The Complete AdWords Audit Part 3: Campaign Settings & Bid Adjustments
One of the first high level checks you should do when performing an AdWords audit is check the campaign settings. There are a couple of reasons for this:
- The default settings as suggested by Google often aren’t optimal.
- Suboptimal settings apply to all keywords and ads in the campaign, which means it can cause a lot of (hidden) waste or missed opportunities.
- Since enhanced campaigns a lot has changed in the settings tab, and chances are your campaigns aren’t fully taking advantage of all the new settings and options.
- It’s relatively easy and quick to fix.
To get a quick feel for the current campaign settings, go right to the Settings tab in the interface, select All enabled campaigns and add all possible columns in this view (except campaign start & end date):
Now you can scroll through your campaigns in this view and answer the following questions:
- Are the location and language settings as they should be?
- Do any campaigns target all networks (search and display, which is never good)?
- Is there any ad scheduling in place?
- Do the bid strategies make sense?
- Does the delivery method (standard or accelerated) make sense?
- Which ad rotation setting is being used?
- Are there any active bid adjustments in place?
I’ll cover the different bid strategies in the Bid Management part of this series. For the rest of the settings, how I see it, there are three kinds of campaign settings and each need their own approach to be set optimally:
- Settings you should always apply.
- Settings you should usually apply, but may have good reasons not to.
- Settings where you need statistically significant data before you should apply or change them.
Let’s walk through each of these categories.
Settings you should always apply
These are the no-brainers you still want to check, just to be sure:
Use the ‘All features’ setting when selecting your Search campaign type. This will give you access to all the advanced options you really need to get the most out of your campaign settings.
Never target the Search & Display Network in the same campaign. You know this already as it’s in every best practice list and for reasons of completeness I also have to mention it in this post.
It’s quite strange Google still gives you this option, just be sure to never actually use it. Search and Display are so totally different that analyzing aggregate performance is meaningless and optimizing for both networks from a single campaign would be a huge pain without the well-deserved rewards. Even the recently launched Search Network with Display Select doesn’t convince me this campaign type should be used by advertisers that want full control and quick insights of their search & display performance.
Target the right locations. This may sound utterly obvious, but there are still many accounts out there spending money in the wrong places. So make sure the geographic locations you’re targeting are in line with the reach of your business. And if you’re an international advertiser, it’s probably best not to target multiple countries from a single campaign.
For example, the US and the UK are both English-speaking countries, but by targeting them in the same campaign, you’ll show the same ads to American and British users, which makes it impossible to create ads that take the many differences between British and American English into account.
Also, be sure to understand how ads are matched to geographic locations and how you can refine this with the advanced location options.
Target the right languages. You may be inclined to just target the language of the ads in the campaign, and this is also what Google recommends. If you’re only targeting the US, selecting English is usually good enough. However, if you’re targeting countries outside the US it’s important to know that users use Google.com as well as their local Google domain. So in those cases it’s advisable to target English (even if you advertise in the local language) as well as the native language (even if you advertise in English). Of course, keep monitoring your search queries and your geographic performance (in the Dimensions tab) to ensure you’re showing for the right queries in the right places.
Settings you should usually apply
You should usually set your delivery method to accelerated and rotate your ads. However, there are some exceptions and implications you may want to consider before blindly using these settings.
Delivery method: for mature and profitable campaigns with no budget constraints (and they shouldn’t, as they’re profitable), accelerated is the way to go. That way, you’ll be sure to capture any (temporary) increase in search volume, as long as your daily budget allows it. One such example would be your branded campaigns, always set those to accelerated with a more than high enough daily budget.
If you’re on a tight budget or just launched a new campaign, you can set your delivery method to standard to spread your ad display throughout the day.
Most experts probably expected this to be in the ‘Settings you should always apply’ section where it should say to use rotate indefinitely (or evenly).
Indeed, if you want full control of your ad testing (and you should), rotate is the way to go. This also means you’ll have the responsibility to take the time to regularly pick the winning ads in each ad group.
If you know beforehand you don’t have the time (or the click volume) to frequently do this in a specific campaign, this could be a reason to choose for one of the optimize settings. Of course, this isn’t optimal, but it’s better than letting a terrible ad run 50% of the time because you’ve set the campaign to rotate and didn’t come back soon enough to analyze the results.
Whatever setting you choose, you should regularly pause the significant worst performing ad in each ad group and write a new one to beat the control. In other words: you should always be testing.
Google could make our lives much easier by automatically treating each ad test as a campaign experiment, showing us how significant the differences in performance are and even warning us when an ad group has a significant worst performing ad (in terms of CTR or conversions per impression / revenue per impression). That way, we’ll always know when and where it’s time for a new ad. So for any Googlers reading this, feel free to consider this a feature request.
In the meantime we’ll have to do this ourselves when rotating and luckily there are some useful tools out there to help us, which I’ll cover in the Testing part of this series.
- Optimize for clicks is usually not the way to go, as it doesn’t take conversion rate into account. Even if CTR, clicks or Quality Score is your main objective, many users feel Google chooses a winning ad before it’s really statistically significant.
