Google Ads continues to be one of the most important platforms for businesses of all sizes and scales. But with constant updates, increasing automation and the loss of many “traditional” control levers, finding the right balance between investment and returns can be incredibly difficult.
At SMX Advanced, Sam Tomlinson, executive vice president at marketing communications agency Warschawski, explained that marketers need to figure out how to work alongside AI and help them do better “because fundamentally we are not going to outtrade a machine.”
Here are 10 advanced strategies that have helped Tomlinson’s company achieve huge results with Google Ads campaigns.
1. Data is your optimization lever – so help the machine learn
It’s vitally important to feed machines the best data possible. Four useful tools to ensure you are giving Google the best data possible:
Conversion Action Set
This great, underutilized option allows you to “group conversions for a similar stage in funnel” and “allow you to have different goals…so that you’re actually optimizing the relevant campaign towards a relevant set of objective without other data coming in and confounding the mix,” Tomlinson said.
Tomlinson revealed his company uses enhanced conversion for every client because it allows you to link any CRM directly with Google Ads and set primary optimization events. Tomlinson said:
- “This means you’re actually giving machines data that they need to make smarter bids and smarter decisions, so you’re getting fewer and fewer junk leads. Honestly, one of the biggest ways to avoid junk leads is to just use enhanced conversions…We saw a 75% reduction in junk leads within 45 days of implementing an MQL-based primary conversion action. That’s fantastic!”
Smart Business Data
Marketers can import business data into Google Analytics and, by extension, into Google Ads.
- “I highly recommend you do this. The more data you give machines, the better they can optimize,” said Tomlinson.
Offline Conversion Import
Enhanced conversion is the first step in this process, followed by conversion adjustment. This allows you to restate the conversion value so you can adjust it up or down.
“You can remove data from the optimization algorithm so that you’re not optimizing flawed data,” Tomlinson said, adding:
- “For example, if somebody makes two purchases by mistake, you can retract one of them…Remember that every piece of data you give Google, unless you tell it otherwise, Google will assume it’s incremental. So if there is a discrepancy between what you’ve initially told Google and what actually happened, use conversion adjustments to change that.”
2. Use data to focus engines on what matters to you
This is a powerful approach for both B2B and B2C businesses because it lets marketers make smaller adjustments to how algorithms bid, based on their business’ unique data – which is something that they can control.
- “Every business has different value structures. So being able to one, understand what those are, and number two, translate them into value rules…enables you to get so much more from your Google Ads campaign.
- “I’ll review probably between 200 and 300 Google Ads accounts per year and I would say less than 10% are using value rules you see. They’re so powerful and they’re so under-utilized,” Tomlinson said
3. Layer audience & business insights on top of a data-centric foundation
Give Google the data it needs to optimize your content better and faster – ultimately, this will get marketers better results from their campaigns.
The best way to do this is by layering your category information about how customers buy your products, such as product type, price, buyer type and so on.
- “What matters is that you are combining the data about your business with data about your audience to give the machine a structure that it can operate within. When you layer information about how customers are buying your products with how you make money on your products, you end up with a matrix – prioritize the matrix of what you should bid on,” he said.
4. Use CPA or ROAS as your campaign’s steering wheel
Too many advertisers are using automation and automated bidding strategies in the wrong way, according to Tomlinson. Commenting on the campaigns his company has achieved the most success with, he pointed out that a low CPA target or high ROAS targets are a lower priority for Google Ads.
- “Your highest priority campaigns should get the highest target CPA or the lowest target ROAS, because that reduces the threshold for them to serve to the very least that you can tolerate based on your business data and your business’s unique cost structures and value structure. “The only way this structure actually works profitably is if you know your business’s numbers and you’re confident and can trust them.”
- “When you start thinking about how to set these CPAs or ROAS targets, Google’s Budget Simulator is actually great. I love it. It’s a really fantastic tool…But remember, anything beyond that optimal spending starts getting into inefficiency and loss. You start to give back the gains that you’ve got. So getting the right target CPA is critical. Use that budget simulator to understand where that is, the second part of that.”
Get the daily newsletter search marketers rely on.
5. Don’t let runaway CPCs blow up your ad account
Get the most out of your campaigns by preventing runaway CPCs, which can blow up your ad account – and not in a good way.
You can do this by setting a minimum and maximum bid for any automated strategy within your portfolio.
