Google is ‘No Longer a Black Box’ at BrightEdge #Share12

BrightEdge Share 12One of the themes of the industry panel I moderated at BrightEdge Share 12 was whether Google is still “a Black Box”… and whether marketers needed their own Black Box to deal with Google’s Black Box. Have marketers entered a new era of automation where, in a Spy Vs Spy scenario, big data appliances are able to measure, analyze and respond to each other? I caught up with Leo Haryono, Director of SEO at Macy’s says and Lemuel Park, CTO of BrightEdge to get their view on this question.

Paid search marketers have enjoyed this situation for years with automated bidding platforms managing their cost-per-click in Google’s own auction and in particular against competitors. Pay-per-click (PPC) marketers have enjoyed the benefits of big data appliances for longer than other search marketers with features such as rules based bidding, bid to position and return-on-investment (ROI) based bidding. Display marketers too have been able to manage and predict demand for even longer and the smart money has been on combining all these black boxes together to make managing PPC and Display as interchangeable skills.

SEOs, however, have not had the luxury of being able to predict or respond to deliver SEO strategies at scale. It has always been the slow burn strategy, which now social media marketers now also must face. In many ways, search and social both occupy the earned media side of the, much discussed, paid to earned media ratio that CMO’s covet. Despite two millennia of modern civilization (relatively speaking), we still don’t know what makes humans tick… but perhaps more frustrating is the fact that after more than 13 years of Google, SEOs still don’t really know what makes the super computer – the tin man of our times, our robot overlord – tick.

Or do we? Leo Haryono, Director of SEO at Macy’s, would argue that the times, they’re a-changing.

Peeling the Onion

“Google is becoming less of a Black Box,” Haryono told SEW before he took to the stage for the #Share12 industry panel. “SEO is becoming more data centric. Whilst we still can’t get a perfect analysis like paid channels, we have more visibility into all the metrics now. This is partly due to tools that Google has put out themselves and also due to the rise of enterprise SEO software platforms, like BrightEdge.”

“It is more like a science now – SEOs have tools and methodologies to measure any initiative. You can create a hypothesis about ranking and the tools make you more confident to measure and make changes. Without the data you are stuck making ‘statutory changes’ (i.e. according to the rules) but with data and segmentation you can dissect the impact of the changes and prioritize each initiative based on the ROI.”

“I call this process of SEO testing, ‘peeling the onion’. As the unknowns become knowns, the onion gets smaller and smaller – the Black Box gets smaller.”

Quick & Dirty Vs Slow & Steady

In light of the discussion, I ventured some questions to Haryono, “If Google’s algorithm is becoming easier to understand, does that not pave the way for Black Hat techniques? Surely the more SEOs know about how Google works, the greater the temptation to game the system.”

“Black Hat Techniques are not ROI positive,” Haryono responded categorically, “these short term tactics tend to have a negative ROI. Google penalties are a decent deterrent. SEO is not only about rankings, it is about free traffic. For me SEO is about building a strong foundation upon which to grow as a business. Any company wanting long term success has to focus on ROI and make a 6 to 12 month bet on profitability.”

macys-logo“Managing SEO for strong brands like Macy’s naturally have an advantage in certain areas. As we continue to grow our brand in fashion, we naturally accrue organic links to our site. Our approach needs to be relevant for users and as SEOs we must do our best to accelerate the brand footprint. This is where social media has a role to play.”

Macy’s focus on user experience. “We aim to be deliver a more appealing shopping experience, for example, using bigger images and site navigation. Whilst an enhanced user experience may not be optimal for SEO, we continue to build for users and optimize for the robot (Google’s crawler) by working with cross functional teams” At Macy’s the creative team manage copy, the merchandising team get the product info and the product management team who build the features of the site. The SEO department drives awareness and evangelizes SEO concepts internally with specialized training courses for each of the different teams.

Combining different data sources across the business with a “cross functional approach” enables Macy’s to mesh insight from paid search and social media to test different how visitors from different channels behave. Haryono continued, “For example, PPC shortens our planning process about what markets to target with SEO. Depending on objectives we can dial up or down returns from specific categories. We have done some limited tests on social’s impact on search, and whilst we are definitely seeing interesting fluctuations, we don’t have any conclusive evidence on what is the right mix to get the best impact.”

