Select Page

How to Diagnose Performance Issues in Google Ads

Google Ads is a very complex system with nearly infinite combinations of settings, keywords, ads, bidding strategies, and targeting options.  Never mind the fact that it’s a dynamic auction-based advertising system.  A competing advertiser’s decision could dramatically change the performance of your program, and you do not have visibility, insight, or voting rights in any of their decisions.  It’s an amazing melting pot of features that leads to an ever-changing environment with new challenges and opportunities presented everyday.

But, because of the complexity, it’s easy to get totally overwhelmed when an issue arises.  Where do I start?  There are 1 million things that could have gone wrong (literally), how can I possibly find the cause before my 11am meeting with my boss?

It can be paralyzing, so over the years, I’ve developed a checklist of things to look at to try to hone in on exactly what is driving the performance change.  Here are the 9 things that I look at first, along with details of what I look for:

  • Search Query Reports – These are GOLD. They tell you exactly what users are typing into Google in order to find your ads.
    • Make sure you sort by clicks, impressions, and revenue/profit (you’ll need to build a custom metric to see profit = [conversion value]-[cost]). 
    • Click volume should be coming from relevant searches.
    • Add negatives for any search terms with super high impressions/no (or low) clicks (this lowers your eCTR, and therefore Quality Score).
    • And use revenue/profit to discover new opportunities – build out keywords around the best performing ones – and ensure you’re not lighting cash on fire in a concentrated few terms (or term types).
    • Bonus Tip: Week over week (WoW) and month over month (MoM) analyses can show you what has changed. What kinds of terms were getting volume last week?  What about this week?  How are they different?
  • Keywords –
    • Make sure you sort by clicks, impressions, and revenue/profit (you’ll need to build a custom metric to see profit = [conversion value]-[cost]). 
    • Try bucketing keywords into groups (Pro Tip: you can use the tag feature to keep track of this in Google Ads) and then analyzing the performance of various buckets.  This might help you see larger trends that might be masked at the keyword level.
      • Example buckets – Head/Neck/Tail terms, high purchase intent vs low purchase intent, city-modified terms vs generic terms, exact/phrase/broad matchtypes (this one is pre-built into Google Ads), branded vs non-branded, groups of similar phrases
    • Compare keywords to search queries – is there a small subset of keywords that are causing the majority of the problematic matches? Do they have enough upside to make the ongoing management worthwhile?
    • Pause and/or add negatives for any keywords that are consistently super high impressions/no (or low) clicks (this lowers your eCTR, and therefore Quality Score).
    • Bonus Tip: Week over week (WoW) and month over month (MoM) analyses can show you what has changed. What kinds of terms were getting volume last week?  What about this week?  How are they different?
  • Bid Strategy –
    • If you’re using autobidders (which I recommend in most cases), make sure the actual is converging on the target over time.  If it is not, and the above checks seem reasonable, you may need to adjust bid limits to pull the bidders into line.
    • If the bidder is converging on the target that you set, but ROI is not backing out, re-evaluate your target and adjust as needed
    • Bonus Tip: We saw success in a dual approach to Target CPA – all campaigns were kept in 1 portfolio strategy, but every adgroup received an individualized target.  If it was a high-volume adgroup, the target was based on that adgroup’s performance.  If it was a low-volume adgroup, an average target was computed across multiple adgroups.  Using this strategy, the target in the portfolio settings is ignored and the 
  • Ads –
    • Make sure you sort by clicks, impressions, and revenue/profit (you’ll need to build a custom metric to see profit = [conversion value]-[cost])
    • Look for outliers that are getting the lion share of volume  Is there a reason for it? Does your ad copy appropriately pre-qualify clicks to try to eliminate unnecessary spending? For example, is there an unrealistic claim that is capturing people’s attention? Do you have automated language (through Dynamic Keyword Insertion, etc) that may be creating unintended claims?
    • Were any ads recently disapproved?
  • Device –
    • Did the click mix shift? Most websites will convert very differently on mobile and desktop.  If you’re just looking at high-level metrics, a change in device mix can make summary metrics look very crazy.
  • Lander Performance –
    • The key to making paid ads (of any kind) work is a website that converts visitors into customers. So the first place to look is conversion rate – [number of conversions] / [number of visits].  If there was a change in this metric, and the checks above seemed normal, there may be a bug on site or a test/feature that is dramatically changing user behavior.
    • On top of conversion rate, it’s also wise to take a look at average revenue per conversion.  Was a low price-point product just rolled out that’s leading to a spike in conversion rate but a decrease in revenue per conversion? 
  • Auction Insights –
    • This gem of a report gives you visibility into the dynamics of the auctions that you’re competing in.
    • If CPCs skyrocket (or fall) and there aren’t any obvious factors contributing to that shift – check Auction Insights.  A competitor may have entered or exited the auction, changing the calculus on how much you will be charged based on your page rank.
  • Top vs Other –
    • Getting into the top 3 spots can have a massive impact on your click volume.
    • While average position was decommissioned by Google in September 2019, they replaced it with “Top” and “Absolute Top” impression share metrics.  You can use these to see if your showing up at the top of the page more often.
    • Be careful – #1 isn’t always best. While you’ll certainly see a massive increase in click volume by increasing your absolute top impression share, you may see a decrease in conversion rate because unqualified users are clicking the first result out of habit.  So you may run into profitability issues as you move up the page.
  • Audience –
    • People that are familiar with your brand are going to be easier to convert. That’s one of the primary reasons that remarketing is such a powerful tool. So if there’s a shift in audience mix, you may see a corresponding shift in performance.
    • Additionally, there will be differences in how audiences convert for your program.  Whether that’s demographic audiences, affinity audiences, in-market, etc – it’s likely that some will perform better than others.  And therefore, if you’re struggling with performance, it might make sense to exclude certain audiences, especially if you’re working with a constrained budget.  Autobidders should take care of this automatically (over time), but that’s not 100% guaranteed.  So it may be worth manually making the adjustment to shift the marketing dollars into higher value segments. 

As I said, there are (nearly) infinite things that you can dissect in the search ads ecosystem.  But that is my checklist that I start with when trying to understand performance and one of the above bullets will usually highlight the issue so we can dig in deeper.  You probably won’t understand the full story from this list, but you will probably have one or two solid hypotheses that you can then go investigate further.

And, based on my experience, it’s almost always well received when I walk into that 11am meeting and say, “There’s a problem, I’m on it and have identified where the issue is.  I don’t have a complete picture yet, but here’s what I know so far and this is what I’m looking into next in order to fully understand what happened and how to fix it.”

Now, compare that to “I know there’s a problem, I spent the morning looking at a bunch of different things but I haven’t been able to find the cause yet. I’ll keep looking.”

You come across as WAY more confident (and capable) in the first scenario.  And, arguably, the only difference is you approached the diagnostic work in a methodical way.  You eventually would look at all the same stuff – but having a process to cut through the noise and highlight the common offenders helps you work more efficiently.  And efficiency means you get to approach your boss with a cohesive story that puts their mind at ease.