Yesterday we took a retrospective look at the ‘Vince’ update, exampling the ‘Poker’ and ‘Life Insurance’ SERPs, and how Google has cleverly managed to identify and apply corrective adjustments to a small number of rankings for big brands.

Today we’re looking at the ‘Holidays’ and ‘Betting’ SERPs and the possible methods behind these adjustments, as well as introducing data from the Stickyeyes data set, enabling us to dig deeper into the back-link profiles of these movers and shakers.

While the previous examples show adjustments to begin on the 26th June, adjustments to the ‘Betting’ SERP began two days later on the 28th June.

Betting SERP - Vince Update

  • is displaced by media goliath sometime in the early hours of Sunday morning.
  • is given a boost to third place, be it fleeting.
  • High street bookmaker displaces both and on Sunday morning 28th June. 
  • On Monday 29th June and are both displaced by
  • also joins the party “above the fold” on Monday 29th June settling in fourth place behind the two major UK high street brands and a global media giant.

Summary of Observations

While the impact occurs slightly later than the “poker” example, we see a similar pattern of displacement, apparently by brand weight, leaving William Hill and Ladbrokes to battle for dominance. Note: by contrast, the “Online Betting” SERP remained unchanged. What’s unusual about this is that the adjustment, which Matt insists is not an update is that the timing is clearly staggered with no apparent consistency. Why would this “change” only affect a relatively small number of queries?

Anyone for a Holiday?

The SEOs engaged with some of these travel brands might be thinking so as they find themselves in a very nice position of the back of Vince.

Holidays SERP Vince Update

  • kicks things off on the 26th, on the same day as two of the previous examples.
  • Later that day, UK travel giant is pushed up into first position.
  • displaces the following day.
  • appears sometime before 3:34 Monday 29th June, displacing, and

While the most significant number of changes occurred between the 26th June and 29th June, we monitored a number of more subtle adjustments which continued to occur over a period of two months. What is clear here is that broadly, known brands to benefited, clearly many deserved to be there, but how was Google able to determine who should rank where?

Technical analysis

Stickyeyes has been recording vast quantities of data across a range of major SERPs, including link structures, on page factors and ranking correlations for all competitors across all the vertical markets which Stickyeyes operates in. Utilising this data and publically available data (see below), Stickyeyes made two key comparisons based on averages, by comparing key factors of sites before and after the adjustment.

  • Analysis 1 – comparison of averages across hundreds of keywords, measuring the key metrics for each top ten, before and after the update.
  • Analysis 2 – comparison of the big winners against the big losers following the adjustment.

Analysis 1: New Top 10s (averages)

  • Alexa Ranking: 53% improvement
  • Domain age: 6% increase
  • Brand Search Volume: 36% increase
  • Google News: 302% increase
  • Total # backlinks: 44% decrease
  • % brand links: 14% increase
  • % targeted anchor text: 3% decrease   

Analysis 2: Big Winners (averages)

  • Alexa Ranking: 79% improvement
  • Domain age: 78% increase
  • Brand Search Volume: 6,574% increase
  • Google News: 282% increase
  • Total # backlinks: 63% increase
  • % brand links: 63% increase
  • % targeted anchor text: 25% decrease

This analysis demonstrates some strong patterns across a number of key areas. However some findings could be symptomatic of something else, rather than actual driving factors behind the statistics provided – Stickyeyes conducted isolated tests in order to gain more accurate findings. These findings raised more questions of course, but some interesting answers.

Q: Is Google factoring in brand search volume?
A: Not in isolation – Instigated by the 6,547% increase in brand search volume in Analysis 2, we tested brand search volume as a ranking factor in isolation. Stickyeyes can confirm that brand search volume is not a factor (at least not in isolation).

Q: Is Google moving away from a link based algorithm?
A: No – We know that Google is considering other factors to determine relevancy including user behaviour, however Google is still fundamentally a link based algorithm. We’ve seen a significant decrease in total back links, which appears to be the result of Google discounting low quality link and rightly penalised the spammers.

Q: Is Google up-weighting non-keyword anchor text (link text)?
A: Not directly – While the evidence implies this, it’s more likely to be symptomatic of the up-weighting of the proximity text in or around the link, another relevancy factor, or the contextual theme of the link. If this assumption were correct, it may explain the 63% brand link increase as well as the 25% reduction in targeted anchor text in our Vince Winners vs. Losers analysis detailed above.

Q: Is this a manual adjustment by Google to favour big brands?
A: Google have always denied manual adjustments
– Taking into consideration the staggered timings, the relatively small number of queries impacted and Matt Cutts cryptic comments, I feel it would be naive not to ask the question. In order to evidence a manual adjustment, would mean catching Google red handed. Perhaps a momentary lack of concentration on the part of a search quality engineer could result in human error to be captured by the SEO equivalent of CCTV.

Food for thought

Remember what Matt said:

I wouldn’t call this an update, but there has been a change… It affects a relatively small number of queries… most people haven’t noticed it.”

Why wouldn’t Matt call this an update? and why would this largely go unnoticed? Why is it staggered? How did Google not only identify these brands, but also position them almost in the precise order of brand equity?

Tomorrow, I’ll present the SEO equivalent to the ‘CCTV’ evidence that may prove this adjustment (at least circumstantially) to be manual. I’ll leave it to you to decide how to interpret the evidence…