5 things we learned from studying automation

Automation was recently identified as one of the strategic priorities of the global GroupM leadership. Within Ad Operations teams, trafficking automation is being supported as a way to reduce costs by reducing headcount and eliminating or reducing time-consuming and low value tasks within the organization. The extent of this impact is estimated to be large, but exact numbers are difficult to come by given the scarcity of data.

Wishing to gain a deeper understanding of how automation could benefit or hinder teams across GroupM, we spent the last six months studying the potential impacts of automation within the Canadian Ad Operations team—both through qualitative discussion and data based analysis.

Here are five lessons we hope will further automation efforts across GroupM:

1. Consider the whole process, even when evaluating just part of it

Automation is frequently understood as a complete change to a process. Perhaps our mental images are colored by visions of physical robots replacing jobs in their entirety but the reality is very different. Most automation is “task-based,” augmenting or eliminating a single task in a broader series of tasks that make up a process.

You can’t implement task-based automation without having a deep understanding of the entire process. When implementing automation, you should make a flowchart of the entire process and be clear about where the task fits, when it’s required, and when the task’s requirements may change based on the outcome of the tasks around it.

After mapping out our existing process, we discovered the greatest opportunities for improvement actually exist between tasks in our process versus the tasks themselves. Streamlining communication, or creating clearer roles and responsibilities, could create quick wins without the high costs of implementing new technology.

2. Be weary of relative gains without understanding the absolute benefit

Automation gains are frequently talked about in the relative reduction of time. For example, a tool might promise a reduction of time by 50%. However, relative gains can only be understood in the context of the absolute benefit, or how long that task actually takes on average.

In our work, we assumed trafficking would make up the largest portion of the Ad Ops team’s day-to-day tasks: if true, a 50% reduction could have a significant impact. We quickly realized our assumption was very far from the truth. Much of the team’s time is spent working with agencies, troubleshooting, and project management. Trafficking, the task affected by our studied automation, is only a small portion of the team’s actual workload.

A significant but relative time-reduction of a task that is only a small portion of the total process time is unlikely to deliver benefits to justify the costs and effort of implementing automation.

3. Decide on a primary goal and understand the trade-offs

Time-savings and standardization are the two most common expected benefits from automation, but we learned these two can be at odds with one another.

We found that completing the automated task, on average, takes the same amount of time regardless of the size of the input. Trafficking a large campaign takes just as long as a small one. This is obviously not the case for a person ie. the larger the campaign is, the longer it will typically take. Many of the campaigns we work on in Canada are so small, the time it takes to complete one manually is less than the overhead of the automation—that is to say, using automation in this circumstance actually makes the task longer. When automation was applied to all campaigns, regardless of size, we only saved 20% of what we expected to save when automation was applied to campaigns that would benefit only.

Although other benefits, such as error-reduction or process standardization, may make this an acceptable trade-off, it’s important to be clear about your primary goal, and how that may negatively impact your secondary goals.

4. Expect variability: there is no single answer that addresses everyone’s needs

In Canada, our clients’ campaigns are highly variable in size and complexity. Trafficking automation benefits are concentrated in clients with the largest and most complex campaigns. For us, the top ten clients make up 80% of our total potential automation gains.

While this is true in Canada, it may vary dramatically by market, team, and the type of automation being considered. Individual situations need to be analyzed with this in mind.

Given a very lopsided benefit, you may choose to dedicate resources to teams or markets who will benefit greatly from automation, rather than seeking a broad-based implementation that creates benefits unevenly and drives implementation costs up.

5. To achieve true long-term gains, we need long-term commitment

We have smart people on our teams, and those people are constantly handling small, ever-changing details that keep our processes running smoothly. Humans are great at handling variation and complexity, but automation is not. Automation that fails to account for the slight nuances of our jobs will be ignored and have its potential gains lost. It’s impossible to account for these details upfront, implementation requires a push and pull from the developers, the users, and all other process stakeholders.

It is ineffective to drop an automation tool into an existing process without bending the process to fit the automation. To those who are implementing automation; be prepared to provide dedicated change management resources, training, and the ability to incorporate user feedback. Making a tool available to users without any further plan is insufficient. Long-term dialogue between the automation’s developers, the prospective users, and team leadership is crucial.

Committing to a successful implementation requires deep thought, planning, and on-going support throughout the automation’s life cycle in order to truly achieve what automation promises.