How marketers can adjust to redefined keyword matching
Keyword matching has changed, but not all campaigns are ready to use them effectively. At SMX Next, Aaron Levy shows how marketers can leverage modern match types.
“I think we’ve gotten away from marketing over the years, and a lot of what is happening with match types empowers that,” said Aaron Levy, head of paid search at Tinuiti, in his presentation at SMX Next. “This allows us to grow our campaigns and businesses by looking at a bigger picture instead of looking at the words that people use.”
According to Levy, keywords are no longer the most relevant piece of paid search campaigns. Rather than remaining the linchpin, they are destined to play a much more tangential role in the future of marketing, especially when it comes to exact match strategies.
“I know that they’re [keywords] still in Google Ads. I know that they’re still in Bing Ads,” he said. “But fundamentally they’re gone. There is no longer an area where we’re going to pick the exact word that someone is going to use.”
“We’ve been predicting this for years. We’ve expected them to go away,” he added.
Despite knowing this change was inevitable, marketers may still be scrambling to craft new ways to target customers based on a more broad understanding of queries. Fortunately, Levy laid out a roadmap to help marketers adjust their keyword matching strategies.
Transitioning to new keyword match types
Successful paid search marketing used to be defined largely by precision — the ability of marketers to exercise control over their campaigns via exact and phrase keyword matching. And while broad match was available, it was never as big of a focus.
“I used single keyword campaigns,” Levy said. “I wanted to have everything in the same place. I wanted to maximize segmentation. I wanted to control every single solitary thing in every message that every person saw.”
“To me, phrase match was a waste of time. It didn’t make any sense to me, so we only used exact and broad match modified keywords,” he added.
Levy noted that this focus on exact match keywords worked well back then; it was a strategy many marketers used to get their ads in front of people more efficiently. However, as Google’s algorithms change, marketing strategies need to shift as well.
“We need to evolve,” he said. “We need to move past this language-only way of doing search.”
Focusing too much on the exact language of queries prevents marketers from seeing the full context of their audiences. To address this issue, Levy suggests marketers consider thinking of their campaigns in terms of passive (broad) and active (exact) targeting, placing more emphasis on the former: “Active is when you are directly trying to make something happen for a given word for a given person. Passive, or broad match, is when you set a theme, let the ‘Google Roomba’ drive around and bump into some walls and make some decisions. Then you see what happens.”
There will always be situations where marketers must manually adjust their paid campaigns. But the key point is that they no longer have to control everything about their matching, and the sooner they let go of that control, the better.
Audience segmentation using modern match types
Segmentation has always been a key ingredient of successful paid search campaigns. These practices help marketers break down their audiences using relevant data such as buying intent or demographics. But changes in digital advertising — especially the move to responsive search ads — have changed the segmentation process.
“Segmentation in modern search with modern match types is a little different,” said Levy. “I’m not trying to have a perfect one-to-one message match because we have responsive search ads. And I know that all of us are mourning the death of expanded text ads, but the fact is all ads are responsive now.”
Modern match types allow marketers to segment their audiences based on actionable data instead of wasting time trying to achieve maximum segmentation. Leveraging broad matching helps marketers further optimize RSAs by minimally segmenting audiences using performance, goals, budgets and other relevant data.
“At this point in time we’re seeing that data value is more effective than it is having perfect control and one-to-one message match,” said Levy. ”We want to segment campaigns and ad groups based on what performance dictates, when budgets change, when goals differ or when messaging needs to be different.”
How AI affects phrase and broad match types
Last year, Google announced that phrase and broad keyword match types are preferred when they’re identical to queries. This is no doubt due to its work with AI, specifically BERT, which has helped the search engine better interpret language and intent.
However, these AI systems are not at the keyword-matching level many marketers expect them to be.
“What we all want, and maybe what search engines are selling, is artificial intelligence with matching behavior that will replicate human thought,” said Levy. “It will mimic the way that we work. It will understand the things that we understand.”
Unfortunately, these systems fall short of this goal.
Yet even though Google’s AI systems fail to understand and apply human behavior to matching, marketers can still use them to improve ad campaigns. They just need to think about them differently.
“A better way to think of match types, and a better way to predict their behavior, is to think of them as machine learning,” said Levy. “They don’t think; they learn. They understand performance. They understand statistical models. They figure out what happened and they learn and grow from it. We give it a conversion goal or a keyword target and it seeks to accomplish that goal as many times as possible, which means it’s going to have some swings and misses.”
He added, “So, it’s a performance-based decision engine and it’s looking for relevance. And there’s a way to predict it.”
Levy asks marketers to think of search algorithms as technologies in need of coaching, rather than expecting them to understand searchers as a person would. Marketers can help these systems match more effectively by:
- Using keywords tools and search listening tools to forecast matching.
- Limiting language targeting to avoid query confusion.
- Negating synonyms, slang and any other language that doesn’t work for humans.
“For anything that you’ve learned doesn’t work, set up a guardrail and protect [the campaign] against the AI,” Levy said. “That will help mitigate some of these bad matches, freeing you up to use match types for what they’re meant to do.”
Marketers that can deploy modern matching in a responsive ad environment will have the greatest chance of succeeding in our transformed paid search landscape.
Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.