Google Ads – Can AI optimize your ads?

Google Ads is a very powerful marketing platform. Whether you work for a corporation or own a small business, if you want to get in front of people searching for your product or service, Google Ads can provide almost immediate results.

Unlike social media platforms, when people use Google and search for specific, relevant phrases or “keywords,” you can usually understand the searcher’s intent and place ads in front of your audience when they are ready to buy or sign up.

This level of targeting is a lot more difficult on social media, as users are typically scrolling through a newsfeed rather than actively searching for something of relevance.

Google and AI
Google has been at the cutting edge of artificial intelligence for a number of years. In 2011 the company started the Google Brain Project, and in 2012, they built a “neural network” of 16,000 computers, which is trained to identify animals in photographs and videos.

With the acquisition of the London-based DeepMind organization in 2014 for approximately $500 million, Google has been using AI for everything from beating world champion humans at the ancient game of Go, to updating the relevancy of search engine results pages and optimizing Google Ads bidding.

Regarding Google’s algorithm updates, the search engine uses AI to help understand search intent and words in the context of the surrounding words. Google’s BERT (Bidirectional Encoder Representations from Transformers) update uses natural language processing to train its own questions and answering system.

Google Ads smart bidding
If you have ever experienced the joy of managing a large Google Ads account, then you will have no doubt consulted with a Google employee about switching over to some form of automated or smart bidding.

While this can often be of great benefit to the individual or organization running the account, many a cynical Reddit user has also pointed out that Google wants to push more users towards AI and smart bidding – because the more data they get, the better smart bidding will work.

Smart bidding can save you lots of time and effort. It can automatically adjust bids based on user behavior and demographics, bidding more for users who are most likely to convert into a lead or customer.

What are the specific smart bidding strategies?
If you head over to Google Ads, and click into a campaign and then “Settings,” – there should be a section with the title “Bidding.” There should be a drop-down menu within the section, which contains the names of all the different bidding strategies.

Target CPA – “sets bids to get the most conversions possible while reaching your average cost per action (CPA) goal.”

Let’s say you run sell answering service software for small businesses, and you are running Google Ads to drive form completions and phone calls from your website. If you choose Target CPA, you set a bid amount, for example, $45 – you are telling Google Ads to get as many leads as possible for approximately or below $45.

If you set your Target CPA bid too low, you will stop the campaign from working. For example, if you enter $5 as your Target CPA, that is unrealistic, and Google won’t drive any conversions for that amount, so you will probably get no clicks or conversions.

If you are not sure how much to bid, use Google’s Keyword Planner tool to get an idea of the bid amounts.

Target ROAS – “sets bids to help you get the most conversion value while maintaining your target return on ads spend (ROAS).”

For Target ROAS to work effectively, you will need your conversion tracking set up so that each conversion is assigned a value. For example, with the answering service software example, you would want to link up your conversion tracking to a CRM to estimate the annual or lifetime value of each conversion/customer.

Using this data, Google can then adjust bids to target users who are most likely to be the most profitable. To use a simplistic example, if all of your most profitable customers are lawyers from South Florida, then Google’s Target ROAS bidding strategy will probably bid a relatively high amount to target lawyers from South Florida.

Target ROAS can work well with Google text ads but is generally more suited to eCommerce sites and Google Shopping Ads.

Maximize Clicks – “set bids to get the most clicks within your budget.”

This bidding strategy will aim to get the most visitors to your website via a click of Google Ads. Be sure to set a maximum CPC bid. Otherwise, Google will sometimes bid extortionate amounts for clicks on your ads. If you set the maximum CPC as $40, Google shouldn’t bid any more than $40 for a click. However, you should still monitor the bid amounts.

Maximize Conversions – “sets bids to help you get the most conversions within your budget.”

Before you use maximize conversions, ensure you have lots of conversion data within your account. Also, be sure that the conversions you are tracking are valuable to the business. For example, a form completion or phone call will generally result in more valuable leads than a live chat conversation or an innocuous conversion such as a click on a footer item.

Target Impression Share – “sets bids with the goal of showing your ad on the absolute top of the page, on the top of the page or anywhere on the page of Google search results.”

You can set bids according to where you wish to show up on the Google results pages. For example, “absolute top of results page” will show your ad in the top position, “Top of results page” will display your ad within the top 4 results.

Why AI bidding doesn’t always work

When you switch over to automated bidding, it is unlikely that you will see an immediate improvement in performance. Smart Bidding needs time to learn, and as a result, for the first few weeks to a month, it could result in a drop in performance.

An overall lack of conversion data is another common reason smart bidding strategies might not improve your account’s performance. Smart bidding uses conversion data to make decisions about how much to bid. Smart bidding needs to understand what variables are driving conversions too.

If you only have 1 or 2 conversions per day in each campaign, it will take a long while to build up a significant amount of data to use in terms of bids and decision making.

If, however, you run an eCommerce store that sells low-value items and you have hundreds of conversions per day, smart bidding will have lots of data to work with and should start to produce a more positive return on ad spend within a few days.

The more conversion data you have, the more beneficial smart bidding should be.

As outlined above, Google and smart bidding like lots of data. The number of keywords can also limit conversion data in each of your campaigns. If you run very small, niche campaigns with only 1 or 2 exact match keywords, again, it could be a bottleneck in terms of the amount of data smart bidding’s AI has to work with.

Frequent changes can also impact smart bidding’s performance. If you add a keyword and remove another one, the machine learning will need to start, and you will effectively lose the data associated with removed keywords.

Drafts & experiments

A great way to test automated bidding strategies is to create a draft of an existing campaign. A draft is effectively an exact copy – but you can then go in and change just the bidding strategy to an automated one.

Side-by-side, you can then see which bidding strategy is the most effective. Just remember to give the automated bidding strategy time to learn and adjust.