There is a profound shift happening in performance marketing right now. The fastest-growing businesses are moving from highly targeted campaigns, which are optimised manually by people, towards signals-based marketing, powered by machine learning and underpinned by automation.
The days of manually hacking your way to ad success are no more. Top direct-response advertisers are now leveraging a specific set of automated ad tactics to unlock new phases for growth.
Put simply, the machines are coming for your marketing plan.
The marketers who will win on Facebook and Instagram this Christmas will embrace this approach and set themselves up for success before we get to the craziness of peak season.
In recent years there has been a lot of talk of “precision marketing” where we layer in data from different sources to “build” highly specific target audiences. Director of Emerging Disruptors at Facebook, Jake Bailey, points of that performance advertisers have used this approach to manually hack their way to success. By learning a lot about their customers’ demographics, they’ve been able to target highly-specific audiences, find pockets of efficiency and relish in the success and control they have to manually optimise their campaigns. This was all good for a time. But times have changed, and this approach is no longer best-in-class.
This approach has limitations. The fact is, this manual, reactive approach just isn’t sustainable anymore, and it’s never been scalable in sustaining long-term growth. The problem is that audiences become very saturated with the same messaging, and gradually, marketing performance starts to plateau, and growth begins to decelerate. That’s because advertisers have created campaigns that are so granular there’s no room to further test or identify new customers. Many marketers have simply optimised themselves into a box, and it’s limited their potential to grow their businesses in a big way.
The fastest-growing companies recognise this and embrace the potential of machine learning to take the manual constraints of targeting off campaigns and automate their media and focus on more strategic opportunities. Many marketing activities can be simplified and performed through machine learning. Through automation and algorithms, machine learning is simplifying account management, driving better campaign performance, and scaling the business more quickly by finding new growth opportunities across different channels and placements.
It’s time to let go of old fashioned targeting
The new path to success on Facebook is to leverage as much signal data as possible, not as much targeting as possible. Your marketing team will simply never be able to pull levers or act on intent signals the way an algorithm can. It’s time to put your trust in machine learning.
That means letting go of the reactive approach to marketing where your teams manually perform reporting and optimisation. It means replacing it with automation that optimises campaigns in real-time based on hundreds (if not thousands) of intent signals and user behaviour to put the right message in front of the right person. Here, machine learning is simply better and faster when it comes to gathering insights, testing your ads, learning what works, and optimising for the best result.
The fastest-growing companies are combining what they know about their customers with our machine learning capabilities to match the right message to the right user at the right time. It’s unlocking new phases of growth at scale by automating advertising, audience selection, and placement for the best possible result.
Targeting is like salt; a little bit can make things better.
But you’ve already got a marketing team to do optimisations for you…right?
Whether you run marketing in-house or through an agency, consider all the time your marketers spend analysing and reporting campaign results. Many advertisers have 10s if not 100s of ad accounts on Facebook and often are optimising budgets separately. Add in manually changing creative, building out separate campaigns for each platform, allocating ad set budgets or creating redundant ad sets.
All of this effort can be eliminated by letting machine learning do the work. A better question to ask is, “What could your marketers be doing in place of this extensive manual effort?”
Where machine learning meets human creativity with brands
Machine learning is great at optimising what you have based on signals. It does this faster and more effectively than people can, but machine learning struggles to make leaps of creativity that take ideas into new and original areas.
This is the preserve of human beings.
No machine would come up with the idea to change the KFC logo to FCK delivering the best apology in corporate history, or take the iconic image of Colin Kaepernick and combine it with the words, “Believe in something, even if it means sacrificing everything”. These big creative leaps that are the bedrock of big brands and marketers.
Ways to embrace machine learning in marketing
The emerging disruptors team at Facebook looked at more than 100 of our top disruptive advertisers and identified the marketing tactics on the Facebook platform that are driving the best performance. Each uses machine learning, and each is impactful on its own.
Though, when used together, they are accelerating the growth of our top disruptor advertisers, driving better account performance, boosting return on investment and lowering CPA’s. Here are the key lessons for any marketer looking to transform direct-response efforts:
1 – Trust in automatic placements
You would never run separate TV campaigns for ITV, Channel 4 and Sky Arts, and yet advertisers continue to create separate campaigns for Facebook, Instagram and our other platforms. Certainly, there are differences between our platforms just as there are differences between TV channels, but there are more similarities and for most advertisers, most of the time automatically running across both dramatically improves performance. By running separate campaigns you essentially put guide rails in place that restricts the ad delivery systems opportunity to run advertising in the most effective placement. This also reduces the collection of signal data. This inadvertently drives up CPMs, lowers reach and increase the cost per sale. Smart marketers are using automatic placements to find the right person with the right message in the right place.
2 – Campaign budget optimisation
Rather than creating lots of individual campaigns, marketers should bring them together under one objective and let the machine determine the most efficient use of budget to get you the overall best results, spending less on underperforming ad sets. With campaign budget optimisation, advertisers can set one central campaign budget to optimise across ad sets and continuously distribute budget to the top-performing ad sets in real-time.
3 – Creative is still your competitive advantage
It is shocking that over 40% of video impressions delivered on Facebook are not optimised for a Feed environment. Spend time on craft. You can now upload different creative assets for Facebook, Instagram, Stories, Messenger, and In-stream. Building creative that works for the environment it’s served in, and the mindset that your target customers are in whilst they use it is paramount to any successful campaign. Equally, marketers need to think about how best they show people products that are tailored to their interests, whether or not they’ve been to your site or app. Using machine learning to find the correct people for each product, and always using up-to-date pricing and availability is driving real results for many brands.
This Christmas leave the machines in the office
Once set up – the temptation we all have is to meddle, particularly if we see our costs going up. This human intervention massively holds performance back.
When campaigns are set up they are initially in the “learning” phase. During this period the ad delivery is purposefully very broad, which allows data from many different audiences to be collected.
Once a reasonable number of conversions (around 50) have happened the ad delivery starts to optimise and performance rapidly improves. When someone intervenes during the learning phase, it resets and the learning phase has to be conducted again. Instead of meddling make sure your campaigns are set up in an optimal way and then relax and enjoy a mince pie.
To learn how you can achieve breakthrough performance with machine learning (and a whole lot more), read our Disruptors Annual Report, Build to Break: The State of Disruption.