When you’re running Google Search Ads, Facebook, or YouTube Ads, how do you know it’s time to pull the plug on a campaign?
If a campaign persistently underperforms its expectations and you wonder if it’s time to stop it, first, answer these questions:
Bad performance has to be relative to something else. If this campaign underperforms, what are you comparing it to? Make sure you have established a baseline as a definition of success. That might be a previous campaign performance, or it might be simply expectations that you feel are reasonable.
Once you have that baseline set, you’d know if a particular campaign fails relative to the benchmark (but still possibly achieving positive ROI) or failing in absolute terms.
The next question to ask yourself is if you’ve given that campaign enough time.
How do you know how much time that is? It really has to do with the data.
When you analyze the campaign KPIs, you want to know if your findings are “significant.” But business relevance, or practical significance, isn’t always the same thing as confidence that the results aren’t due purely to chance and have statistical significance.
You could take as an example the A/B tests you might be doing to improve a landing page’s conversion rate. To ensure statistical significance, you’d need to pay attention to the sample size.
If you know your baseline conversion rate, the minimum detectable effect you want to measure, and the confidence level – or percentage of statistical significance you want to achieve – you can calculate the sample size you need to run the experiment. With this data, you can also determine how long the experiment needs to run – how much traffic these pages will get – to have statistically significant results for your A/B tests.
Similarly, for paid ad campaigns, you need to make sure that you’ve given the campaign enough time to run and collect data to achieve a 95%+ confidence interval.This way, you’d know that if this campaign were to continue, you could expect that this poor performance will continue, and it's not about Google’s or Facebook’s algorithms needing more time to learn and optimize.
The next question to ask yourself is if any external factors affect the campaign results in question, that did not affect your baseline campaigns.
One factor could be seasonality – if this is the time of year when people are traditionally searching more often – or less frequently – for your services. Another could be a big social event that spreads across the web. Think of the US presidential elections, for example. Such an event focuses so much attention online that people simply don’t have the same bandwidth to register your online marketing campaigns.
So when you have a clear baseline, no external factors, and a large amount of data that show your campaign is not performing, then kill it and kill it fast!
The last question to ask yourself is where to allocate that campaign budget – it would make sense to either use it to test a different approach or increase the budget for your best-performing campaign.
If you’re curious, take a look at a solution we’ve implemented with SaaS clients that allows us to track off-site conversions and feed the data back into Google Analytics and Google Ads. With this solution, we can effectively train the Google Ads algorithms to optimize for paying customers, not for free-trial starts so that we can see the real Return on Ad Spend.
My job is to match talented, motivated marketers with high-growth companies, arm teams for success, and then get out of the way.
.jpg)
