Skip to main content

  • July 2, 2019

Humans vs Machines: When to Automate Your Paid Media Campaigns

Decorative Illustration Claude Shannon, the late American Mathematician, largely known as the ‘father of information theory,’ once remarked that he can “visualize a time when we will be to robots what dogs are to humans.” Artificial Intelligence (AI) and machine learning are no longer mere buzzwords. They’ve evolved to be an integral part of society, powering everything from self-driving cars to Amazon’s Alexa. Within the MarTech space, it’s now hard to find a software package or technology that doesn’t tout its use of machine learning. Indeed, machine learning and AI have been incredibly powerful in the paid media space, automating tasks that practitioners could do, but would rather not. This is especially true for tasks that take a disproportionate amount of time relative to the benefit earned.

Honestly, the idea of automating many of the daily, necessary aspects of paid media management sounds quite appealing. After all, who actually likes spending their afternoon in spreadsheet hell, writing ‘if/then’ statements to figure out max CPC bids and dayparting schedules? The reality, however, is that while the machines and technology are amazing – and necessary – we humans, the intelligent practitioners, are even more amazing.

It’s known that the paid media space is becoming more crowded and expensive. Machine learning technology, such as the bid and budget algorithms that we use here at Happy Cog, are great at finding efficiencies to mitigate those issues. But, at the end of the day, while the technology is evolving at an incredible rate, the accounts that will win are the ones that are managed by a combination of strategic thinkers AND smart technology.

Let’s start with the things that can (and frankly, should) be automated in a typical paid media account. In general, automation is perfect for time-consuming tasks that are largely repetitive in terms of pulling data and changing a variable based on that data. Some examples include:

Bid and Budget Decisions

Assuming your campaigns have enough click and conversion data, changes to max CPCs and daily budgets can absolutely be automated. Smart technology will track the impact of incremental changes to bids and budgets and learn where the proper equilibrium point between volume and efficiency lies. As accounts become more complex with bid adjustments to not only keywords, but device types, demographic characteristics, audience profile, and more, it becomes overwhelming for a paid media manager or marketer to keep up with the impact of each change. Not to mention, they also have to try to remember exactly what they’ve actually changed. As long as your bid strategy has constraints (for either CPC, CPA, or ROAS), let the machines do the number crunching. Just be sure to regularly monitor whether the set constraints really make sense for your business goals.

Finding Underperforming Creative and Keywords

When tying your paid media to business outcomes by using conversion tracking and analytics, a task that is perfect for automation is pausing underperforming ads, keywords, and even campaigns. Machines are capable of determining if something is performing below your goal or underperforming relative to other things in the account. Machines can also test the impact of modifications and then ultimately make a statistical determination that a campaign needs to be paused.

Adding Negative Keywords

If you are a Paid Media Manager, you’re likely familiar with the search query reports that can be run in Google and Bing’s platforms. Sifting through thousands of queries that your broad or phrase-matched keywords have fired for is the definition of tedious. Instead, you can leverage automation technology to bin and bucket irrelevant queries and add them to the appropriate campaigns and ad groups.

So what tasks are best left to actual humans?

The truth is, the more strategic decisions are much more difficult, if not outright impossible, to properly automate – at least at this point in time! The following should be led by the practitioner:

Account Structure

While technology is strong in terms of analyzing the performance of an account structure against a target or constraint, no machine is going to know how to properly break out campaigns and ad groups in the same way a marketer does. Account structure comes down to thinking about the sales funnel and customer journey, business priorities, and more. You should take the lead on this process.

Ad Copy

Tech tools can aptly perform testing and analyze the impact of a particular title and description combinations. They’re not great, however, at figuring out what your unique value proposition is or determining which call to action is going to resonate best with the market. Effective ad copy needs to appeal to a user’s emotions – which requires a deep understanding of the market. The machines simply aren’t there yet.

Audience Targeting

Nobody knows your customer like you! Marketers should take the lead on defining an audience, their demographics, and their affinities as an initial starting point based on tacit knowledge of the business. Then, you can use automation technology to validate and test your assumptions.

Non-Tangible Results

Every campaign in a paid media account should have a KPI-driven goal. Frequently, however, there are goals that go beyond metrics that can be tracked with analytics and conversion tracking. This could include brand lift, market penetration, and awareness building. Humans are needed to make the determination (often using non-web analytics data) on whether or not campaigns are meeting these secondary, but no less important, objectives.

The recurring theme here is that it’s best to leverage the technology when you can, but don’t let your campaigns run on autopilot. Also, be sure you feed the machines the proper inputs to let them optimize properly. Marketers should constantly be monitoring and fine-tuning their automation criterion to ensure that they make sense in light of seasonality, a changing business climate, and other external factors. So, the ideal approach to managing paid media circa 2019, isn’t to pit humans against machines, but rather to unite them for the best blend of strategy and efficiency.

Illustration by Ashlie Boyce

Back to Top