- Optimize for conversions takes both conversion rate and CTR into account to pick the ads that deliver the most conversions. However, if there isn’t enough conversion data, ads will rotate using the optimize for clicks data. So this is actually a worthwhile setting to try out in campaigns with many conversions in each ad group.
- Rotate evenly (for 90 days). For most ad groups, you should be able to pick a winner within 90 days (often much sooner). In campaigns with enough clicks and conversions, this setting will provide you enough control, but prevent low performing ads to run evenly more than 90 days.
If ads in an ad group are unchanged for 90 days, the campaign will automatically begin to optimize for clicks or conversions (the latter in case the Conversion Optimizer or Enhanced CPC is enabled).
- Rotate indefinitely. This is the setting where you keep full control of the split testing process by keeping your ads delivered evenly into the auction forever. Especially in low volume campaigns, you might need more than 90 days before you can pick a winner in an ad group.
Settings that need statistical significance
Before knowing which bid adjustment to set, whether or not you should enable search partners and whether or not you should enable close variants, you need some actual data to see what works best in each campaign.
This means waiting for statistically significant performance data about mobile devices, locations, day-of-week and time-of-day for your bid adjustments. It also means trying out Search Partners and close variants and disabling them in campaigns where they significantly underperform compared to respectively Google Search and their exact and phrase match siblings.
One of the most welcomed improvements of enhanced campaigns is that we can bid differently for different locations in the same campaign. Ad scheduling principally got a facelift and devices, well, I won’t go into the lack of control on devices as many have rightfully done before, I just hope your website is very responsive. Let’s see how you could best determine each of these bid adjustments.
Location Bid Adjustments
From the moment you have a significant amount of data on the performance of a specific location compared to the campaign average (or your target), you can set a bid adjustment for this location. What metric you compare depends on your goals, it could be conversion rate, CPA, ROAS, Revenue per Click, etc. For example, if the average CPA of your campaign is $100 (and you’re happy with that) and Chicago has a CPA of $80 you could adjust the bid for Chicago to (100/80) – 1 = +25%. You’ll find this data in Geographic view of your Dimensions tab.
For further reading, these great posts will provide you all the details you need to determine this bid adjustment: How To Determine Your Mobile & Geo Bid Multipliers For Enhanced Campaigns & Smart Geographic Segmentation & Bidding With Enhanced Campaigns by Benjamin Vigneron and Geographic Targeting In An Enhanced Campaign World by Brad Geddes.
The idea is the same as for locations: get the campaign data from the Dimensions tab (Time: Day of the week and Hour of Day) and look for statistically significant differences from your average performance. To be even more precise you should combine the day-of-week and hour-of-day performance data by downloading a report and adding a segment like this:
With this report you’ll get a combined insight of the performance of each hour of day on each day of the week instead of the average performance of one of both which could lead to suboptimal bid adjustments.
And don’t forget: ad scheduling is based on the time zone of your account, not of the user!
Mobile Bid Adjustments
In case you have a website that provides a bad mobile experience, it’s probably best to set this adjustment to -100% and make it a top priority to create a responsive or mobile website in the meantime. While you’re at it, make sure your site provides a great experience on tablets as you’ll advertise on them anyway.
If your websites provides a decent or even great mobile experience, you’ll want to advertise on mobile devices. The hard part is gauging the full value of mobile, which is not just the macro conversion on the mobile site. Many advertisers say: “My CPA is twice as high on mobile devices, so my bid adjustment should be -50%”. That’s only the case if the only thing you care about are direct mobile online conversions.
Before applying this logic you may want to consider that mobile value is also generated by:
- Calls: see the previous measurement post on how to track phone calls.
- Cross-device conversions: these are now included in the recently released Estimated Total Conversions column (if Google has enough conversion data to give an estimate, currently set at 50 a day on the account level).
- Apps: see the previous measurement post on tracking app downloads and/or use App Analytics to track in-app conversions.
- In-store: also see the previous measurement post about tracking in-store conversions. Google will also include these as part of Estimated Total Conversions soon.
Not all of the above may apply to you, but the ones that do should be taken into account when setting your mobile bid adjustment. Google even created a fancy Full Value of Mobile Calculator to help you with this.
The best way would be to determine a value for all types of mobile conversions that apply to your business, add those values together (also see the micro conversions part in the goal setting part of this series) and use this to estimate the value from mobile devices. Then apply your mobile bid adjustment based on this total value. And just like with geo bid multipliers, Benjamin Vigneron’s post can help you on your way, even including an Excel file that will do calculations for you.
Search Partners, the blackest of the black boxes in AdWords. For some reason Google won’t show us performance by search partner so we can optimize based on this data. That leaves us with two options:
- Let the Conversion Optimizer do the bidding, as the Search partner site is one of the factors the Conversion Optimizer takes into account when setting bids.
- Simply keep Search Partners enabled when they meet your goals and disable them when they don’t.
The performance of search partners can vary strongly per campaign, sometimes it’s great, sometimes it’s terrible and sometimes it’s pretty close to performance on Google (in terms of CPA or ROAS). As they can easily provide an additional 10% to 20% search volume, I wouldn’t recommending opting out by default, unless you’re on a very tight budget.
