Every few weeks or so, when speaking to someone that works in search engine marketing, the topic of campaign automation comes up. There are a ton of tools out there for automating a lot of the manual tasks we do in campaign management which can save PPC analytics lots of time. On top of just saving time, these tools always also promise that they can apply statistics-based optimization with some revolutionary algorithm that’s capable of doing a better job than any human analyst can ever.

I get it, the notion of doing less work, and having it done more accurately is pretty enticing. That’s why so many people look at me like I’m crazy when I tell them that aside for some basic scripts, I generally stay away from automated tools, including those offered by Google AdWords like the CPA optimizer.

Once in a while, we’ll test some automation in instances where it seems to makes sense, but the results have yet to convert me into a believer. Recently, we ran an experiment where we tested using Google’s CPA optimizer against manual bidding using Google’s “drafts and experiments” feature. We tested a lead generation campaign and split the traffic 50/50. Half the traffic used our manual bidding and the other half ran on the CPA optimizer with a target cost per acquisition that was lower than what we were currently paying per customer. As you can see below, we received more conversions with the CPA optimizer, but at a cost that’s 30% higher per lead, with statistical significance.

On occasion, I find that automation does work a little better, but more often than not, it simply does not perform as well.

I have 3 basic reasons for preferring work-intensive, geeky, old fashioned analysis on MS Excel to automation.

1 – Conversion Data Can Get Corrupted

Automated tools rely on conversion codes (aka Pixels) to deliver conversion information to their analysis tools. Then they base all these fancy statistical models off this data and run their algorithms to make campaign changes. Analysts that manually manage campaigns work off the same data. But sometimes, there’s a glitch. Like when some developer inadvertently tampers with the code when he/she is working on the website. The result is that far fewer conversions than actually took place are reported. If you’re an analyst seeing this, you can adjust by assuming conversions are happening or by checking some other conversion data source like another analytics tool or the site shopping cart. Then you just make manual adjustments to your data until the bug is fixed.

But automated tools don’t have that luxury. They see a problem and start acting like it’s a doomsday scenario. There are ways to limit how aggressively these tools make changes, or even to partially take them off of auto-pilot and have an analyst take-over. But either way you slice it, the data models created by the tools during this period become corrupt and will impact decisions made in the future.

2 – They Can’t Adjust for Seasonality the Way People Can

Most industries experience some level of seasonality in traffic and conversion rates. So, while automated tools allow you to make campaign adjustments in anticipation of these trends, they don’t allow you to change your data after the fact. This means that you can prevent the campaign from losing a lot of money, but you can’t tell your tool that this period, which is some cases can be several months, is not representative of your long-term conversion rates. So, the tool ends up cutting your traffic because it thinks there’s something wrong, whereas a human being would adjust his/her target ROI metrics from day to day or week to week. Humans can also adjust for conversion rates changes that result from things like super bowl Sunday or a hurricane that has their customers glued to their screens for updates.

3 – Their Answer to Every Problem is Cutting Your Traffic

Most automated conversion tools focus on managing bids. So when they see a drop in conversion rates or ROI, their default action is to cut your bids and lower your traffic. A savvy campaign manager on the other hand, knows to look for problematic search terms and ads or recent campaign or landing page changes that may have caused the drop. They understand that businesses need long-term consistency from the leads and sales generated by their campaigns, and cutting down traffic is not something that should be done at the drop of a hat.

So the bottom line is, I know that my aversion to automated tools sometimes makes me sound like some dinosaur that just hates change, and that may be true. But I just don’t think the technology is there yet to fully execute at the same level as a seasoned PPC manager. Maybe AI will eventually change that reality, but my instinct tells me it’s not going to happen for a while.