Author:

There's a difference between being similar and being relevant. Image sphericalnotions.com.

 

 how does broad match actually work?

When analysing the search queries which have been activating your ads, it’s easy to become confused by the depth of strange search queries that Google shows your ads to. Why is the search query ‘football fixtures’ ever considered relevant to the keyword ‘rugby socks’? The truth is Google doesn’t directly measure relevancy in the semantic way humans do – it’d just be too complex – how would they know that ‘Usain Bolt’ isn’t related to ‘nuts and bolts’, and that ‘Cheryl Cole’ once was, but is no longer, relevant to ‘Simon Cowell’? Google puts it to the user to decide, and uses relative click through rates to measure how relevant a term is. After a few tests, Google can confidently show an ad with the keyword ‘diy tools’ to the search query ‘red paint’, but won’t bother when someone searches ‘paint the town red’.

 

why broad is bad (kinda)

So this leaves us in a sometimes difficult situation because we’re susceptible to users who are just coincidently browsing. A user searching for ‘irish food recipes’ may still click an ad for ‘buy scottish food’ and so Google would still show an ad, even though they’ve no intention of making a purchase. In fact, if enough users were to click, your broad match keyword could show an ad for absolutely any search query possible!

 

using modified broad and negatives

So with normal broad match you actually have ZERO control over where your ad appears. There’s no semantic relevance and your ads will be eligible to appear anywhere there’s a chance they’ll get clicked. But using a mixture of modified broad and other match types will easily filter this out whilst keeping the keywords as flexible as possible. In our example above, lets say that we’re absolutely sure that any search query mentioning recipes will be irrelevant to us, so adding ‘recipes’ as a negative will block out any further traffic like this. However we wouldn’t add ‘irish’ as a negative incase a user searched for ‘scottish and irish food’ – so instead would tweak the original keyword to be modified broad on ‘scottish’, ensuring that all search queries targeted would at least have to include that term.

 

should I add search queries?

This brings us to an interesting point – If there’s no semantic measurement of the relevancy of a keyword to a search query, and it’s (generally) all done by measuring click through rates, then why bother adding new keywords from the search query report into the same ad groups that activated them in the first place? Example: if my keyword ‘rugby socks’ picks up the search query ‘buy rugby socks uk’, then is there any point in adding it as a new keyword? Regardless of which keyword picks up the search query, it’s going to receive the same click through rate either way and therefore both be deemed equally relevant for the search query ‘buy rugby socks uk’.

 

no, don’t add search queries

You could argue that adding search queries into your ad groups as a keyword is a waste of time, as your current keywords are already picking them up and they’ll receive the same click through rate from your ads either way. In most cases however, this highlights the importance of segmenting your ad groups as it gives you the opportunity to tailor the ads further and therefore improve your click through rate on these search queries. If you can create an ad that better suits the search query ‘buy rugby socks uk’ than the current ‘rugby socks’ ads, then the search query needs to be added into a new adgroup with tailored ads.

 

yes, do add search queries

In reality though, as you may have guessed, there is still a point to adding search queries to your current ad groups as new keywords are one of the easiest forms of keyword research. Each search query will have a circle of search queries which are relevant to it, with each one of those search queries having another circle of slightly different search queries also relevant to it. By regularly adding search queries as new keywords, which are very similar to your current keywords, you can slowly build out the coverage. You don’t add the search query into your ad group because you want to target that keyword theme ‘better’, you add it because you want to use it to target the other search queries which are ‘relevant’ to it.

 

so wait, what should I do?!

Firstly, remember the sobering fact that the search query report shows only a fraction (often below 10%) of the actual search queries that are activating your ads. Do not rely on it as a way of auditing your traffic – use a combination of match types, negative keywords and a lot of common sense to filter out unwanted traffic in advance and don’t wait until you’ve already paid for it before deciding to block it. The Google Keyword Research Tool is a great resource for finding negatives in advance, and with a few tweaks of it’s settings can be really powerful in doing this. If you do find search queries from blatantly unwanted traffic coming through into your account, chalk it up to experience and get back to doing some negative keyword research pronto.

Secondly, when you uncover new search queries triggering a particular ad, ask yourself if you’re really doing that user justice by serving them that ad. If you have any inkling at all that you may not be, then get the search query added as a keyword in it’s own ad group and give it the ad it deserves!

Finally, if the search query really isn’t doing any harm receiving the current ad, then ask yourself if your ad group would benefit from having it added as a keyword? Could there be even the slimmest bit of chance that there are search queries relevant to this new keyword that your current ones may not pick up so easily? Remember that it’s measured by user’s clicking, not semantically how the human brain works, and therefore trying to accurately predict what AdWords will throw up as being relevant is not (technically) even humanly possible.


Tags: , , , , ,