The Search Marketer’s Dilemma: Reporting Vs. Optimizing

Cross-channel revenue attribution – arguably the most widely agreed upon concept in the analytics space – is ironically the source of the search marketer’s dilemma.

The dilemma is simple conceptually but digging into it brings up some fundamental flaws in conversion modeling and highlights the underlying catch-22 that marketers face. It all starts with a simple question: Should you optimize a specific marketing channel based off analytics data or proprietary channel data?

To answer the dilemma, let’s set the stage of why the dilemma exists in the first place: it’s impossible to avoid using multiple conversion tracking systems.

Moreover, it’s borderline impossible to have different tracking systems show the same performance metrics for one channel (ex: AdWords and Google Analytics will show different daily conversion counts for the same campaign; or 3rd party bid management systems will show different conversion counts than AdWords for the same campaign). This is because each tracking system collects proprietary conversion data and uses cookies to determine if their ads drive conversions.

One of the core competencies of an analytics system is to resolve multiple cookie issues and distill a complex conversion path into a logical decision of which channel gets credit for a given conversion (attribution modeling; ex: first click, last click, linear, reverse decay, etc). But the flaw lies in the purpose of an analytics system (reporting) versus the purpose of a bid management system (optimization).

Here’s a classic example: Google Analytics utilizes a 180 day cookie window (actions are associated with a given channel up to 180 days following their visit via the marketing channel) assigning conversions to the day the conversion occurred. On the other hand, AdWords utilizes as 30 day cookie window assigning conversions to the day the click occurred.

Now we get to the conundrum: I fundamentally agree with the logic behind both AdWords honoring the day of the click and Google Analytics honoring the day of the conversion. They’re both right – in order to understand the value of a click, you need to link click costs with associated revenue and a true reporting system should display total revenue captured in a given day.

That said, by assigning revenue to the day of the click for optimization purposes makes it impossible to determine how much revenue a specific channel drives any given day.

Solving The Dilemma – The Best Of Both Worlds

Several third party bid management solutions offer the ability to integrate Google Analytics or other 3rd party analytics data into their software – allowing users to manage bids based off revenue captured in analytics. So long as the bid management solution is able to integrate at the click level (via unique IDs per visit), it becomes possible for the system to assign analytics captured revenue back to the click that drove the revenue.

Because AdWords (and every other proprietary marketing tool such as AdCenter, Facebook Ads, etc) uses a proprietary conversion tracking system, each system will take credit for a conversion regardless if the click was the first in a cross-channel funnel, last, or somewhere in the middle.

So if you total up conversion across all proprietary marketing tools, your total will be much higher than the numbers in Analytics. Using an analytics system is the only way to ensure conversions aren’t being double counted and optimization efforts reflect true conversion data/revenue.

If using a bid management tool isn’t in the cards for you or your company, there are ways to take the reporting flaws into account, minimizing the impact on optimizations using proprietary tools.

Start by calculating the average daily delta between your analytics system and the tool. I recommend using several weeks worth of data, ideally a full 30 days:

Calculate the average daily delta by channel. Now do this for several different 30 day periods and average those numbers to derive a normalized multiplier. This is the multiplier to determine how goal CPA or ROAS should be adjusted.

For example, in the diagram above, the normalized delta is 9%. So by adjusting conversion metrics (divide CPA by 0.91 or multiply ROAS by 0.91) you can more accurately optimize based off de-duplicated analytics data.

Concluding Thoughts

In an ideal world, there is a free tool with a universal cookie allowing users to report and optimize on true/de-duplicated conversion data. The reality is that no such free tool exists and third party [not free] tools are imperfect and/or expensive.

No matter how you choose to cope with de-duplication and optimization, it’s a concept which must be addressed in order to provide accurate insight into marketing channel performance.