There’s been a number of significant changes to reporting and measurement in paid media over the past few years. From the recent shift in Google Analytics from UA to GA4 to iOS 14 impacting Facebook ads, attribution has become a hot topic for many agencies and in-house marketing teams.
I think sometimes it’s easy to get swept into the minutia when it comes to data, analytics and attribution.
What do we need to know about attribution?
I believe that broadly understanding the user journey, how channels take credit for their touchpoints and how to use business-level performance as a check and balance is enough to grow and scale regardless of these changes…
..after all, they say that ‘perfection is the enemy of good’.
What is attribution?
In the context of digital marketing, what we’re talking about here is a particular channel, campaign, ad, email, etc, taking credit for a sale and the associated revenue.
I’ll give you an example: If you’re running a Google Ads campaign, when a user clicks an ad and comes to the website, a cookie is dropped onto their device.
If that user then makes a purchase, the system is able to report that a sale at x revenue has occurred and attribute that value to a particular campaign, ad, keyword etc.
Where it becomes more complex and interesting is that the user who clicked our Google Ad and was cookied was more than likely also cookied by other channels such as Facebook ads, Klaviyo or may have even first visited the website via organic search and now has come back and converted via paid search.
We end up in a situation where each channel will take full credit for the sale and revenue, over-reporting the actual value by 2-4 times what’s in the bank.
E.g. one customer makes a purchase, the following occurs:
Google Ads: 1 sale reported, £30 revenue
Facebook remarketing: 1 sale reported, £30 revenue
Klaviyo: 1 sale reported, £30 revenue
Total reported marketing sales: 3 sales, £90 revenue
—
Actual website sales: 1 sale, £30 revenue
It’s not something that you can prevent. Even before we had Google Analytics, businesses would have had the same challenges with offline marketing. Companies were still able to make decisions and grow, though, so we’re not in new territory from that perspective.
Being aware of the outliers is very useful as it allows you to ring-fence and protect your attribution givers and to spot the inflated attribution takers!
Most businesses don’t need more data; they need insights.
Like most things, what we’re talking about isn’t a completely linear scenario.
Even with overlap, there is likely added value whereby you will still want all of the activity touching the user journey.
Attribution givers
Let’s look at a typical conversion journey:
There are common sources that drive valuable, qualified initiations of this journey. Non-brand product searches from Google are a great example.
When we start running this activity, when we start getting it right or when there’s a significant breakage, such as a seasonal drop in demand, account performance suffers noticeably—more often than not, the business really feels it too.
Not to the point of perfection, but you can literally see the impact here, positive or negative, on the P&L.
The reason for this is that we’re driving new potential customers, who are actively in-market for what is on offer, to the website—and a reasonable number of them transact as a result.
Of course, it’s not just paid search.
Organic search traffic from product terms, non-remarketing audiences in paid social and relevant affiliate partners can all drive qualified traffic at volume through the start of our conversion journey flow.
For me, this is why getting activity right on this type of audience is essential as a starting point. Doing so is where you can optimise for the highest volume of cost-effective, new customer acquisition.
Other supporting activities can be incredibly useful; however, they’re optimising for efficiencies that our attribution givers are creating throughout the conversion journey.
Attribution takers
At each stage of our conversion journey, we see dropouts for a variety of reasons. These cause inefficiencies in the overall conversion rate for the business.
Some of the tools at our disposal, such as Facebook remarketing, can work to plug some of these gaps.
For example, when someone has looked at a product page and then leaves the site without purchasing, we’re offered the opportunity to serve a reminder, to hopefully drive them back to the site to complete their purchase.
Where this causes an issue from an attribution standpoint is that everyone who’s viewed a product page will then see the remarketing ad immediately.
This means that a sale will be counted by the campaign, and this can be with or without even a click being needed.
You will see an inflated conversion rate, ROAS, and deflated CPA in remarketing campaigns (this applies to any remarketing campaign, not just Facebook).
A certain percentage of users who visited and left would have come back to purchase anyway, so now you have a situation whereby this additional campaign is adding another layer of measurement to the same set of customers rather than initiating new customer acquisition.
1 + 0 = 2
It isn’t an assumption to say that a particular percentage would have come back anyway. All websites have a time lag.
Time lag is the average period of time from the first site visit to when that same user closes into a sale or lead, depending on your business.