There’s an easy way to find out if you should keep Search partners enabled: choose a long time range in the AdWords interface (e.g. the last 12 months) in the All online campaigns view, sort by Cost and click on the Segment button to show the segment: Network (with search partners):
In this case we compare the CPA (this could also be ROAS) on Google search with Search partners and we see Search partners have a lower CPA, so we’ll leave them enabled in these campaigns. But if Search partners are significantly performing worse than Google search (let’s say 50% worse or more), you could improve the performance of your account by disabling Search partners in those campaigns.
Also note the much lower CTR on Search partners, this is usually the case and no reason for concern as the CTR on Search partners doesn’t influence your Quality Score on Google.
Keyword matching options
Many advertisers didn’t like the liberty Google took in May 2012 by matching phrase and exact match keywords with plurals, misspellings and other close variants and therefore disable this matching option by default to keep full control.
I’d say that’s fully justified if you’re 100% certain that all your phrase and exact match keywords also have active (modified) broad variants in your account. In that case you’ll want the broad matches to catch the close variants. In other cases: data beats opinion. In many campaigns I’ve seen these close variants delivering additional conversions for an acceptable (or even good) CPA or ROAS. Besides, you can (and should) always add the unwanted queries as negatives.
If you want to know how these close variants perform on average on a campaign level, there’s a way to find out:
- Include these variants in the Keyword matching options setting. This is the default setting.
- Compare the performance of close variants with the performance of exact and phrase match per campaign with a Pivot Table.
To create this Pivot Table go to the Dimension tab of your account, select a long enough time frame and select view: Search Terms. Then only select the following Columns (Total conv. value if you’re an e-commerce advertiser tracking revenue with AdWords):
Now download this report and open it with Excel, if it gets too big you can add a filter before downloading it to only include search terms with at least 10 impressions (or any number higher than 1 to get rid of unique queries).
In Excel we need to re-label the match types to make a comparison on an actionable level. This means comparing the average performance of close variants (both phrase & exact) with the average performance of phrase & exact. To do this replace as following in the Match Type column:
- Replace all ‘exact (close variant)’ by ‘Close Variants’
- Replace all ‘phrase (close variant)’ by ‘Close Variants’
- Replace all ‘exact’ by ‘Exact & Phrase’
- Replace all ‘phrase’ by ‘Exact & Phrase’, make sure you only replace this in the cells that contain ‘phrase’, not in the cells that already contain ‘Exact & Phrase’
Now we can create a Pivot Table that will show us the aggregate performance of these match types (and broad match) per campaign.
The fields of your Pivot Table should be as following where CPA is a calculated field you created by dividing Cost by Conversions (and in case of ROAS you would divide Total conv. value by Cost):
Your Pivot Table should look something like this and now you can compare the aggregate performance of close variants with the aggregate performance of exact & phrase:
In the example above you can see close variants perform worse in Campaign A, but perform better in Campaign B. So we could disable them in Campaign A and keep including them in Campaign B.
Campaign Settings & Bid Adjustments: Your Audit Checklist
Are the search and display network always targeted in separate campaigns?
Are all search campaigns set to include ‘All features’?
Are all campaigns targeting the right locations?
Are all campaigns targeting the right languages?
Does the delivery setting make sense for each campaign?
Is the ad delivery set to rotate by default? You can try optimize for conversions in campaigns with many conversions.
Are location bid adjustments set based on significant performance differences?
Is ad scheduling set based on significant performance differences?
Are mobile bid adjustments set based on the full value of mobile?
Have Search partners been tried in every search campaign and have they been disabled where they significantly underperform?
Have close variants been tried in every search campaign and have they been disabled where they significantly underperform?
Opinions expressed in the article are those of the guest author and not necessarily bgTheory. If you would like to write for Certified Knowledge, please let us know.
Can you elaborate on why it’s probably best not to target multiple countries from a single campaign? I was considering setting up a Worldwide campaign to target a specific service and look at the data after it had run for a while to see if it merited being split out into different country-specific campaigns. How does one best handle international targeting?
it’s perfectly fine to start out that way. Indeed, keep a good eye on the geographic performance in the dimensions tab. If you see significant differences in performance for different countries, you can use bid adjustments for locations to reflect those differences.
The advantage of splitting it out in country specific campaigns is that you can write different ads that may appeal more to the local users. You also get a better view of the performance per country (this is mostly a reporting advantage). The disadvantage of course, is that it’s more work, so it has to be worth it.
But as I said, it’s fine to start with a worldwide campaign to gather data. Just make sure your keywords don’t mean different things in different countries 😉
Brilliant, thanks for your reply!
Regarding Search Partner performance, apparently your Google rep can access a report comparing Google and search partner performance. I was going through an account with mine and he offered to see how partner performance was going. I heard in his tone that he was hoping Google was doing better. He said, “Oh, you’re actually doing better on the partner network, so you should leave that turned on.” Not sure how he determined “better”, but I was happy to move on to other areas.