Reviewing your current Google Ads set up and best practices is a healthy habit to get into.
What may have worked for you two years ago may no longer generate the leads you need in a more competitive market, or you may find yourself not utilising the full range of avenues open to you within Google.
This is a process that we have been through as an agency recently, and our thoughts are shared below.
These best practices are based from auditing and running Google Ads accounts that spend from £2k to £600k a month, across verticals including travel, retail, software, education, and healthcare.
Search Campaign Types
- Generic (incl. DSA)
If you have competitors bidding on your brand, or you do not have a top of page SEO rank you should be bidding on your Brand terms. If both of these are true, you absolutely should be bidding on Brand.
A slightly less legitimate, but common, reason for running Brand activity is to increase the overall profitability of your Search channel.
We usually see this when a company knows they need to secure more marketing budget in order to fully invest in upper funnel, generic, campaigns. However the business case for that often relies on Search as a whole being profitable.
If you’re not sure whether running Brand is worth it, you can always run a paid/organic test. For example, an AB script can pause your Brand activity for every other hour over a two week period.
Then, using Google Analytics data, you ought to be able to compare how paid and organic revenue as a whole was affected year on year, or period on period.
Generic (aka Non-Brand)
There isn’t much to say about this – if you have a Google Ads account, presumably you are looking to expand your customer base. If you are looking to expand your customer base, you’ll do that through generic campaigns.
We have come to the conclusion that, however old school it seems, it is a good idea to have a Generic DSA campaign alongside your regular Generic campaigns.
We’ve found DSA campaigns to be particularly good at serving ads for searches that are otherwise deemed too low search volume if set as keywords.
As with all accounts, the most specific (relevant) searches, are the most valuable, but if you happen to sell obscure products or products with very low search volume, it can be hard to capture this intent.
Take an industrial machine parts manufacturer for example. They may have hundreds of thousands of distinct products that are differentiated by necessarily obscure model names and numbers.
In this case, each individual part’s name will be low search volume, but that doesn’t mean that demand isn’t there.
Instead, we can ensure we are able to serve ads by creating DSAs.
The usual best practices for these still apply:
- Use a page feed if possible to create and maintain your target placements
- Use negative placements to avoid paying for traffic that your organic results or Google My Business ought to pick up (contact numbers for example)
- Add your Brand as a negative keyword
- Add the keywords from your non-DSA campaigns as negative keywords to avoid cross contamination
Specific (aka Product)
This campaign type is a halfway house between Generic and Brand.
You would use it, for example, to contain keywords for the brands of products that you sell, particularly if you are a sole supplier, and/or you have an agreement with the manufacturer that you can bid on them.
NB If you are also a manufacturer of that product, those keywords ought to go in the Brand campaign.
Another use for this type of campaign would be if you sell a product with a specific use, but could feasibly be searched for with more generic searches also.
For example, if you sell ethical clothing, you may decide to have specific campaigns which contain long tail keywords to do with ethical and sustainable fashion and products – i.e. organic cotton skirt, fair trade clothing, sustainable swimwear.
Those can then sit alongside more generic campaigns that contain shorter tail, less specific keywords – for example, cotton skirts, clothing, swimwear.
Note that you will need to ensure your longer tail, specific, keywords are negatived from the more generic campaign, and visa versa.
This structure allows you to differentiate between different levels of intent for reporting and budgeting purposes.
A way to expand into more generic campaigns from specific campaigns also is to create RLSA generic campaigns.
That is, launch campaigns containing vaguer keywords, but have them targeting audiences who have engaged with your site previously, thereby prequalifying the traffic more than pure generic.
Generally speaking bidding on your competitors’ brand terms is a good idea if any of the below are true:
- A competitor is bidding on your Brand
- You have a good amount of budget
- Your aim is to gain or maintain market share
- You’ve maximised viable volume across Brand and Generic campaigns
- Exact and BMM version of each campaign, with a 70:30 split of budget
- Script to negate close match variant matching, but carefully monitored
Exact & BMM are the two match types you want to be using.
It still stands that these two match types are the best options. We consistently see them perform best in terms of CPA across accounts and verticals.
We would recommend avoiding phrase match, as it tends to inflate CPCs, and rarely produces anything that can’t be replicated by using BMM.