- “It’s free insurance because sometimes, Google does get it wrong – and this creates downside protection. There is a possibility that you could have an outlier conversion rate or an absolute fantastic runaway success, but how much are you going to shape that risk? Just eliminate the bad hands and you’ll be fine.
6. Use a modern structure to minimize time learning
The more you split your data, the harder it is for machines to learn from it. So marketers need to focus on minimizing fragmentation and segmentation to the points where it is necessary, and provide an actual marginal return that justifies it, according to Tomlinson.
- “That is a hard pill to swallow for a lot of old school SEM people, but it is one you absolutely must swallow. And start consolidating.”
Tomlinson advised consolidating 20 single keyword ad groups into a single ad group. He then recommended combining audience targeting and data targeting with keyword targeting.
- “This will create a combination of passive targeting and active targeting. That’s what gets you a better result.”
PPC experts were also urged to stop competing with AI and instead take on more strategic tasks that will help machines. For example, by writing good ad copy, you can help machines to get a better understanding of your content, explained Tomlinson.
- “Creating unique, highly relevant and highly informative landing pages is important as well. It is doubly important now that Broad Match looks at landing page content as a signal for what types of users and what types of searches might be relevant.”
- “Finally, make automation your friend. Negative goals are a great way to do this. If a campaign performance deviates from the expected by a significant amount, turn the campaign off. Something’s not working the way it’s supposed to, so before it gets worse, stop the bleeding so that you can then go in and adjust; restructure, and move keywords and products. Teach the machine what it needs to look for so that it can do its job better.”
7. Always add audiences… even if it is just as ‘observe’
Marketers must remember that they are targeting people rather than keywords, and so it’s important that they use audience tools more effectively. One way to do this is to go to GA4 and use predictive audiences.
- “This is an incredible feature,” said Tomlinson. “Brands have historically spent billions of dollars trying to figure this out – and here it is for free!”
GA4 can tell you what audiences are likely to engage with your marketing and which people are unlikely to return to your website in the next seven days.
Marketers can also create custom segments based on what specific audiences search for in Google. Alternatively, they can identify which audience browse certain types of websites or use certain types of apps.
- “It’s incredible,” added Tomlinson. “You can look research people who are searching for solutions, people who are searching problems that we solve, people who use alternatives to our current service – you’re limited only by your imagination.”
8. What you exclude is more important than what you include
As machines get increasingly broad, as match types continue to change, your exclusions become so, so important.
- “So please, please, please think more about what you can cut out,” said Tomlinson. “Can I cut out the absolute garbage versus can I include everything I need to? The machines will find the stuff that you need to include. Your job is to seep out the stuff that needs to stay out. “
9. Test RSAs using variants
Variant testing is the best way to test RSAs, according to Tomlinson.
- “What I recommend doing is one of three tests that look at composition,” he explained. “Say I want to test brand differentiator call to action versus proof point. I’m going to test them against each other and for each of those structures I’m going to use pinning and multiple headlines. The pin ensures that the headlines in each position match the structure I am testing, and the multiple headlines gives me some flexibility so that Google doesn’t completely hate me.
- “You can test your offers and your value props or a seed and iterate where you just put in a bunch of headlines, see what comes out, see which combinations work best, and then use that as an inspiration to do testing. Any of those approaches is right and fine.”
- “Variant testing is something that is so rarely done that it hurts. It genuinely hurts to go into an ad account and see one or two RSAs with no variant testing because candidly it just doesn’t work right.”
10. Ensure you’re doing multi ad group testing
It is essential for advertisers to have RSA testing structures – especially those working on smaller account. The best way to do this is to use variants and look for patterns across your headlines and descriptions.
“You’re basically testing messaging across your entire account versus just one ad group,” explained Tomlinson.
- “That gives you statistical significance faster and allows you to take learnings and apply them across your account. Multi ad group testing is amazing – but it’s just not done, which is really unfortunate because what you really want is to understand which assets and which combinations of assets lead to the best results.”
Remember labels for brownie points!
Tomlinson urged PPC marketers to make better use of labels for monitoring purposes when changes are implemented.
- “Most of us don’t remember what happened last week, let alone remember why you made a certain change seven months ago.
- “Whenever you make changes, whenever you add stuff or do stuff, always label your changes and use a standard set of labels so that you can actually group them, pull the data and understand what’s happening.”