Not sure if machine learning has a place in your business and the way you report and optimise your marketing? The team at Distract share their thoughts.
As technology progresses, machine learning is becoming more ingrained in our everyday lives.
After all, it was found that 40% of marketers believe better data would improve their marketing outputs.
And we understand why.
Access to better data can afford you more confidence to make data-driven decisions. And with machine learning in place, you could even get intuitive recommendations based on campaign performance.
So in this blog, we’ll explore:
Let’s get stuck in.
Machine learning is a form of artificial intelligence that allows software applications to better predict outcomes by using historical data and analysis.
If you work in the digital marketing space, you know that machine learning is unavoidable and in many cases, can lead to significantly increased business performance.
There are all kinds of artificial intelligence tools out there that support businesses with machine learning.
From a marketing perspective, there are five key types of machine learning software out there:
Let’s look at each in a little more detail.
This is pretty much what it says on the tin. Recommendation machine learning can use purchase history data to offer products to buyers that other buyers like them have purchased.
By offering personalised offers and recommendations, businesses can uplift sales and refine their targeting.
Generally speaking, paid advertising targets users based on particular demographics or historic activity.
There’s also predictive targeting, however, which is where you can target users who are most likely to make a purchase.
LTV is calculated as part of a report meaning it’s retrospective. Ideally, LTV should be calculated before a customer leaves. And even better would be to be able to predict when a particular type of customer will consider churning and to use that to kickstart a campaign to keep them onboard.
It can also be used to measure the impact of marketing campaigns on LTV as opposed to just leads or sales.
Similar to the above, churn rate forecasting gives businesses predictive marketing opportunities to keep customers onboarded for longer and to identify key causes of churn early.
Last but not least, reporting and big data analysis is a key future trend for machine learning in marketing.
While tools like Google Analytics serve you all of your marketing data in one place, it doesn’t offer much in the way of insights or recommendations.
We predict an uplift of software offering more predictive analytics from a marketing standpoint.
Historically, marketers have relied on tools like Google Analytics and native reporting within tools like Google Ads and Facebook Business Manager to understand the quality of their marketing campaigns, particularly their paid ads.
But with the death of the third-party cookie and privacy updates coming into place, analysing user metrics for optimisation is becoming harder.
Related: What is first-party data and how to collect it
Due to this, marketing platforms are leaning into machine learning, taking sensitive data out of marketers’ hands and forcing them to trust the smart optimisations that the platforms are making.
Marketing platforms such as Google and Facebook allow you to pick a goal for your advertising campaigns to maximise for certain conversions, reach a specific ROAS, or simply drive traffic to your website. These platforms claim they can use detailed metrics (which they do not divulge exclusively) about users to reach these goals for you.
However, Google has drastically cut the amount of user search terms you can see, and with the transition over to GA4, it will be near impossible to analyse the navigation of your website without having a large data set.
Related: Key differences between Universal Analytics and GA4
iOS 14 has had another crucial impact on Facebook advertising, with demographic targeting no longer available when looking at conversions.
Limitations like this leave marketers with no choice other than to trust the machine learning systems.
Related: How to mitigate the impact of iOS 14.5
When you’re entrusting these platforms to optimise your campaigns to reach your marketing objectives, it is important to ensure the data you provide them is correct.
Making sure your conversions are set up correctly and are in your platforms will set you off on the right foot and mean the platforms are optimising for the right thing. If you do not provide platforms with relevant data, the machine learning process will be much longer. It’s also likely that won’t see the full benefits.
The more data you can provide for machine learning, the better. Finding significant similarities in a group of 10,000 users will be more accurate than in a group of 100, meaning smart optimisations based on a large group are more likely to result in a positive outcome.
Google has many campaign types where machine learning is at the forefront. The newest of which is Performance Max, where you enter a number of assets and Google advertises across multiple Google platforms on your behalf to users it decides are likely to bring you conversions.
Whilst this is an extreme form of machine learning, you don’t have to dive right in and trust Google to take everything out of your hands.
Automated bidding strategies are also a form of machine learning. You can keep control of your keywords and copy whilst allowing Google to show those ads to people that will convert.
Providing Google with the most amount of data possible will put you in the best place to bring in the correct users. One key benefit of using Google platforms is that all Google tools work seamlessly together. This makes conversions easy to set up using Google Tag Manager, Google Analytics, or Google Ads itself. Once a conversion is added to Google Ads, you will benefit from machine learning and smart optimisation right away. Google will use machine learning to achieve your goals depending on your bidding strategy.
With Ruler, calls or forms can be imported into Google Ads in a number of ways. Ruler calls can be automatically imported into Google Analytics as an event which can then be created as a goal and imported into Google Ads.
Related: How to track phone calls in Google Analytics
Alternatively, both form and call data can be exported from Ruler and uploaded into Ads. This method will backdate conversions and associate them with ads using GCLIDs. Google can then use this information to dictate smart decisions.
Getting third-party platforms to talk with Facebook and other social media platforms can be difficult. It is best practice with social platforms to add UTMs to each ad so action on the website can be tracked back to the social campaign, ad group and specific ad.
With Ruler, these UTMs can be analysed to determine the number of calls and forms which can be attributed back.
However, another way in which we can utilise machine learning is by creating custom audiences. Audiences can be uploaded to social platforms using identifiers such as phone numbers and email addresses. By doing this, this custom audience can then be targeted for remarketing or for look-a-like audiences. These look-a-like audiences are where machine learning comes in.
Social platforms will analyse your custom audience list and use key markers to identify what these users have in common. They can then create a look-a-like audience based on additional users who also possess these key markers. In theory, this audience will be similar to your remarketing list, meaning they are more likely to take similar actions.
The more data you can provide to any platform, the better the smart optimisations will be. It is recommended to have a list of 300+ users when creating a custom audience.
If it is not feasible to create a list of past purchasers or converters, then look at the step before conversion or the step before that. Creating an audience of users who have completed key actions on your site (whether that be the last step in your conversion journey or not), makes any similar audiences more likely to be one step closer to converting than a completely new, broad audience.
Machine learning cannot be escaped in the digital marketing space.
Be proactive and set up audiences and tracking now. Later down the line, these can be used in campaigns, and platforms will already have data to create decisions.
Remember, attribution tools are the first stepping stone to machine learning. Invest in one now to get a head start on your marketing data.
Ruler Analytics is a leading attribution tool that can integrate with all your key marketing tools. Book a demo to learn more.
Or, if you need support creating marketing campaigns end to end, then reach out to the Distract team.
About Hannah Langton
Hannah is a Senior Search Advertising Account Manager at Distract. With five years’ experience, Hannah is looking to the future of digital advertising and machine learning.
About Distract
Distract is a creative digital marketing and advertising agency based in the city of Lincoln in the East Midlands. Distract prides itself on innovation and creativity, enabling clients in the B2B and B2C sectors to stand out amongst their competition.
Distract’s Search and Social teams work creatively with clients to create campaigns that constantly deliver sales leads and enquiries and raise brand awareness. Distract achieves this through a combination of clear personal branding, innovative strategies and the use of creative elements produced through design, video and social media.