Is subdomain spam killing local SERPs?

We’re all aware of the *cough* ‘quality’ of some of the local SERPs results we’re served these days. Poor sites, with keyword density spam beyond legibility and crammed footers, full of towns and postcode lists. Frankly, they suck, but more annoyingly they still seem to be working. In the wake of Panda and Penguin, supposedly targeting ‘low quality sites’ and blatant web spam, did local search get left behind?

You don’t have to look hard to find what we’re talking about, a simple {local service:town/city} search should suffice. Aside from actual site quality, sure we have a splattering of exact match domains as you would expect, but one thing that becomes apparent very quickly is the number of subdomains ranking highly for these search terms.

I decided to have a deeper look at the problem and how this dated, spammy technique is alive and well in our local SERPs. I’m not in the business of outing, so to get some sense of the problem I picked a random city and used ‘keyword:city’ for 3 or 4 popular services that a person might need. Lets say you couldn’t get into your house, or had a leaky tap, you get the idea.

Of the 4 services I picked here’s how things panned out.

Excluding the Google+ (places)

12 subdomains
9 subdirectories
16 directories (Yell.com etc)
3 exact match domains

The next step was to take the subdomains and research which of them were indulging in spam practices. I did this by undertaking a [site:] search on the top level domain. The aim being to produce a result something like this.

Forgiving the blurring for a second, examples like this one, which was using town and postcode spam were shockingly commonplace. All the pages themselves were simple lead pages of duplicate content, with the chosen keyword dropped in for apparent relevancy.

After looking into results for both sites using subdomains and subdirectories, over 80% of these first page ranked sites had at least 10 permutations. Some like the example above had hundreds.

This is no empirical study, every town/city and choice of service will be different, but where subdomains and sub directories are being used the correlation between there appearance and usage for spam practices were undeniable.

Whether this is just a short term problem of newly built sites gaining quick wins before they’re removed, its unclear. What is clear is this ageing technique is still out there, still giving results and ruining Google’s ‘apparent’ drive for search quality.

Google Instant and The Polarising of PPC

The introduction of Google Instant last Wednesday has thrown up many questions for PPC Account Managers the world over. Will my CTR plummet? And Do I need to change my whole account structure? Are just two on an ever increasing list. As we get to grips with the impact of Google Instant over the coming months those dedicated ones amongst us will quickly identify, evolve and act to increase on the KPIs we already measure ourselves by. Its in our nature, as it is with all those passionate about PPC to do so. Below are some of my thoughts on what could occur and become best practice as we move forward with Google Instant.

Death of Traditional Broad Match

The release of Modified Broad Match earlier in the year was a great addition to the PPC armoury. Those advocates of traditional broad match teamed with a lengthy, quality researched negative keyword list (which includes myself) still found rich pickings targeting niche markets and the tail. Where (session-based) broad match was applied by Google with little defence or opt out from those broad matchers amongst us, Instant Search could prove more deadly. The ability of instant search to serve new ads every three seconds makes short tail broad match laced with peril. Aside from irrelevant impressions being served along the search in real time, the application of broad match keyword usage could prove just too costly. Both in subsequent irrelevant clicks and the cost in effort involved in optimising an account to defend against an immeasurable related broad match threat.

3 Second Impressions

Google’s ’3 second rule’ on impressions is frankly about as brutal as it could be, given their own admission that the average user takes 9 seconds to enter a search term. I’m no touch typist but certainly no slouch, however the vast majority of people can’t find the correct key to type next in three seconds, let alone know at that stage their search term before they’ve begun. Yes SEO circles may say you’re not charged by impression so why the panic but indirectly you are and the cost will be a high one. Increased irrelevant impressions harm CTR, which lowers quality score which raises CPC. As mentioned by @richardfergie on the recent @kelvinnewman podcast, these problems will affect all PPC advertisers on a level playing field so its a universal handicap. Clients the world over though will want to know why they are paying more for their orders and leads and the explanation, although valid, may not cut it.

Quality Score

The quality score algorithm has always proved a problem for those of us that see the dreaded ‘low search volume’ enter our accounts. Experienced PPC Analysts and Account Managers that target niche markets or the long tail do so for the great ROI and CPA that can be obtained, only to see poor quality scores attributed down to lack of keyword history and historical CTR data. This has been a flaw in the quality score for too long but hopefully instant search may just start to address some of the issues. Instant search already seems to be pushing the long tail, albeit through suggested search. This process of auto fill will help those of us looking for a solution to delve that bit deeper into the web for our answers and for the masses at large, condition their searches to a format that both us and Google would like them to use. Hopefully this change in the methodology of search will start to create richer history’s for niche and long tail keywords to combat current quality score issues and cement the foundations for low volume keywords to start finding favour in search.

Increases In Short Tail

In this current age of impatience people don’t like to type any longer than they have to. All of us are guilty of using Google suggest currently to get to our info that bit quicker, especially auto completing for sites we visit regularly. It’s become an intrinsic part of the search process for us all that we would surely miss it if the functionality was removed. Instead of with Google Suggest where we see the correct keyword and click this has now moved on with Google Instant to there’s what I’m looking for and click. This change will make bidding on the short tail essential especially for brands and those trying to pre-emptively rank for high CPC mid length terms.

Reduction in Mid Length Keyword Searches

Google Instant as discussed above looks to be an attempt to polarise search. For one part driving more short tail to brands and informational searches, the other to more expansive long tail. This newly conditioned way of search should suit both Google in serving more accurate content and and advertisers for getting exposure to niches and the long tail. A browse through suggested terms on Google Instant does seem to back this up. A reduction in the size of the suggest box from a 10 pack down to a new 5 pack ,although necessary so as to not dominate the SERPs results, has led to a point where Google has to cover all the bases in less space. The architecture of Google Instant has dictated this so suggested keywords featured should see an increase in traffic to the detrimental affect of those mid term keywords.

Account Architecture

One final thought on account structure. As discussed above polarising of search should make PPC account structure based on separate short tail and long tail ad groups a must. Many of us already use this technique to marriage with user intent based on an ever refining search cycle, from initial search to final conversion. Google Instant however should make user searches by intent more defined for advertisers as to better identify, analyse and target through ever more appropriate ad text.