Recovering Quality Score in Google Adwords

by Joel on September 18, 2013 · 4 comments

in PPC

Quality Score is a key factor in the success of an Adwords account. It contributes significantly to the profitability of an Adwords account as it is a key determinant of the actual CPC that an advertiser pays.

If you are in need of a refresher on how quality score is calculated and works then you can read our short guide to Adwords quality score.

Why is Quality Score important?

Simply; the amount you pay for an Adwords click depends on your Ad rank.

Your Ad Rank (position) is determined by Maximum CPC Bid X Quality Score. If your ad rank is higher than your competitors then your ad will show above theirs.

This table shows how, for a fixed bid of £1.20, quality score affects your ad rank.

Quality Score ranges from a low score of 1 to a high score of 10.

Max Bid Quality Score Ad Rank (position)
£1.20 1 1.20
£1.20 2 2.40
£1.20 3 3.60
£1.20 4 4.80
£1.20 5 6.00
£1.20 6 7.20
£1.20 7 8.40
£1.20 8 9.60
£1.20 9 10.80
£1.20 10 12.00

As you can see the, having a high quality score means that you can achieve the same ad rank with a lower bid or a higher ad rank with the same bid.

Google do not tell you the ad rank of competitors but what they do tell you is that you only need an ad rank slightly higher than a competitor to rank above them.

A poor Adwords quality score can be poisonous

It is not often that we are stumped by an issue with Adwords or cannot find a solution for it. But we had a client that whatever we did we could not recover the accounts quality score – it just kept edging closer and closer to a fatal ending. On one day in June it actually dipped below 3 on average for the account.

You can see the past 5 months quality score average for the account in this graph:

Recovering Quality Score on Adwords

And to prove that this was not a blip, here is an update on 18th October:

QS Improvement Progress

 

We now have an average (normalised by impressions) account quality score of 5.58. This has meant incredible cost savings for our client who is of course smiling this weekend.

If you want to track quality score like this then read our post on how to track quality score at account, campaign and ad group level.

Did we just sit around and watch it die?

Hell no; It was not that we were not being proactive. We literally tried everything:

  1. Mining search query reports to build negative lists to improve keyword to ad relevance.
  2. Constantly testing and writing new ads that improved the quality score
  3. Working on specific ad groups and campaigns to try and gain momentum in one area that would hopefully breathe life back into the broader account.
  4. Creating new landing pages that drastically improved keyword to landing page relevance.

None of it worked.

It was over a period of around 1 year that it just declined and declined. But then as you can see in the graph – we recovered! Well – at least we are on the road to recovery. 5.58 is still a bit low for an account quality score but with having been sub 4 for over 1 year it was a major breakthrough and continues to accelerate upwards, pulling costs downwards.

How we recovered from a poisonous Adwords quality score

This is a big Adwords account and we needed to understand first exactly where the problem was (what keywords, what campaigns etc) and then to understand why exactly the problem ad groups and campaigns had a low quality score (low CTR, relevancy etc).

We were also aware that even if we did find the problem we could not possibly have worked on the whole account in one go so needed to prioritise fixing the largest problems first.

So first of all we exported all KWs into Excel to see the number of poor, average and good QS keywords (we didn’t have QS script at this time).

Recovering quality score image 1

Please note that these are eligible active keywords only. I wasn’t interested in paused keywords or paused adgroups/campaigns. I was also not interested in low search volume keywords because QS can be high for these keywords and this can screw up the data.

Recovering quality score image 2

If I had included all paused keywords and low search volume keywords then the above data would have looked like the data in the table to the left, which does not look that bad (about 27% of kws with >=7 QS compared to only 0.22% in reality).

So now I knew the real problem was that most of the keywords had a very low QS of 3 & 4 and only a few KWs have >=7 QS. This clearly shows there was something not working at the whole account level because most of the keywords had poor QS. As I didn’t want to work on the whole account I focused on 2 specific campaigns where we spend the most money. These campaigns were the highest traffic campaigns (and also having highest negative impact on QS).

We therefore thought that if we could improve the QS of these 2 campaigns then it would have the greatest overall impact in increasing the overall QS.

