Over the past few years, match types have changed, smart bidding has become a little more intelligent and new campaign types have been introduced. These changes once again have advertisers reconsidering their account structures.
How are search marketers evaluating account structure and organizing campaigns in light of all these changes – and why?
We found out during a roundtable discussion at SMX Advanced, featuring myself, Duane Brown from Take Some Risk and Aaron Levy from Tinuiti. Brad Geddes from Adalysis moderated.
The session covered many of the changes the major search engines have made recently, including:
This was a lively session with lots of great information, so buckle up!
How has campaign organization changed for you?
The prevailing thought on campaign organization was that the hyper-targeting of the past is no longer necessary. I have moved from granular campaigns segmented by match type to incorporating all match types in the same campaign, along with more automation and smart bidding.
Levy said he used to use single keyword campaigns segmented by audience. Now, they’re focused more on input optimization that’s more performance driven. Giving the right signals to the machine is key.
Automation was another theme here.
Brown is a fan of Performance Max campaigns for his ecommerce clients. However, he cautioned that Google is still not great at context, so figuring out when to consolidate and when to keep things separate is important.
What are the big things you look for when planning campaign structure?
The key here is intent. All three speakers mentioned understanding search intent as the basis for campaign structure.
Levy suggested that if performance is expected to be different enough to require separate targeting and messaging, split it into its own campaign. However, his default is to let the machines make better decisions faster.
I think about campaign structure in terms of intent, goals, and budget. Where any of these vary, make a separate campaign. Budgets may be client-driven by region or even content asset. But aside from that, don’t split out unless it’s necessary.
When asked about brand search, all three speakers agreed that brand should usually be broken out separately, to maintain brand protection and its place in the buyer journey, which is usually bottom of the funnel.
What do you no longer segment on that you used to?
I no longer segment by match types. This used to be done at the campaign level but is completely unnecessary now.
It’s also not necessary to use every match type. Pick the best match type for each keyword instead.
Levy added that for ecommerce, they used to segment by product type. They still do this when performance dictates, but it’s not something they always need to do.
How has ad group structure changed?
Fewer ad groups and no more breaking out by match type was mentioned by all panelists.
While broad match performs better now than it did a year ago, Brown noted that he does break out this match type for testing.
Levy is a fan of consolidation. They used to structure ad groups by match type, but now incorporate all match types in the same ad group.
What about RSAs? Some people are using one headline per product in the same RSA and assuming Google is matching them up with the right product. Are you still separating RSAs by theme?
I still segment RSAs by product, with each ad specific to the content and theme and using variations of headlines and descriptions.
Incorporating multiple products into a single ad is better as a DSA. DSAs are used very little in B2B and have to be carefully set up to be successful.
Levy added that they still split out DSAs. DSAs have improved but still make questionable calls, so they use them primarily as a research tool.
Levy is also writing separate headlines by theme for his RSAs. Be careful of including the keyword over and over in multiple headlines – make sure the combinations make sense.
Do keywords still matter?
Brown said yes, but maybe not as much as they did 5 years ago. Keywords still have value, though.
Brown uses DSAs as a “sweeper campaign” to find unusual keywords they wouldn’t think of otherwise.
Levy addressed the common theme of “keywords are dead” by saying that keywords died and came back as something else.
Google has a branding problem with keywords and match types. Neither are what they used to be, and the names do not mean what they used to.
Keywords are now thematic, audience-based or language-based. But they’re still important.
Do you trust that Google will show your ads for all the possible search terms?
Sometimes Google is surprisingly good at matching keywords to search terms. Google seems to have figured out that B2B advertisers are looking for B2B customers.
But I still see a lot of bad matches, where Google doesn’t seem to understand what the advertiser is selling. Advertisers should be cautious not to put all their trust in Google.
We still need keywords and negative keywords. As I mentioned during the roundtable discussion, Google hides about 75% of search queries in many of her B2B accounts.
In the end, it comes down to signals. The more signals you can give Google with your keywords, audiences, and goals, the better results you’ll get.
Levy added that we now have a combination of image and language search. When someone is searching for an image, what do you bid on?
He believes intent will take a larger role in the future, since intent is a mix of people and language. For B2B – who is searching? An administrative assistant, or the decision-maker?
Advertisers need to go beyond just keywords. They need first-party and third-party data. Google makes decisions on what they can see, we need to feed them as much as we can.
It’s possible to have the same search term showing in multiple ad groups. Should you include every single search term as a keyword? Does Google really match them up exactly? Are you adding all the queries or letting it ride?
Brown said that if there are enough conversions, he will add the search term and adjust bids. But if query volume is low, it doesn’t make sense to add the keyword because it won’t move the needle.