For an e-commerce site with a sub £50 AOV, most users would have likely completed within 3-5 days from the initial site visit. A small number would still remain afterwards.
Knowing that many users don’t transact immediately, you can start to look at this objectively.
Yes, this activity will plug a gap and add some value. The gap that’s plugged will likely be marginal.
This is where looking at the business level in the store itself can tell you if there’s been any noticeable uplift to balance against the in-platform reporting.
In reality, from an ads perspective, if the cost is low enough, that’s worth being on to be on the safe side; that’s the decision made—as you know, from a user behaviour perspective, this has ‘a purpose’.
A marginal value on a large enough volume of users is a worthwhile endeavour.
Email pop-ups are another culprit for interjecting through the existing conversion journey (pop-ups that trigger on page load).
We have our customer journey; the user is going through the steps, and a pop-up then appears.
The pop-up provider will send an email from the form submission, so will now count a sale, as will the channel that drove the start of the session beforehand.
If we also have remarketing running, that campaign will do the same
1 + 0 = 3
Sure, in this instance, we’ve captured first-party data, and that has ‘a value’.
Those who completed the conversion journey and became customers would have likely ended up in the email list anyway, though.
Of those left who filled out the form but didn’t purchase, what’s the efficacy of converting those into customers in the future?
Typically, pop-ups also offer a discount. In doing that, you’ve just given away margin by lowering AOV. This affects paid media return-on-ad-spend (directly incurring cost) and general profitability in the business.
In-platform, the email-attributed revenue will look like it’s hockeystick’d. In the sales report in the overall store or in the P&L, this will very likely not be the case.
You can also check this by looking at the overall website conversion rate for all channels. If so many more website visitors were retained, the overall conversion rate would be higher as a result, often it is not.
Bearing in mind all of the cost implications, overlap, what’s the delta?
If there’s a net benefit, then it has a use, of course. It’s likely very far away from what it looks to be in-platform, though.
Let’s look at one more common attribution taker—Brand search.
Someone actively searching for your brand term is more than likely ready to buy. They aren’t just in-market for the product—they’re in-market for the product from you.
Often, this is the second touchpoint from a first non-branded product search or paid/organic social interaction. When you lose the second touchpoint, you don’t close the sale from the first.
Capturing these valuable users is something that would ideally go through your organic listing. Organic isn’t enough, though, as competitors come in (after you’ve done all of the hard work) and bid on your brand term to hijack the sales.
Adding a brand search campaign has ‘a value’. It blocks out competitors and ensures you continue to receive high-quality traffic and sales. You get extremely low-cost clicks in these campaigns, given the ad and landing page relevance to the branded keyword that you’re targeting.
Clearly, though, as these users are already pre-qualified to buy from you, we see inflated performance metrics. Conversion rate and ROAS are inflated and CPA is deflated—much like in our remarketing campaign.
It’s another instance where the cost is relatively low, and although the benefit is marginal, it does add enough value to exist.
We need to interpret reporting differently on this, though, and take what’s attributed with a pinch of salt, given the context.
You also see something similar in brand search with affiliate marketing.
If you onboard established voucher site partners, due to their domain authority, they end up ranking for queries which are your brand term + discount code.
What happens is that space on your brand search results page is lost to a site where not only have you now discounted your AOV, but you also have a commission to pay on top.
Once again, at least you got the sale, and perhaps this is worth doing in your vertical. However, the affiliate platform here will show an inflated return, low CPA, and high conversion rate..
Summary
A lot of these instances are equivalent to someone flyering the queue for the checkout in the supermarket and then taking credit for the sales that occur.
In their own way, all of these activities add ‘a value’ though so they do have a place within a multi-channel strategy. It’s not the activity that needs to be changed, it’s the way that we interpret the results.
I don’t think we need or will ever be able to achieve perfect attribution.
We need ‘enough data’ to be able to make an informed decision.
At a point, data, measurement, and attribution become blockers to taking action and finding out—which is what most businesses need more than anything else.
I say that as someone who’s spent hours per day, for the best part of 10 years, living in Google Analytics.
We can benefit from overarching principles that help us identify the outliers. For me, with attribution, we need to apply Carl Sagan’s:
“Extraordinary claims require extraordinary evidence.”
Know what you’re measuring, where it sits within the conversion journey and how it interacts with the customer. Doing so can only benefit your decision-making.
Interested in finding out more about our services? Get in touch with us below.
We create momentum
Delivered, measured, built upon
