How to Optimise Smart Bidding with Google Ads

smart-bidding-strategies-google

There are two ways of setting your campaign in Google ads – with Smart Bidding or Manual Bidding. Many advertisers still use Manual Bidding or their own scripts and bidding solutions. Google’s smart bidding strategies optimise bids and try to achieve a certain goal. They adjust bids automatically based on the search queries and the user’s browsing history as well as multiple other factors. This is what is making them reliable and in many cases superior compared to manual strategy.

But how to assess performance? A lot of advertisers feel reluctant to give away control on bids as they think they will not get better results. There is  one easy way to assess performance and make an informed decision by looking at the smart bidding report.

Control period

Before moving to smart bidding strategies, make sure you have a long enough control period. This has to be time when no major corrections have been done on the campaign. You must also consider any big seasonal events which might influence performance over the period. A long enough control period is minimum  2-3 weeks of uninterrupted advertising. Consider also the effect of other campaigns you might be running simultaneously online on other channels or offline.

Learning phase

Once you set your smart bidding strategies the campaigns will be in learning period for some time. During the learning phase the algorithm will try different combinations to get the desired result. Smart Bidding strategies are learning from data of trials and errors, as well as previous performance. If the learning phase lasts longer than 2-3 weeks, you might have set too aggressive a goal for the campaign. For example, if your desired cost per acquisition(CPA) is much smaller than what you were getting before, you might consider increasing. The higher the CPA, the higher volume you are going to get in most cases. Same with return on ad spend(ROAS). The lower ROAS, the higher amount of online sales you are going to get.  

The most important thing you should know about the learning phase is that this is not a good representation of the results form your smart bidding. You might see jumps in the cost per conversion, or dips in the volume, which worries a lot of advertisers. The trick is to keep it running and do not make any changes during that time.

Limited by Budget

When a campaign is limited by budget smart bidding does not perform well. In fact, it is likely to perform worse than manual bidding. In order to have a representative sample of smart bidding performance, you must exclude any periods when your campaign has been limited by budget. It is very likely that your campaign starts reaching its daily cap once you switch to smart bidding, so it is important to increase this cap to the recommended level. Smart bidding simply needs a higher ceiling of spending in order to perform. As smart bidding is designed to bring more results, the more you spend, the more you are likely to get out of it.

Change of status

You might see periods when the campaign went through different statuses. This affects the smart bidding performance. Pausing and unpausing sets the campaign in a learning phase, as does any time when you make any changes on the campaign. These periods will be indicated in your smart bidding report and must also be excluded from the assessment of performance.

Conversion lag

Towards the last days of the campaign you will see that there are more conversions to be expected. This period depends a lot on the amount of time it takes from a user clicking on an ad to them completing an action on your website. In some cases this might happen immediately, in others it might take a few days. It is important to exclude the last days of the campaign performance, as during that time frame there might be more conversions to be recorded than what you see.

Once you have excluded all the learning, change of status, conversion lag and budget limitation phases you must have a long enough period in which the campaign functions uninterrupted. This period has to be compared with an equally long control period when the campaign was running with manual bidding. Only then you compare apples to apples and have a clear vision on the performance differences.

Making adjustments

With smart bidding there are few adjustments you can do after the comparison has been done. You might see lower conversion volume at a lower price. You can try increasing the tCPA to see if you can get better volume on a higher cost per acquisition. Also if your cost per acquisition is high but the numbers are good, you can try to lower it gradually to get the optimum balance. Make sure you do not change the CPA with more than 10% and allow the learning period to finish before you make any other changes. Every change you make on the campaign will set it in a learning phase, so be careful.

Attribution

Another factor to consider is your attribution model. Attribution means how you allocate value across the different interactions with your ads. Most advertisers run with Last ClIck attribution, i.e. the click immediately before the conversion gets all the credit. This does not show the full picture, since the customer journey is often complex across multiple touch points and devices.

This is why it is a best practice to use an attribution model like DDA(Data-Driven Attribution) if it is available. (Currently the requirement is 15000 clicks and 600 conversions in the account last 30 days). If not you can try a Linear, Position-based or Time Decay, but in any case stay away from First and Last Click models. Warning: changing your attribution model will set smart bidding strategies in learning phase!

About me: I am an experienced digital marketing professional and a Googler. I have taught over 3000 people in online marketing, sales and PPC. Get my new book. I blog weekly and teach on Udemy with more than 2,700 students enrolled in my online course. Subscribe to my YouTube channel. Follow on Instagram and Twitter @odolena

Leave a Reply

Your email address will not be published. Required fields are marked *