Over the past few weeks, the PPC world has been focused primarily on one thing: Google’s new policy on sharing search term data. A notice on the Google Ads platform stated, “We are updating the search terms report to only include terms that were searched by a significant number of users. As a result, you may see fewer terms in your report going forward.” And in a statement to Search Engine Land, Google cited privacy concerns as the reason for the change.
So that is the background of the discussion—Google rolling out a vague new policy saying it will show less search query data of low-volume searches in an effort to protect privacy. We are now about three weeks removed from this change, and, all over, people are starting to see how much data Google is considering “low volume.”
This post is going to dive into what we, at Inflow, are seeing across our client accounts, as well as some thoughts I have around the future of Google Ads. There are many good posts out there that are showing data from their clients as well (including this great one from Seer), so we wanted to cut up our data differently to help and provide another context. For all data and graphs, we compared 9/3/2020–9/16/2020 (post-change) against 8/3/2020–8/16/2020 (pre-change) to see just what kind of effect this had on our clients.
Lead Generation vs. eCommerce
At Inflow, we focus primarily on eCommerce but do still have some legacy lead generation (lead gen) clients, so the first split we looked at is how this change is impacting each group.
Although lead gen accounts show a smaller percentage of missing clicks than eCommerce after the policy change, the percentage of missing search term cost data is much higher for lead gen accounts. This is likely caused by a combination of two things:
- Cost per Click (CPCs) for our lead gen accounts are generally higher than the eCommerce accounts.
- From what we have seen, the search terms we lost visibility on have a higher CPC than the ones that we can still see.
The other major takeaway from these data sets is that we went from seeing search term data on over 99% of conversions from lead gen queries to only seeing about 76%. This means that not only have we lost visibility on what is driving nearly 40% of spend in our accounts, we have also lost visibility on the good terms that are driving almost a quarter of all conversions.
The eCommerce side of things isn’t as drastic as the lead gen. However, we still went from having visibility on 91% of the spend to only having visibility on 71% of spend, and went from having visibility into 95% of conversions before to only 81% after the change. Missing 19% of search query data for what converts in eCommerce is concerning. We use search queries to develop our shopping-tiers strategy (our number 1 strategy here), not to mention optimizing the current campaigns to cut bad spend and increase conversions.
The second cut we did was to look at how the change impacted different account sizes. We broke our accounts into three tiers based on the ad spend in September. Tier 1 is low ad spend, Tier 2 is medium ad spend, and Tier 3 is high ad spend. We used these thresholds to get a relatively similar amount of accounts in each tier, and they tend to be a good indicator for us on the number of campaigns, types of campaigns, and strategies. Here is how each tier looked pre and post Google’s change to conceal more search query data.
The main insight from these cuts confirms what we were already starting to see in the accounts. Smaller accounts are impacted more by this change than larger accounts. Although the percentage of clicks are relatively close, the percentage of spend missing is 10% higher in both Tier 1 and Tier 2 than in Tier 3. Smaller advertisers are losing more visibility on where their budgets are being spent than advertisers that are spending more per month—this is crazy! Tier 1 and 2 advertisers have less budget and need that data to optimize their campaigns even more. Don’t get me wrong; ideally, that data doesn’t disappear for any account, but the fact that it is hurting smaller advertisers more is a major issue.
The other thing that popped out on this cut was within our Tier 2 (medium spend accounts). Why were we losing so much data in August—before this change? Was Google testing this change on a few accounts before rolling out across the platform? After digging into accounts where we saw a higher discrepancy in August, we can’t find any reason for that missing data. Has anyone else noticed accounts missing some of this search query data in August?
A Few Action Items
All right, so back to Austin Powers—what does it all mean? How is this changing how we are setting up and optimizing accounts?
The short answer is, we don’t know yet. There is currently a petition circulating (started by Marketing O’Clock) that is getting a lot of signatures, but I think this lack of data is the new normal, unfortunately, so we have to adapt. The specific action items will depend on the account and how much data is missing. We are looking into more testing. Previously I wrote about a target ROAS test on an account, but it may be time to revisit that test since the Google algorithms can still see these missing queries. Also, it will be more important than ever to use keyword data to pause/adjust bids on keywords with low CTRs or low conversion rates. Since you will not necessarily be able to see and negate queries coming to that keyword, it may be easier to turn off keywords with high costs and low conversions or where you see the click through rates far below other keywords in the account.
Other ideas we have seen that we will be trying out (most of these come from others in the industry and #ppcchat on Twitter):
- Use Bing search query data to add negatives to Google
- Add keyword suggestions from Google as negatives that don’t match your intent, because apparently, Google thinks they are relevant
- Enter your landing page into keyword planner and see what Google suggests and add the irrelevant ones as negatives
- Comb through old search queries and add as negatives bad queries that maybe weren’t significant spend-wasters at the time
Don’t forget about keeping up with your best practices:
- Analysis and bid adjustments on time of day, day of week, device;
- Pending data volume and analysis, consider breaking out the above into their own campaigns;
- Layer in in-market audiences and RLSAs;
- Review demographics performance and adjust accordingly;
- Regularly test ad copy;
- Utilize all ad extensions possible;
- Get even more granular with your keyword/ad group segmentation.
Ensuring you keep up with all of the above will help mitigate losses on missing search term data and negative keywords research.
Finally, we recommend that while you’re keeping a close eye on overall performance pre/post-change, make sure you keep context in mind. For instance, reviewing our account pre- vs. post-change, we’re seeing about 45% of our eCommerce accounts improve in ROAS. However, where we’re seeing decreases, for the most part, it’s either accounts that are seasonal, are expanding by testing new strategies (which tend to start at a lower ROAS), or are showing minor fluctuations. Our lead gen accounts are showing improvements on 40% of the accounts, where the other 60% are either accustomed to showing more significant swings in CPA month over month or are seasonal. We’ll keep you posted on any other insights as we continue testing—feel free to do the same!