We began by looking at several QS indicators such as:

  • Ad copy CTR
  • Relevance between keywords and ad copy
  • Landing page relevance

We realised that in many cases Google was rating all 3 of these important factors as ‘poor’.

One of the main reasons we found for this was that the relevancy between keywords and ad copy was a little weak and this was affecting the click through rate (CTR) too.

As you will gather form the example below our client works in foreign exchange transactions so we have a tight focus around keywords with countries and currency names in them.

Both campaigns were using Exact and Broad (modified) ad groups with only exact keywords in exact ad group and broad match modified keywords in broad ad groups. We found that the selection of broad match modified keywords was not very targeted and it was creating irrelevancy between keywords and ad copy. For example, the +send +money +Germany keyword was using ad copy with the headline such as ‘Send Money to Germany’ but because of the broad match modifier keyword some irrelevant queries would also trigger the ad copy and because of the intent of the search query the existing ad copy was not relevant to the searcher at all and thus affecting CTR.

We concluded that the main problem was the broad match modified keywords were focusing only on 1 kind of search query (in this example send money to germany and not on send money from germany) and it was affecting relevancy, CTR and in turn QS.

This set up was repeated across the whole account. We therefore decided to get rid of broad match modified keywords and to create new Phrase match ad groups to replace them. Phrase match keywords would give us more control than broad match modified keywords.

+send +money +Germany was triggering ad for terms like ‘send money from Germany’, but it was also triggering ads for searches like:

‘send money to kenya from Germany’

and

‘send money to france from Germany’

That was affecting relevancy.

So we went more detailed on the long tail and used to/from country as phrase match keywords so for example “send money to Kenya from Germany” and “send money to France from Germany” and put them into 2 separate ad groups.

We analysed all of the country specific search queries over the previous few months and made a list of about 80 countries and then created 80 x2 ad groups for to/from with Germany/UK/France etc.
These became our ad groups for say ‘send money to France from UK’ and ‘send money to UK from France’ with only highly relevant keywords.
Then we created the same ad groups with transfer money (instead of send), transfer funds to/from country xxx from/to country yyy and so on. This exercise gave much more granularity and relevancy which helped us to increase the click through rate (CTR).

We noticed the first positive differences within a week and then noticed QS was slowly improving.

Recovering quality score image 3 We looked at the same data yesterday and can see a huge improvement in # of keywords with >=7 QS (34%).

Note that the number of eligible keywords are now 69,356 which was 36,391 before. This means we created many more targeted adgroups with lots of new keywords to make sure that adgroups were more tightly themed and that the ad copy included relevant keywords.

See below the comparison before and after we worked on this project:

Recovering quality score image 4

If you are suffering from a poor quality score then please leave any questions you have in the comments section below.

{ 4 comments… read them below or add one }

NCode Technologies September 24, 2013 at 10:20 am

Thanks for sharing such a good information. In my account i am using broad match and modified broad match keywords so now i am understand what match is useful for my campaign.

Reply

Joel September 25, 2013 at 2:21 am

Hi There
Yes; if you are using broad and broad match modified keywords together in the same ad group then you will probably end up quite confused and with a lack of direction.
Just get rid of the broad match.

Reply

NCode Technologies September 25, 2013 at 7:08 am

Hi Joel,

I am using in modified broad match in different campaign. is it ok ?

Reply

Joel September 25, 2013 at 9:58 am

Well I am not sure that I see the purpose of using both broad and broad match modified keywords in the same account.

Normally when we use multiple match types we separate into different ad groups and then use negative keywords to ensure that there is no conflict between them.

i.e. if we have:

[big red t-shirt] in one ad group and then +big +red +t-shirt in another ad group we would add the [big red t-shirt] to the broad ad group so that it does not compete with the exact match ad group for the same search term.

However; when you have two ad groups one with
big red t-shirt (broad) in one and then +big +red +t-shirt in another there is no way for you to stop them competing against each other for the same searches.

I have not yet found a situation where broad match performs better than broad match modified (you cannot run a test at the same time because of the issue stated above).

This post explains what I am suggesting would be a better structure:
https://deepfootprints.co.uk/online-marketing/ppc/adwords-account-structure-simplified/

Reply

Leave a Comment

Previous post:

Next post:

Privacy policy and cookie use