For example, if the keyword word order is of particular importance to you, you can always add phrase match negatives to a campaign, negating the word order you are trying to avoid.
Towards the end of last year, Google made Exact match less exact, meaning that close variants can also match into your keywords. To ensure that you are appearing for the searches that you believe you are, we’d recommend using a script that ads close match variants as negatives.
You can find the script at number 45 on this list. We use this as best practice across our accounts because analysis we have run on close variants demonstrates consistently that they perform worse than their match type equivalents.
This means you can keep your ads as relevant as possible. We’d also recommend that you keep reviewing the keywords being added as negatives to see if any of them are worth adding as keywords.
If you are struggling with low search volume, in the first instance we would suggest running a DSA campaign rather than making your keywords more broad.
However if that doesn’t help, you can try changing a selection of your keywords to broad. Bear in mind these should be labelled and monitored very carefully. Also expect lower CTRs and higher CPCs.
NB only do this for searches which you can see competitors are appearing on but you are not, due to low search volume.
Campaigns should be split by match type – i.e. you should have one Exact match version of each campaign, and one BMM version
This is because the match types behave differently to each other, and so you want to be able to allocate budget and report on them accordingly.
For example, your BMM campaigns will be getting higher volume with a lower CTR and higher CPCs. That’s due to capturing more searches, and your ads being less relevant to each search.
Their primary purpose is to capture longer tail searches, that you will then add as Exact match keywords in your Exact campaigns.
Adding Exact match keywords ensures that you can have headlines and descriptions in your ads that exactly match the search being made. That means you’ll have a better quality score (through improved ad relevance and CTR) – and ultimately pay lower CPCs.
Because Exact match campaigns better capture intent, they ought to account for 70% of spend, compared to 30% on BMM (including DSAs).
- Single keyword ad groups
- AB test 2 ETAs per ad group
- Initially 1 RSA per ad group with as many headlines as possible
Always use single keyword ad groups
Having one keyword per ad group means you can tailor every ad to every keyword and definitely remains best practice.
That said, if you currently have 2-3 keywords per ad group and these tend to be plurals or misspellings, you’ll see a limited benefit from this.
The same goes for Brand campaigns – because ad relevance and CTR ought to be really high for Brand campaigns regardless.
However, if you have 3+ keywords per ad group, particularly in Non-Brand campaigns, it’s definitely worth addressing.
Run at least 2 ETAs per ad group
You should always be running ad tests. Improving your Google Ads performance is an iterative process; in order to continually improve you need to be consistently testing to statistical significance.
This holds true for ads, whether that’s headlines, descriptions, or landing pages. The best practice is to have at least 2 ETAs per ad group.
You can have more than 2 if you have more than one meaningful variation you are testing, and if you have enough traffic to reach a significant result within a reasonable timeframe.
I’ve written a guide on AB testing within Google Ads, but to summarise:
- Always AB test using an AB script
- Set your campaign settings to rotate ads evenly
- Vary one element only
Avoid before and after testing and avoid testing more than one element at once, as neither will provide you with a rigorous experiment. AB scripts will help you split traffic more evenly than Google’s inbuilt tools.
Also add at least 1 RSA per ad group, with as many headlines and descriptions as possible
ETAs are great for being more scientifically split tested, and working out which messages are impactful.
RSAs on the other hand allow you to show on inventory that you otherwise may not be eligible for with just ETAs. That’s thought to be because Google’s machine learning algorithms determine you to be relevant to more searches when you have a wider variety of headlines and descriptions.
You do need to relinquish some control over testing and traffic splitting for RSAs, but you can use ETAs for your ground for more scientific testing.
- Use query sculpting to drive traffic to the relevant campaigns but monitor closely as it isn’t a perfect system
- Build single product ad groups
- Trial smart shopping once you have the same number of conversions you’d be happy running a bid strategy on
Use query sculpting to create Brand and Generic campaigns, as you would in Search
Query sculpting is an intimidating phrase, but is a simple premise. If you have multiple Shopping campaigns with all of your products in, searches get funnelled to different campaigns depending on their priority setting.
The settings dictated by Google are simply high, medium, and low, with traffic initially going to the high priority campaigns.
If you want to be able to set campaigns up to reflect different search intent, and so bid accordingly, you can do this adding negatives to your high priority campaign for searches you do not want to be picked up there.
