If you’re not up to speed with the Google Previous Query technique, it’s quite simple – and the name says it all really. Put simply, Google is starting to display their advertising placements based on the key phrase you’ve searched for and the phrase(s) previous to that.
As an example, a customer will search for [travel] and then [insurance]. The second search will produce advertising results for insurance but with a slant or higher emphasis on those bidding on travel insurance terms. It’s fairly easy to see why Google are approaching it this way.
Firstly, they know that a large percentage of searches are tiered – becoming more specific as more words are added to the phrase e.g. 1st search = [insurance], 2nd search = [travel insurance], 3rd search = [travel insurance quotes]. Usability studies also show a small but not insignificant percentage of users breaking (for example) [insurance] [travel] [quotes] into three separate searches. You can see whyGoogle are looking at the previous query technique as a way to add some contextual relevance to the customer’s session/search sitting/previous query.
Secondly, it will undoubtedly add another layer of mystery to the results pages. Each of the paid search advertising systems has moved away from the pure bid ranking model. Quality score (page content relevance, CTR history and the like) has added an element of sophistication to paid search optimisation – all designed for the benefit of the customer of course!! A search engine sceptic could be forgiven for thinking that the previous query technique doesn’t really benefit the customer and simply adds another layer of confusion for them and the advertiser to think about.
Although it is only being trialled now (by any users who accept cookies – not just those searching under a Google sign-in), the previous query technique is also rumoured to also be aimed at natural search results. In theory, this should help to reduce the dominance of highly competitive one or two phrase search queries – but that is arguably less likely in the personal finance market. One thing is for sure, customers are likely to be confused by the results returned as they change their search intention during a session e.g. 1st search = [loans], 2nd search = [car], 3rd search = [hire] …is the customer looking for the loan or hire of a car or did they start searching for loans and then switch to an unrelated search for car hire?
Capital One
Credit Made Clearer
lovemoney
Relationship with Your Money
CashQuestions
Personal Finance Problem Solver
Life Cover
Insurance Made Easy
{ 10 comments… read them below or add one }
lol! I noticed this a week or two ago when I did one search, and then did another search for a completely unrelated topic and was still seeing ads for the previous search term.
It seems to me like Google are just finding excuses to give more impressions to ads, and therefore more income from AdWords. Which then makes me think “Are they doing it for customer usability, or corporate greed”?
What a shambles, 1st search, car insurance and the usual suspects appear as you would expect, 2nd search spanish holidays and hey presto nothing in sponsored search that has anything to do with holidays in spain, it’s all about car and travel insurance in spain, well done G that is really relevant for the customer!
Sorry for being a bit dull but I just don’t get it!
“Sorry for being a bit dull but I just don’t get it!”
I don’t think customer will either Dave!!
I suspect they’re trialling it more and more to perfect the revenue – I mean usability – aspect of it, but some of the results are quite a way off the mark and I would’ve thought that the CTR would fall on the second search.
Mind you, do you think there’s an added weight for the more expensive terms? For example, searching for [car insurance] and then [spanish holidays] brings a strong element of car insurance into the spanish search. Almost completely dominating.
On the other hand, if (starting with a new session, browser and clearing cookie) you search for [spanish holidays] first and then search for [car insurance], the insurance searches still dominate their own query and there’s extremely little impact from the spanish context.
Revenue. Revenue. Revenue.
‘Mind you, do you think there’s an added weight for the more expensive terms?’
Oh yes, I think the whole thing is geared towards the Finance/Insurance sector, it’s where the big money is after all.
I think this is aimed a revenue, but I guess it will affect the longer tail, off beat phrases more, whereby google will bring in the higher paying advertisers.
Let’s put this in context though, how many people will search for car insurance and then holidays staright after? Surely there is a time cut off on this anyway, it’s not going to base ads on what you searched for last week.
I think they are trying to maximise revenues from same industry searches, I can’t see many normal users flitting from one industry to another in one session.
Also, this is “previous query”, so the user mustn’t have found what they wanted on their first search, do you think this will affect the likes of “car insurance” – how many people re-search for a different phrase – not many I bet. So in terms of organic seo, again I think it will affect the off-beat phrases where people haven’t founf what they were looking for 1st time round.
Interesting point Clive. I guess stats on search intentions per session would be useful to qualify that. I can see your point but ‘in theory’ Google ’should’ be serving relevant adverts to any one particular query.
What may happen however – and this is something we all know people do – is that the longer and less competitive key phrases may be invaded by the top key phrases.
OK, I’ll explain that a bit better. Searching for [car insurance] as the lead phrase – it’s common for the customer to then follow up with a more refined intention – such as [female car insurance]. They could refine another lever to [young female car insurance].
Do these less competitive terms where the relevance to the advertiser is currently higher get washed away because Google is bringing in more adverts from the previous, higher-paying, generic [car insurance] term.
It doesn’t create a completely off topic list of adverts but it certainly makes it harder for the small, niche advertisers who thrive/survive on the specific bits of business they’re set up for.
Also, what about the thought that they’re looking to bring this into the natural search listings?
@Ernest
Agree with you on those points, could be a bit of a rich get richer scenario.
But, it would be interesting to know how many people “tweak” their searching as in the example you gave, I don’t suppose there is any way to know for sure, but judging by impression/click projections that the likes of adwords give, the overall volume sticks with the main terms….
There has been talk about this sort of adserving being introduced for sometime. I am don’t think it is anything terribly innovative though. As long as there has been marketing, advertisers have been trying to second guess consumer requirements and needs.
I will often promote life insurance against mortgage searches. Does not get a lot of clicks, but when that same user does search for life insurance they may well remember/recognise the brand meaning you do not have to bid so aggressively.
What I don’t understand at this point is how it works with the relevancy algorithm, which is supposedly part and parcel of the ranking model. Secondly, an advertiser will buy against a key word, the price of which can vary wildly dependent on vertical. When your secured loan creative gets served against a credit card search, how and where are you paying? Is there a new bidding system which I am missing?
No new bidding system that anyone has commented on – just another level of mystery!!
If your interested there are a few patents on this which were covered last year on this topic. I listed some links to further coverage last week here; http://www.huomah.com/Search-Engines/Algorithm-Matters/Query-analysis-confirmed-as-Google-ranking-signal.html
What is interesting is considering what other related signals they may be tracking beyond query analysis. There are more than a few other user performance metrics that are outlines as well in those patents. Interesting stuff ….