Building a Black Box to Watch the Black Box

BrightEdge logoWhen Lemuel Park, CTO of BrightEdge presented some new S3 platform feature updates at #Share12 a particular thing he said stuck out to me, “as mobile search share continues to grow, we wanted to solve the problem of mobile rankings and see how they differ across cellphones, desktops and tablets, but in order to solve mobile rankings, we had to solve the problem of local rankings first.”

It struck me as interesting as he hinted at a layer of intelligence required to develop the next layer of intelligence, which would have substantially slowed the development time for the feature request. What i wanted to know was how did he know that he could risk keeping customers to develop a feature set that may be short of what they desperately needed?

Catching up after the conference, Park told SEW, “the most important thing is to index all the different data sources in the right way. “We use a lot of relational database and Hadoop concepts,” he said, “start-ups are lucky because they can stop on a dime and change directions, which is important when you are dealing with something as changeable as Google.”

So actually the risk was not as great as I had imagined. He said, “we run focus groups with key customers to test certain ideas and assess future feature requests. What often happens is that common goals emerge which sets a kind of roadmap.”

Of the features announced all of them seemed to solve a typical Black Box problem that many of us have been asking ourselves:

  • How Volatile are Search Rankings? BrightEdge’s Daily Rankings update aims to measure the volatility of search rankings and gauge how much they shift over time to deliver a more accurate average position.
  • How Do My Rankings Vary by Location? BrightEdge have expanded their Local Rankings offering to support an additional 81 cities around the globe, to provide city by city comparative analysis with dashboard and management functions so that companies “can fill gaps in coverage” (a concept explained by Raj Rao from 3M).
  • How Do My Rankings Vary Across Devices? BrightEdge’s Mobile Vs Tablet rankings feature combines data from both daily rankings and local rankings and then pivots it by device type, whether that be via mobile or tablet.
  • How Often are Relevant Discussions are Taking Place on Twitter? BrightEdge’s new Twitter Trending feature “listens” to the Twitter “firehose” of data and matches the keywords you are optimizing for – so that movements in rankings can be measured against social signals, or alternatively, unexpected spikes in discussion around your products on Twitter can be capitalized on to rapidly retarget your Twitter ad campaign.
  • How Do I Deal with Keywords (Not Provided)? BrightEdge’s new Page Manager feature pivots your SEO and ROI data from keywords to the landing page to create a content centric analysis. In particular, this feature could help mitigate reporting limitations created by Google’s Encrypted Search Referrers and also the give a view on the impact of Google Search Plus Your World personalization.

Park continued, “there was a certain serendipity to how the product roadmap emerged.” When BrightEdge started building daily rankings Park’s development team decided to go for a page by page analysis (i.e. show paginated results as Google does) rather than delivering results for an automated query that requests 100 results per page. “We wanted to report search rankings in exactly the same way that users would see the data. We found that when we added a ‘num=100’ attribute to the Google query string via the API, Google actually organized the universal results differently (to how they would show on the typical paginated display a user would see) – Universal Search Results would be ‘clumped together’.”

Park also highlighted to me that Google Korea has a completely different way of organizing Universal Search Results compared to the rest of the world and that in their own tests, BrightEdge found that Mobile Rankings differ depending on whether the device is connected by data or by Wi-Fi. He explained, “We had to do mobile on a page by page basis as it seems that a Mobile search on data connected device is less stable than on Wi-Fi connection – it always seems to be triangulating location.”

For Park, the central tenet of creating BrightEdge’s counter-Black Box is accuracy, “with encrypted Search Referrers, traditional analytics start reporting less profit from successful keywords,” because the data is not provided. With Page Manager BrightEdge is reporting on the ROI generated from the page itself, “which is a new perspective to overcome the data loss from encrypted Search Referrers.”

Learn More at BrightEdge #Share12

This is the last in the SEW series covering BrightEdge Share 12. To catchup with our recent coverage, check out:

BrightEdge today also announced that Adobe has selected BrightEdge S3 to manage The BrightEdge platform integrates with the Adobe Digital Marketing Suite allowing the Adobe SEO team to forecast, manage, and report on large scale global SEO campaigns.

Related reading

Exclusive interview with Craig Campbell Golden nuggets every SEO needs to know
How to understand searcher intent to boost SEO rankings
How to master technical SEO Six areas to attack now