That traffic will then be sent to your next highest priority campaign (let’s say medium), unless the terms are also negatived there.
Finally they will come to the lowest priority campaign
Query sculpting isn’t a flawless system – if a higher priority campaign is limited by budget, queries usually collected there will be forwarded onto another campaign.
So for example if you have the below set up:
- Generic campaign – high priority, Brand terms as negatives
- Brand campaign – low priority, no negatives
And your Generic campaign is limited by budget, the generic searches will be default be sent to the Brand campaign.
Also the idea that priority settings alone dictate the order in which traffic is funnelled should come with a caveat.
We have seen that when low priority campaigns have particularly high bids in comparison to higher priority campaigns, traffic can be rerouted to these first, even if the higher priority campaigns are not limited by budget.
That is, potentially bids supersede priority settings in some circumstances.
Build out single product ad groups
Having single product ad groups in Shopping afford you an extra level of control over bidding and negatives.
We want to be able to bid at as granular a level as possible, and grouping products together is not a good way to achieve this.
You’ll regularly find that it is actually a handful of items that are spending the majority of your money within a grouping, but very different items that are converting.
However, if you have categorised them together, you are applying a blanket bid across the lot, which clearly doesn’t make sense.
What you could do, is to put all of your products into one ad group, and then split that ad group by item ID to get single product product groups.
This provides the opportunity to bid at a product level, but it limits you in terms of being able to add negatives at a product level (as these are added to ad groups).
Moreover, if you want to use custom labels in order to boost bids across products that you are trying to promote for example, or those that make you a higher margin, you can’t achieve that via single product product groups.
Instead what you can do is make single product ad groups. The titles of these should be the item contained within them, and concatenated with the values from any custom labels that are relevant for you.
That way, you can bid and set negatives at a product level, but you can also adjust your bids based on your custom labels by filtering for them in the interface.
There are a variety of third party platforms that can help you with that, or it can be achieved via building a spreadsheet template.
Test Smart Shopping when you have decent conversion volume
Initially we were a bit reticent to use Smart Shopping. All you can do really to control it is change budgets, and as a performance marketing agency there is an instinctive mistrust of anything that puts so much power in the hands of Google.
However, we have seen great results for a number of retail clients from using it – as long as they have a steady volume coming through their Shopping campaigns in the first place.
We have also seen it blow through low volume accounts’ budgets and just not convert, whereas traditional shopping has been fine.
- Organise your campaigns by targeting type
- Exclude apps
- Use responsive display ads but don’t forget to update your creatives
Design your campaign structure by targeting type
And make sure you are using all the targeting types available to you (audiences, placements, keywords etc.).
Within each campaign, each ad group should be targeting one thing only. For example, in your Audience campaign, if it were remarketing, you might have the following ad groups:
- All visitors L7D
- All visitors L8-15D
- Cart abandoners L7D
As with all granular and clear structures, this allows you to see precisely which targeting options are performing the best.
It also means you can bid depending on intent initially – for example, the most recent visitors to your site may be the most valuable, as may be the ones who got the furthest through your check out process.
Exclude app traffic unless you have a specific reason not to
This seems like quite a strong statement to make off of the bat for an account, but every GDN campaign we have ever seen has had much higher CTRs from app placements and lower CvRs.
That is compounded when you segment by age and look at users of 65+.
The theory here is that it is very easy to accidentally click through on an app when trying to close an add or just generally whilst using your mobile – a theory corroborated by low CvRs.
Higher CTRs also eat into overall GDN budgets, and therefore bring down performance and potential volume across the channel.
Of course test this if you’re not comfortable excluding them initially. If you are however, then the most efficient way to do so is via Editor – select “All apps” from the negative mobile app categories menu.
Use responsive display ads
There’s not too much to say on this – we’ve seen responsive display ads consistently outperform standard display ads, so it’s no surprise that they are becoming the default ad format for GDN.
As with RSAs, to ensure that you get the most out of Google’s machine learning, make sure to include as many creatives as possible.
Moreover, don’t forget to refresh them. Unlike in Facebook where you can easily see a handy frequency metric, it is easy to forget that with GDN searches can also experience ad fatigue.
That is especially true for your audience based campaigns, or when your targeting is quite narrow.
Do you have any new best practices which you want to share, or have anything to add to the above? Get in touch!