Looking at PPC in 2018: Predictions for the New Year

1. More Audience Targeting Features

The most significant changes in PPC over the past five years have been dominated by audience targeting features, and 2018 will not be an exception. We predict audience targeting will advance in several different ways.

Expansion of Customer Match Targeting

Customer match is audience targeting based on your CRM(Customer Relationship Management) data. It allows you to show ads based on data about current customers that you choose to share with Google. At the moment, you can use this targeting across Google’s properties, such as Search and Gmail ads. However, it is currently unavailable for third-party general display advertising.

Throughout 2018, we should see an expansion in how customer match targeting can be used. Since this targeting is based on CRM data, we might see some restrictions in the EU and other areas. These areas have a lot of privacy laws that restrict how customer data can be used in advertising. However, we can still expect to see increased customer match targeting options everywhere.

Expansion of Similar Lists

Similar lists are lookalike lists to an audience list you create in AdWords. Presently, if you don’t meet some minimum data size, you rarely see these lists as being available. As machine learning gets better at patterning out user behaviors and matching them to other people, expect to see more similar lists available within your account.

Easier Third-Party List Management

One of the least used features in AdWords is the ability to use non-Google third-party lists in your PPC account. This feature has more than 92,000 audience lists available to it. These lists come from third-party data providers, such as BlueKai.

There are two reasons these lists are not visible on most accounts:

• Most people don’t rent audience lists
• They pay on a CPM (cost per impression) basis

While most smaller accounts will never need to layer third-party data with Google’s targeting, sometimes it is useful, especially when you are trying to reach a very narrow market segment. Unfortunately, when you pay on a CPM basis within a CPC product, the combination of payment methods can get complicated.Suddenly, an impression is no longer free when using this method. With AdWords, you pay for the click; with CPM, you pay for the impression. Mixing these two opposing payment systems can be tricky, so you have to be very careful when using third-party lists.

Expect integration of third-party list data to be easier to manage some time in 2018.

The Dreaded Unknown Keeps Declining

The last few years have seen several new demographic targeting features. However, for many advertisers, these features have had little meaning, as their demographic data was primarily comprised of ‘unknown’ users as opposed to users whose age, gender, and other characteristics are known.

The unknown category declined significantly for many in 2017. We predict that in 2018, more users will fall into the known categories, making many of these advanced features useful to a larger number of advertisers.

Audience Targeting Adoption?

In the world of PPC there are two main milestones for examining features:

• When the feature is available for general use
• When it actually gets used (if ever) consistently

Most companies have tried remarketing across the display network. When we examine large amounts of accounts, many have sampled it for the search network, but have not fully adopted remarketing for search. When we look at similar lists, customer match and demographic insights, these features are not utilized in most accounts. The reason for the lack of adoption is threefold:

• A lack of understanding and education in this area
• Not enough data for small accounts to utilize
• Google has not provided tools to easily scale all the features across enterprise accounts

 With the change of Google’s interface and a large set of features becoming available, 2018 should be the year when Google makes it easier for people to adopt these features at scale. Hopefully, 2018 will also become the year that audience targeting is embraced across most PPC accounts.

2. Attribution Management Changes

If a user clicks on three search ads, two organic listings, reads your email and then converts, what visits get the credit for the conversion? That is the question attribution management seeks to answer. In 2018, we’ll see two main changes to attribution management.

Bidding by Attribution

At the moment, most companies bid by “Last click”. That means that if a person visits your website six times and then buys, the last click receives all of the credit. In a direct response world, that might be fine. In a long consumer journey process, that type of bidding ignores how a user found you initially. It also has a tendency to overvalue brand clicks, since brand clicks are commonly the last click before a conversion. So, bidding attribution needs more focus on fine distinction options to be useful in the PPC process.

In your conversion settings, you can choose how conversions are counted within your account by changing the attribution model from "Last click" to any of the other options:

Once you make this change, you will start to see fractional conversion data throughout your account. Any bids you make based upon conversion data will be made based upon the attribution model you choose. While this feature has been available for a while, it has been underutilized. With the rise in awareness about attribution and the ease of implementation, we will see many more accounts bidding based upon an attribution model that is not last click.

Cross-Channel Attribution Management Conversation

If you change your attribution model to be "Position based" and turn on CPA bidding, then you have automated your paid search bids by an attribution model. However, those bids completely ignore the social, email, organic and other clicks. The clicks used in this attribution model are primarily your PPC clicks. Cross-channel attribution analysis can now be done at the channel level. Once you start connecting the individual touchpoints—including what search queries, actual email offers, and social mentions occurred during the entire conversion cycle the models fail. This failure is not mathematical in nature, but rather a result of how scarce the pathways become when you dig that far into each click path before a conversion.

While we won’t see this type of bidding happen in 2018, what we should see is more conversations about how to properly handle these scenarios. We need to start building models and tools around analyzing the entire consumer pathway as it relates to necessary touchpoints, bids, and budget allocation.

3. Machine Learning Gets Smarter

Machine learning is highly useful during initial stages, however, it seems that it gets dumber over time instead of smarter. More data should mean better learning, but this is not always the case. In 2018, we should see machine learning get smarter. Consider this simple scenario: You have one ad in all of your ad groups that is appropriate to searchers over the entire year. A week before Black Friday, you create one ad in your ad groups that mentions holiday sales. Machine learning then starts looking at how these two ads perform. It will quickly learn that the ad that mentions holiday sales are faring much better than the ad that doesn’t mention a sale. So, it then serves the holiday ad almost all of the time.

Then, the holiday season passes and as January begins, special holiday sales ads are no longer appropriate. Any smart marketer would agree that it is time to pause the seasonal ad, however, the machine does not agree. It continues to serve the holiday sales ad most often because it’s still working based on the most recent data. Eventually, the machine will see that this ad isn’t doing well and will stop serving it, but this takes time. It could be March or April before machine learning catches up to current trends.

Machine learning is good at finding a pattern and making a decision. The problem is that it takes time to catch up or learn new patterns when the data or inputs change. Machine learning simply isn’t good at changing its mind.

AdWords has been using machine learning for several years and these types of scenarios are becoming more prevalent. Throughout 2018, we should see machine learning get smarter. In 2016-2017, machine learning needed lesser data to make a decision which makes for faster learning. Now it needs to learn that you can alter previous decisions when the scenarios have changed.

How’d We Do?

It can be fun to play fortune teller and make predictions about the future, as it gives you a reason to step back from the everyday minutiae and take a more holistic view and approach to PPC.  You realize that as a marketer it is important to look at the past, present, and future to better serve advertisers. Success depends on a delicate mix of many things, such as an examination of what has happened in the past, what engines are currently capable of coding, where conversations are going in the future and what will logically follow in the upcoming year.

In 2018, we plan to continue to write articles and produce webinars that examine PPC, current developments and how to make your account more profitable. Our final prediction is that 2018 will be a wonderful year for PPC marketers, and we hope that you continue to follow us as we share the latest tips for PPC success.