5 Ways to Adapt Your PPC Campaigns To Machine Learning

What is machine learning in PPC?

Machine learning has taken off over the last few years and transformed the world of PPC through automation. Thankfully it’s becoming much rarer for marketers to have to spend hours laboriously working on bid management, targeting and timings of campaigns. As a result of advancements in machine learning across Google Ads, Facebook Ads and other ad platforms, PPC automation has become a lot more efficient.

These platforms use a vast number of data points, including hundreds of 1st party signals, to try to serve your ads to the right people, at the right time, and at the right cost for your goals. It streamlines the process of testing and optimising, but it can only go so far, and can lack the wider context of a business strategy that only human managers have access to.

Over the last 10 years, Clickoo has been at the forefront of PPC automation adoption. We believe in working alongside the machine, using both human and AI skill sets to optimise campaign performance, instead of giving it total control or rather working against it. So what are the big challenges facing your PPC account manager in 2021 and how can you adapt your campaigns to machine learning?

1. Identify the right PPC bid strategy for your goals

Target ROAS, Target CPA, Maximise Conversions… they all sound like they’ll make your life easier, but knowing when to use each one can be tricky. Often when you create a new paid search campaign, Google will automatically recommend that you use “Max Conversions”, which essentially means you set a budget and the machine will decide how much to bid on each of your keywords following first-party intent signals. The idea is that by generating as many conversions as the machine thinks is possible for your budget, you will gather more data, and the more data you have, the more the algorithm will be able to refine and improve campaign performance. Indeed, an effective approach can often be to start with a broad objective like Max Clicks or Impression Share, then gradually work to a conversion-based strategy.

However, we don’t always advise trusting Google’s bid strategy recommendations blindly. PPC specialists should have the knowledge and experience to choose the right strategy for different businesses, taking into account contextual factors which the machine doesn’t know. For example, businesses that are highly seasonal or operate in niche sectors with low search volume can be much trickier to manage and may not be suitable for automated bid management strategies. It’s also worth taking into consideration the learning period – it can take several weeks for machine learning to take effect, so don’t expect immediate results.

2. Build an account structure that supports machine learning

Historically, PPC campaigns were set up with a granular account structure. The old advice was to split them as much as possible, with the idea that it allowed you to see exactly what worked and what didn’t, and have the most control over your investment.

However, when machine learning is involved a slightly different approach is needed. If you get too granular, you’ll limit the data the machine has access to. This will ultimately hinder its ability to optimise your ad campaigns’ performance. By grouping more campaigns together strategically, you can feed the machine with more data so it can learn quickly, adapt to different intent signals, and end up with better results.

3. Adapt to the new loss of search terms visibility

In September 2020 Google announced it would be reducing the amount of search terms visible in accounts to “only include terms that were searched by a significant number of users”. PPC account managers relied on search query reports to filter out irrelevant or unconverting terms as negative keywords, thereby improving the quality of traffic to a landing page.

Essentially Google hopes that you trust them enough that even if you are targeting broad or phrase match keywords, the algorithm will learn to only serve ads on the most relevant search terms. We definitely wouldn’t advise trusting the machine 100% on this one just yet, and your PPC manager should be finding other ways of adding negative keywords or adjusting the bids on your broad match campaigns to retain some control over investment.

4. Develop a more holistic view of attribution

Another key way to adapt your PPC campaigns to machine learning, given the importance of feeding the algorithm with quality data, is allowing it to access data from the whole user journey, rather than just the last ad they clicked.

If you only let the machine optimise for final interactions, your conversion volumes will be restricted. Despite the vital part they play in starting the customer’s journey to conversion, upper-funnel terms won’t get the level of investment they may need to take your performance to the next level and really get the most out of the machine. It’s worth exploring other attribution models available through Google Ads and Analytics, as well as third-party tools, to get a better understanding of the role of your paid ads at each stage of the funnel.

5. Implement effective creative testing

With the machine taking over a lot of the heavy lifting when it comes to data management, human account managers have more time to create and test engaging ad copy and imagery for search ads, carrying out A/B testing at a greater scale and with machine assisted precision.

Be careful not to go overboard on testing all your campaigns’ creative at the same time if you’re working to a restricted budget. Otherwise, you’ll end up with lengthy learning periods, as each ad variation needs to be tested against each different targeting criteria enough times at a live auction to gain statistically significant data.

The machine has come a long way but it can’t write or perfect ad copy tailored to specific audiences and goals (just yet..!)

What is the future of PPC automation?

Machine learning is only going to get smarter and smarter. Eventually, we’ll have less control over where and how budgets are spent, but we’ll also have more data points and more personalised ad experiences. We can’t fight PPC automation, but we can embrace it and form a symbiotic human-AI partnership, ultimately creating ad campaigns that work for both humans and ‘machines’!

Lydia Rutter
Lydia Rutter
Lydia Rutter is an SEO Executive at Tecmark who joined us in 2019. She has a degree in Public Relations from the University of Oklahoma, and an MSc in Digital Marketing from the University of Salford. She brings us experience in digital marketing for the medical and healthcare industry, and in agency settings.

Are you ready for more leads, more click-throughs, and more sales?

Speak to one of our experts today.

Related articles

Call Now Email Now