One of the most common questions we receive is along the lines of “how often will you be monitoring and optimizing our campaigns?” Marketers want to ensure their efforts are getting the necessary attention and most of us have been burned at some point in our careers, resulting in some professional trust issues (I’m looking at you, Google budgets).

The reality is, campaigns should be monitored regularly for pacing and performance and to intercept any trends that could be hurting results. We schedule frequent check-ins by at least two sets of eyes to ensure accuracy in setups and strong results throughout. But when it comes to optimization, the answer is a bit more complicated. Stop, start, increase bids, update targeting, add an A/B test – digital campaigns can turn on a dime, but that doesn’t mean they should.

The Case for Patience

Many buying platforms, such as Facebook, Google and various programmatic DSPs, have a learning phase during which the algorithms are trying to determine how best to achieve the results you’re aiming for. How long they stay in this phase can depend on a lot of factors, but ultimately the platform is looking for a stable sample size of results to optimize against. Facebook wants at least 50 optimization events before it hits its stride and depending on the size of the campaign and specific action, that could be hours or it could be weeks. 

If you are tinkering daily in a campaign, you may be resetting those learnings, essentially hitting restart on the algorithm each time. Just as importantly, if those changes are being made too frequently, it’s difficult to isolate their impact to determine if they are actually helping or hindering. It’s akin to doing a chemistry test while on a moving treadmill – it may work, but more likely it’s going to create a mess with inconsistent results. Isolating measurable variables and allowing enough time to observe changes with stable sample sizes are fundamental steps to using digital as a test-and-learn medium.

For shorter campaigns, we need to consider whether we can achieve the scale needed to optimize effectively or instead choose a different objective so the campaign can run effectively. For an eCommerce campaign, that might mean optimizing to cart adds or checkouts rather than purchases if sales volume is likely to be limited during the campaign window. For an awareness campaigns, that might mean choosing an optimization goal around distribution and impressions rather than a performance metric. 

While it’s fun to say we’ll be turning dials daily, sometimes, as humans, we just need to get out of the way and let our machine learning overlords take the wheel.