Automated advertising: How AI responds when $h8# hits the fan.

By Mike Hill, Paid Search and Social Supervisor

Automation. Bid strategies. Dynamic Ads. These are all buzzwords that have appeared in most digital marketing plans for the past several years. And, for good reason: case study after case study provided by all major advertising platforms tout impressive gains when utilizing automated advertising across search and social.

The general concept of automation in advertising revolves around using the massive quantities of data collected by each platform to optimize campaigns. Platforms can adapt to both technical data (time of day, day of the week, and browser) in addition to personal data (gender, age, and previous user interests).

But using data gathered over such long periods can also be an issue. What happens when user behavior changes rapidly? When privacy laws change how audiences can be targeted and bid on? When a brand stumbles into controversy? When a pandemic completely alters how consumers interact on the web and off it? What happens with all this automation when $h8# hits the fan? And what can be done about it?

Automation doesn’t mean autopilot

User behavior trends have shifted because of COVID-19 with everyone flocking online while adhering to stay-at-home orders. This is reflected across search and social. Facebook and Snapchat have reported increases in total users and user interaction. On a C-K search account, searches around toilet paper, hand sanitizer and cleaning chemicals have increased over 300%. These increases in volume have caused large fluctuations in KPI performance for brands. This leaves brands that rely on standard automation particularly vulnerable.

Standard Automation Campaign Vulnerabilities

• Automated bidding can cyclically bid higher on keywords deemed more efficient.
• Automated creative rotations can continue to show messaging that had been historically effective but fails to resonate with large user changes.
• Automated budgeting can spend inconsistently and in areas that may not be ideal for a brand.

Even outside of this extraordinary time, volume changes can occur over the course of a campaign flight and can throw a wrench in automation. For example, search volume for shoes increased over 20% in a single day during Black Friday and Cyber Monday. While this is a good increase in volume, negative brand news can also lead to large shifts in search and social that can impact automation. Google Search Trends shows interest in Boeing increased over 900% after the Boeing Max 737 crashes. Brands like H&M, Volkswagen, and Papa John’s all had significant negative press over various issues that significantly changed search volume and social interaction.

Regardless of positive or negative changes in volume, these swings can happen to any brand and can significantly affect automation. But, by applying some best practices throughout planning and campaign creation, brands can maximize performance during positive influxes and minimize impact during negative influxes.

AI Still Needs HI (human intelligence)

Strong team collaboration and planning from the start of a campaign, and continued maintenance, will lead to campaign success. Here’s what to consider:

Automated bidding

Automated bid strategies should be carefully planned to take into account KPIs, budget, platform user behavior, recent trends and more. This planning will involve which specific strategies to select and what bid caps may be placed on each bid strategy. Choosing the right strategy and caps initially is imperative for success.

Additionally, these bid strategies should be monitored throughout a campaign flight, especially during drastic changes in volume. Campaign managers should check to make sure the tactics in place are performing on the brand-aligned keywords and/or audiences with appropriate allocations. Bid caps likely need to be changed as volume changes.

Manual bidding should be used as a tool to “reset” campaigns after times of high volume, as the automated systems are typically still expecting those levels of volume. For example, WordStream recommends running bids manually after Black Friday and Cyber Monday due to the influx of data.

Automated creative

It’s no surprise that, when utilizing automated creative delivery, creative should be monitored to provide insight on the performance of the various executions. This is especially important during large volume changes as new users may interact differently with different creative.

For example, recent fitness trends have shifted from gyms to online coaching, online classes and home workout equipment. By having multiple creative executions at the start of a campaign, a fitness brand could capitalize on home workout equipment interest without any changes.

During times of crisis, brands may want to add or pause creative. For example, Boeing now has search ads addressing the Max 737 issues.

Automated budgeting

When running multiple campaigns, as volume fluctuates, brands may want to shift spending. Platforms have addressed this by allowing for both total campaign budget caps and daily budget caps.

However, most of the daily caps aren’t necessarily hard caps as platforms anticipate volume changes and can allow campaigns to overspend or underspend within certain percentages on a single day. With that in mind, budget pacing and monitoring are especially essential with automated budgeting in times of fluctuating volume.