for this change to impact your forecasts.
Also, be sure to format the tag properly. The tag contains an ‘=1’ parameter – this value should always remain as ‘1’. Do not setup your tags as ‘utm_nooverride=2/3/4/etc.’ as the parameter will simply not function.
Why It Is Important
I’ll illustrate how impactful this change can be with a scenario.
Here’s the branded and non-branded search marketing KPIs for Husky’s Furniture Co, a fictional retailer selling furniture through their eCommerce and catalog channels. Due in part to the massive success of their catalog, Husky’s branded search channel is exceptionally large.
Husky’s executive management team has dictated they want to maximize non-branded profits so long as average margin does not drop below 10% – meaning they are looking to pocket at least 10% of sales as profit, and willing to re-invest the remainder, though priority is given to maximizing profit dollars so long as the average margin at which profit is maximized does not drop below 10%.
Above, you can see that on the default Google Analytics attribution model, branded accounts for 72% of revenue (pre-catalog matchbacks) during the past 30 days. Meanwhile, non-branded is where the bulk of the investment in paid search is currently going, yielding $350k in sales on $82.5k in media spend. After COGS/VOH, the retailer walks away with 10.4%, or $36,500 in contribution. With the current non-branded program, there is very little room to generate incremental sales through further investment without falling below the 10% margin threshold.
Now let’s re-attribute those sales which were closed by branded but were previously impacted by non-branded. This is what the ‘utm_nooverride’ parameter will accomplish.
Husky’s branded/non-branded overlap was equivalent to 50% of orders and 50% of revenue. After re-attributing those sales to non-branded, you can see how the picture changes. Non-branded gains $450k in sales, lifting contribution margin to 23.7%. In this example, none of the remaining branded sales were impacted by other web channels, though that may not be the case with your data.
Given leadership’s decision to maximize non-branded profit with a minimum of a 10% margin, this retailer now has significant room to invest further in the channel, growing sales and acquiring more new customers – accelerating the current and future growth of the business.
Hypothetically, Husky’s likely has in the ballpark of $109,500 in contribution which can be re-invested into the channel, given impression share is not maxed out ($82k + $109.5k = $192k total media spend, which at $800k in revenue, yields a 10% margin). If they were to invest those dollars into the non-branded channel, with diminishing returns causing A/S (ad to sales ratio) to jump from 10% to let’s say 24%, they would still drive an incremental $450k in revenue at a margin of 10%. Now the rate at which returns will diminish is unknown at this point, but if we calculate the incremental investment on the figures above, then Husky’s can expect to increase non-branded revenue to $1.25MM, driving an incremental $43,500 in non-branded contribution for the business.
We often find that improving the quality of account measurement allows for the most significant financial improvements for our clients. While the utm_nooverride=1 parameter is not the only option for adjusting your attribution model in this manner, it is currently the only one which works with the Google Analytics API, eliminating the need for ad hoc analysis and allowing you to see this data re-attributed down to the campaign, ad group, keyword, etc. level. If you do intend to implement this, I recommend having a discussion about the implications with your executive team first, as it can potentially change the dynamic of your paid search channel quite radically, rendering current budgets and forecasts obsolete.