If you have had a new Ad Account with Google set up in the last few years, you would have noticed that Google have become very aggressive pushing suggested changes to your ad account. These changes are called “improvements” but we have found often they have a crippling effect on an ad account.
These suggested changes come in two forms:
- Prompts from within the account, sent via email, recommending changes to “improve your optimisation score”. With the click of the button you can implement all suggested changes and see a nice little tick of approval and green optimisation score. You can even check a setting to “automatically apply future suggested changes” effectively putting your account on auto pilot.
- Phone calls from Google representatives, urging you to schedule an appointment and implement their suggested changes to your ad account as instructed over the phone.
What’s the Problem With these Suggestions?
First let me start by saying not all the suggestions are bad. Some make sense and can improve the account. So they are worth considering.
The problem is that the suggestions are auto generated by the Google system and have not taken into account the human intent behind search phrases.
Recently we noticed one of the ad accounts we managed through our PPC Management Service started to decline in clicks, spend and conversions. Initially we reviewed our negative search terms and match types, thinking that perhaps market habits were changing and we needed to keep in step with the market demand.
We quickly noticed however when we dug a little deeper that our client had access to the account too, and had changed the setting which allowed the Google system to auto apply the system generated suggested changes.
One of the changes implemented by the system was conversion based bidding on keywords with a target CPA (Cost per acquisition). The system had looked at the cost of conversions in the account and decided that a $15 CPA was a good number and happily implemented it.
A 52% decline in clicks and a 43% decline in conversions.
Like most ad accounts we had an ad group for branded search terms. When users search for a brand, they are often ready to convert and as a result the cost per conversion for this ad group was relatively low.
Other ad groups had a much higher cost per conversion as they were non branded search terms. Even though they cost more, these non branded conversions were far more valuable as they represented visitors who had searched, discovered the business and converted.
We see these type of conversions as the heart and soul of an ad account and can open a business to growth, bringing in customers from outside their current network and reach.
So in round numbers, about half of the conversions were branded at about $5 per conversion, while half the conversions were discovery based and cost about $35 per conversion. The Google system looked at our average cost per conversion at being about $20 and implemented automatic bidding with the goal of getting $15 conversions on average.
Sounds good right?
The problem is to achieve this the system implemented lower bids for discovery terms and higher bids for branded terms. The system saw conversions coming from discovery terms as expensive and failed to recognise the value they brought to the business.
The result was that we no longer had any discovery traffic coming through the ad account. The brand searches were not that high, so both traffic and conversions dropped.
Fixing the Issue
The first step was turning off auto implementation of ad suggestions.
The second step was adjusting CPA for each ad group. For the branded ad group, we reduced the CPA as we were previously getting leads for $5. We increased the CPA for the other discovery ad groups, based on the market intent in each group and value that represented to the client’s business.
The point of the story is that there is only so much a machine can do. It doesn’t understand human intent. It doesn’t understand the value of some searches to a business vs other searches.
While Google is making a great effort to understand Natural Language Processing and human intent, it’s not quite there yet. So in the meantime Ad Managers like us have a job (phew!). Actually I think we will be ok for a while 😉