Also, remember that as a PPC manager, you have to manage your campaigns. Will adding the search term help, or just be more work?
I usually add the search terms if they’re converting. I also still try to do some keyword sculpting, although that has become less important.
Also, a term may convert, but on a keyword that’s not as relevant as another keyword elsewhere in the campaign. It’s important to think about the best place to add the search term – don’t just click “add” in the search term report.
Regarding automation: It’s generally believed that performance max and discovery campaigns are the least liked campaign types. How are we thinking about these campaign types? Are they core for you, or backfill, or are you just not using them at all?
Levy quipped that in golf, a “Max” score is for the bad players! He feels that while performance max may take a bigger role in the future, it will not be the only option.
Levy is cautiously optimistic about performance max. It’s performing well but is under-baked. It’s not a core campaign for him as there are too many unknowns and not enough targeting, and it makes bad decisions, especially in B2B.
Brown is bullish on performance max. He thinks that in a year, people will say they love it. After all, people love things when they work. The black box is what people don’t like.
Google needs to test things in the real world to find problems, and agencies need to learn use cases for each new offering.
Regarding discovery campaigns, Brown loves them. The right audience is necessary. Discovery campaigns can convert for prospecting, but you need to change creative faster, like you would with Facebook as opposed to Google.
What’s your advice to someone with 3-5 years’ experience in PPC?
Brown recommended forgetting everything you knew 5 years ago, because performance max has changed the game. Think about what you want to show for and what you want to match for. Our job isn’t to push buttons but to understand what’s making more money.
Levy reminded us that those with 3-5 years’ experience are still learning, so it might actually be easier for them. Those who are long time PPC pros have established habits that are hard to break. Learn new resources, approach with skepticism but devote time to learn how the sausage is made. Influence decisions and adjust accordingly.
My suggestion: managers should think about the signals they’re giving to the machine and let the machine do the manual work.
Search is not a list of tasks to check off. Dig in and see what’s working and why.
Think more strategically: “what are my objectives and what do I need to do to make it better?”
Bonus Round! Highlights from the Q&A:
What about industries with compliance concerns?
I have a lot of B2B clients with compliance issues. Sometimes you have to do old-school segmenting across match types with lots of negatives.
Align things as best you can while still explaining to your client that exact match isn’t like it used to be.
In heavily regulated industries sometimes broad match is out of the question. Often, you really have to go old-school and avoid many of the new campaigns and automations as much as possible in order to maintain compliance.
How do you handle match type breakouts with brand search?
Levy breaks things out based on definitions that are the same across similar brand terms. With brand, he is doing more old-school segmenting and more negatives, because brand can bleed into non-brand and vice versa.
How do you go about increasing reach and volume while maintaining relevancy?
Brown believes in adding more keywords rather than using broad match, and in adding audiences. That said, broad is worth testing. Also, retry things you tested before – they might work now!
Are you even doing query mapping anymore?
Levy is letting Google figure it out unless messaging is critical. Usually, Google does a good job with this.
Geddes added that his high-spend clients still do query mapping, and see a huge conversion increase from doing so.
There are struggles with RSAs due to lack of data. How do you think about structure in an RSA world?
- Levy: “People try to clone ETAs to RSAs. Make sure each ad element says something new or else it becomes repetitive.”
- Geddes: “Pinning is really flexible. You can pin multiple headlines to a single line.”
Google has been pushing using RSAs and broad match. Is this useful?
Levy has done this a lot. They are starting to see some cannibalization, but 90% of the time Google picks the right match – IF you are using smart bidding.
Do not do this with manual bidding. There are too many variables.
What about ecommerce, non-Performance Max campaigns? Any quick structure tips?
Brown suggested using analytics to break out top products and low-performing SKUs. You’ll need multiple campaigns.
Performance Max vs. shopping – are you getting worse ROAS with performance max?
Brown has done a lot of testing with Performance Max. ROAS is equal to or better than smart shopping, but you need to use the right creative and ad formats.
Is Google performing better or the same now with broad match?
Broad match, which used to be a B2B nightmare, is performing way better. If the keywords are long-tail enough, broad match does well.
Lightning round! Where are we going with match types?
- Me: “A lot less visibility into performance, but there’s lots to test.”
- Brown: “Give up control and hope we get insights.”
- Levy: “Make it ok for the machine – put guardrails in.”
- Geddes: “Override the machine if needed.”
Watch the roundtable discussion at SMX Advanced 2022
To listen to the roundtable discussion yourself, get your free pass here.
Already registered for SMX Advanced? Log in.
Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.
New on Search Engine Land