Commentary
Revisiting Return On Rank: The Ultimate SEO KPI?
- by Ryan DeShazer , August 7, 2013
Yet despite the considerable evidence against its use, many SEOs still rely on rank reporting to communicate campaign successes. Unlike other communication mediums, organic search has a certain intangible quality to it. It can seem like witchcraft and voodoo to the lay observer (or lay consumer of services), and so SEOs point to the rank report as the key piece of evidence that their counsel is having a demonstrable impact.
The fact that this is often the normal mode of operation is the result of two truths: 1) SEO is a service that is sold as one that will improve a business’ position (i.e. rank) across the search results pages; 2) a viable alternative has yet to present itself.
What is needed is a replacement measure that would better equip the SEO to understand, and report on, the quantitative impact of their efforts. What is needed is a new KPI. I call it return on rank (ROR).
Similar to KPIs return on investment (ROI) and return on ad spend (ROAS), used to quantify returns from SEM campaigns, ROR would help SEOs quantify the return from their efforts and enable them to prioritize future rounds of optimization. Revenue generated from organic search is tracked just as it would be from other channels; the “cost” of SEO is unique to each company, and involves human capital + technology investments + other miscellaneous costs. The return then becomes a fairly straightforward calculation, with some known caveats:
- The impact of SEO will typically be evident for many months (years even) after the work is performed. In situations where a big SEO project is performed without any subsequent refinements, we can still expect month-over-month gains in ROR (return on rank). This is very different from the world of SEM, where the out-of-pocket investment is needed to generate returns.
- That said, given the real-time nature of today’s Web, we should expect results to wane over time as content becomes stale. Consistently placing time-based or cohort-driven restrictions in an ROR calculation would allow for a more consistent assessment of program impact. For SEOs who are constantly noodling over their sites, or introducing new content, a cohort approach may be best (e.g., group all August efforts as one cohort).
- Accounting for Google’s “not provided” organic keyword referrals for authenticated users. A potential workaround could be pulling user query and average rank details from Webmaster Tools.
- Recognizing that organic search referrals would be present, even in the absence of SEO. A different weighting for brand and non-brand terms would offset this somewhat.
Beyond the core ROR calculation, a powerful predictive model can be introduced that identifies anticipated incremental revenue based on average rank improvements on a per-keyword basis. SEOs would then be able to best prioritize their immediate next step actions. Couple that knowledge with insights from Moz’s Keyword Difficulty tool, and a formidable intelligence mix emerges. SEOs would know with relative certainty which keyword fights were worth picking before expending any effort.
We believe that with enough observations over time, this can become both a legitimate SEO ROI calculation and a smart predictive analytics resource.
This is my first contribution for the esteemed readers of Metric Insider. I’m eager to read your thoughts and field any questions.
Maybe I'm missing the point but as someone who has been overseeing in-house SEO for several years, measuring the return ($$) on organic search has always been the main measure for success. Rankings has always been used as one of many metrics specifically a metric that is more tactical to guide the SEO efforts.
Han - thanks very much for the comment. I'm inviting people to call BS on this one if it won't fly! But to unpack my thinking a bit more: 1) revenue from SEO should be easy enough to identify (especially in ecommerce environments), but cost is a bit trickier compared to paid search or other paid media channels; 2) cost is crucial to calculate, and standardize in some way when assessing SEO's impact relative to multichannel investments; 3) the predictive components to the calculation help prioritize opportunities and next actions, both within SEO as a single channel and as a channel among many. For those of us on the agency side, or overseeing a brand's marketing investment overall (including SEO), this can be powerful. It would put SEO "ROI" on an even playing field to other marketing channels. It would also be helpful to be able to distinguish between SEO wins versus organic search "gimmes" (brand term referrals, perhaps). And I deliberately chose not to broach the subject of multichannel attribution in order to keep it simple, but that would need hefty consideration too.
Hi Ryan,
I like your thought process. As far as the "brand keyword gimmes" go, the same fallacy lies on the paid side. Most tools and companies don't breakdown their reporting between the brand keywords and all other ones. Otherwise, I like your RoR. It would also be cool to give it a global value (e.g. a project that it's worth 75% of each month's revenue.)
Hi Ryan - thanks for outlining a method that's sophisticated and impressive as far as it goes, but it still suffers (as do many other models) from "last click" attribution errors. This method wouldn't pick up, for example, a visitor who first "discovered" a brand through organic search, then later returned via a different path (such as a branded search or bookmarked link) and converted into a lead or sale. It also wouldn't adequately account for a scenario where most conversions are generated through an email newsletter -- but many of those subscribers first discovered the site through organic search. It's a tough nut to crack.
Tom - thanks for the thoughts. Read my response to Han above; I mention that multichannel attribution was kept out of this discussion to keep things as focused and clean as possible. But you do raise a good point. Two solutions emerge: 1) look only at the organic search channel (through an independent application) that would eliminate other channels from the conversion noise; 2) introduce this methodology into an organization's existing attribution technology (like a ClearSaleing). I'm happy to show you what we're building, btw. Multichannel attribution is most definitely a key consideration.