By now you know that meeting business goals is essential for the growth of a company and an important way to ensure you are getting the most out of your PPC program. So it also might come as no surprise if you find your boss breathing down your neck, asking “Why has our ROAS dropped from 150% to 130% over the past month?!” After doing a quick cost-benefit analysis to determine whether or not you can make a comfortable living by leaving your job right now -- and realizing you can’t -- you are now tasked with figuring out why Return on Ad Spend (ROAS) dropped so significantly, and ultimately why your PPC performance is suffering.
But where do you even start?
Ultimately, you need to identify the root cause of the issue. Your program might contain hundreds of campaigns, with thousands of ad groups, and tens of thousands of keywords or product groups. With so many avenues to explore, how on earth will you identify the root cause of PPC performance? We’ve provided an easy-to-follow guide.
Starting out, you’ll want to treat your root cause analysis like you would your PPC account structure. First, identify poorly performing accounts, followed by campaigns, ad groups and finally keywords/product groups. By following this structure, you’ll be able to thoroughly identify poor PPC performance impacting almost all aspects of your SEM program.
More often than not, you’ll find your campaign structure follows some variation of the 80/20 rule; 80% of spend will come from 20% of campaigns.
Clearly, it’s not the best use of your time if you find yourself spending countless hours exploring bad PPC performance in campaigns that make up 1% of your total spend. Instead, look at the PPC performance of each campaign, sort them by descending order of spend, and pick out those with the poorest PPC performance. Filtering out these campaigns early on will set you on a path to identify the root cause of poorly performing campaigns that have the most significant impact on your program.
Now that you’ve identified a poorly performing campaign or two, it’s time to look at the ad groups. Like campaigns, ad groups tend to have fragmented spend metrics in the sense that a smaller portion of ad groups are responsible for the majority of spend within that campaign. And as you did with the campaigns, you’ll need to calculate the PPC performance for all the ad groups, sort them in the order of descending cost, and find the highest-volume ad groups that experienced the largest drop in performance.
As you can probably guess at this point, keywords/product groups, like ad groups and campaigns, also tend to have an uneven spend distribution. Because of this, you should be able to pinpoint the keywords/product groups that are responsible for the drop in PPC performance. You can do this by finding the highest-volume keywords/product groups with the worst performance.
You are now at the point where you’ve found what is causing PPC performance to drop. For example, you might have found two or three keywords that were previously driving the majority of revenue, but have recently declined. Or, you might have found high click-volume product groups with diminishing conversion rates. But WHY is the PPC performance so bad from these segments? To determine this, take a few minutes to think about the cause. Perhaps most importantly, have you made any changes recently that might have led to poor performance?
In fact, there are a LOT of changes that can lead to this type of drop. Here are some of the most common:
Landing page changes
Making changes to landing pages is common, and essential for growth of your business. However, the wrong changes can be detrimental to your paid search performance. Questions to ask yourself following landing page changes include:
These types of changes will be evident in the quality score of the keyword, which will drop if the landing page changes aren’t good.
Business model changes
Have you made any changes to your business model recently that could affect the ROAS? For example, a change to item pricing could potentially hurt your performance. If you sell ten shirts one month for $15/shirt, and you spend $10/shirt, you made $150 and spent $100 for 150% ROAS. If the next month, you sold ten shirts but for $13/shirt and still pay $10/shirt, you would make $130 and spent $100, resulting in 130% ROAS.
In theory, conversion rates might have been the same month over month, but if you’ve reduced the price on an item, you could be artificially lowering your ROAS.
Conversely, have you NOT made any changes recently that might have led to poor performance?
In today’s competitive paid search space, you HAVE to constantly make changes to accommodate the evolving bid landscape. For example, a $10 CPC one day might put you on position 2 of the search results page. While the next month, the same CPC might only put you on position 4. You might have outdated CPC limits affecting your keywords/product groups/ad groups, which realistically need to be changed to help you achieve and maintain your goals.
Keeping the same business goals month over month might not be feasible if seasonality plays a big factor. For example, if you are selling Christmas trees, and your goal in December is 200% ROAS, you probably wouldn’t expect to hit that same metric in January, when search interest in Christmas trees is bound to drop off. Make sure you are adjusting your business goals in general to take seasonality into account.
If you have enough historical data, you can look at data from past years to get a sense of how relevant seasonality is. Some examples of KPIs to consider include impressions, clicks, spend, conversion rates, conversion volume, revenue per click, revenue per conversion, CPA, and ROAS.
Google search trends
These charts will provide insights of search interest over time, and can help you figure out where seasonality peaks and drops. For the Christmas trees example, you can see a significant drop in search interest between January 2017 and December 2018 for ‘Christmas trees’ across the US:
Now I understand the idea behind finding the root cause. But with so much data, what tools can I use to do this?
With so many resources available, effectively troubleshooting my data in the fastest way possible isn’t a question of what tool can I use, rather it’s a question of which tools should I use?
Small data sets
For data sets with <300,000 rows, Excel or Google Sheets will do the trick. Combining pivot tables along with sorting options will help you quickly identify high-volume, poorly-performing segments of your program.
Large data Sets
Excel isn’t compatible with larger data sets. Because of that, using a scripting language such as R, Python, or SAS will help you quickly identify poorly-performing segments of your program. There are inherent tradeoffs; if you don’t have programming experience, it will be a time investment to learn how to use one of these big data tools. On the bright side, once you learn how to use them, you can scale these checks to quickly identify poor-performing segments of your program in a matter of seconds.
To improve PPC performance you need to understand all parts of your program -- including, and perhaps especially, your weaknesses. Troubleshooting poor performance may seem like a daunting task. However, with the right approach and the right tools, you can effectively identify the root cause, monitor certain segments more closely and start making changes to bring your performance back up to speed.