Building ROI is the main goal when someone starts his business or spends money. Here we will focus more on Building ROI with SEO strategies- when discussing pain points with prospective clients google search volume api – see a remarkable number of SEO strategies that are, at best, decoupled from business performance, and, at worst, totally unengaged with wider business performance.
Failing to employ a framework “See Think Do Care” (STDC) framework – or indeed, Google’s Micro-moment mobile variation, which explicitly uses the STDC framework for KPI measurement – is surely not planning for success in 2017.
So how can we, as marketers, do better for our clients? I have my own framework for mapping SEO strategy down to hard £ (or $) impact.
Start With The Growth: Kill brand!
Before thinking about attribution or STDC, let’s consider SEO fundamentals. If we include brand terms in our assessment of performance, we are looping in all sorts of other marketing activities which we should not be taking credit for. So: filter out brand.
How do we achieve that in practical terms? Google Search Console allows you to quickly and simply filter based on terms or partial terms, like a simple REGEX function. So, great! We can simply run exclusions based on the simplest version of our brand term, and we are done.
We only have 90 days data from Google right now. So, either we use a handy TARDIS to go back in time and collect the last year’s worth of data to do a year-on-year comparison to smooth out growth variation due to market seasonality… or, we accept we can’t run year-on-year with brand filtering until this time next year (or more precisely, in 360-90=270 days time). Of course, this is assuming we consistently export and stash Search Console data for our own analysis.
By the way, if you need a hand with that part, I’ve written about exporting Search Console data before.
Even with no TARDIS-delivered historic data, we can at least take the last 90 days percentage brand rate and remove an equivalent percentage from the organic traffic coming into our website as a starting point for the current generic performance delivery.
From there, we can measure the improvement in total click volume from generics over time and apportion that across ongoing organic traffic performance. This will essentially measure incremental improvement from our start date that is delivered by generic term performance improvement.
Step one, complete. We have objective measurement of overall improvement in organic performance excluding brand. That means any remaining performance improvement comes from activity you would expect to find in a solid SEO strategy: content adjustment, ranking improvement, SERP click-through rate improvement, etc.
Take it to the Awesome Revenue Part
Now that we have a moving objective view on organic performance, we can map back to revenue in our analytics package of choice. I’d like to assume all sites have well set up, fully tagged, e-commerce enabled (or revenue-attributed) event tracking in place. If that’s not the case, take an action and get on it!
Mapping to revenue should then be easy. Next, we should take into account any costs you are incurring with your agency. A simple ROAS – (Revenue – Cost) / Cost) – approach is to take our generic-only tracked uplift and compare it back to your agency cost at whatever rate you wish. Typically, it’s worth assessing this weekly or monthly. Higher frequencies are useful for higher spends.
More useful, however, is to move beyond ROAS to true ROI. To achieve that, you need to know your average profit margin for revenue sold. Doing this to the category level is usually enough to start focusing your keyphrase strategy in the right areas – where ROAS looks good, but low margins means ROI is negative.
So in simple terms, think of your ROI calculation as (Profit – Cost) / Cost, a subtle but important difference required to ensure you are driving profitable growth.
And Finally: Build in Attribution
At this stage, you are exporting web analytics data that is showing incremental organic improvement mapped to generic growth, and you are looking at reporting that maps from keyphrases to their revenue delivered with an ROI calculation on top to show if you are picking the right battles.
Now it’s time to improve our accuracy by removing our SEO silo and allowing ourselves to fit into the full, multi-channel, customer conversion path.
In simple terms, the aim now is to pick an attribution model that most closely matches your typical customer conversion path, and apply it. You can get a sense for that by looking at the MCF report in Google Analytics (or using API export and analysis in Adobe). You can tweak the session analysis period to reflect if your product typically enjoys a fast STDC process, or a multi-month STDC path.
A typical model that works well is a blend of time decay and position-based, which you can mock up as a custom model.
And of course, if you are on the Google stack, you can also take advantage of their newly announced free AI attribution analysis tool.
If you are serious about attribution accuracy, there are limitations to Google’s approach: It doesn’t easily ingest non-digital marketing data silos to reflect the true attribution picture, and of course Google Analytics (not to mention Omniture) are built on sessions rather than customers. So there is an approximation there.
If you want to get serious about AI analysis that is truly granular to individual customers, then you have just shown interest in my favorite topic, and I’d be glad to chat about the latest tech in that area – and indeed, that seems like a good subject for future columns!