In the ever-evolving landscape of Google Ads and paid search marketing, understanding the nuances of keyword matching is crucial for maximizing campaign performance. This guide delves further into what we can learn from the results of a comprehensive automation test aimed at optimizing broad match strategies.
To paraphrase Arthur C. Clarke, advancements in technology can seem like magic at first. In the case of Google, broad match was introduced with promises of smart campaigns that would be able to do the work of finding new customers for us. But as with any new technology, it takes time and testing in order to fully understand the scope of its benefits, and more often than not, its drawbacks.
Our research to this point can shoulder some of the pressures we face as marketers to predict the future. Results will differ from campaign to campaign, but through time and concrete testing structure, we’ve learned a number of things about broad match and how it works. We’ve learned what it will do when left alone, and what kind of guidance and support broad match requires to produce the best results possible.
This post outlines the key findings from our journey while exploring the hierarchical dynamics of keyword types and the importance of maintaining a balanced keyword strategy. Whether you’re an experienced digital marketer who’s been around for decades or just starting with Google Ads, understanding these principles will empower you to leverage broad match keywords more effectively.
First, a Bit of Backstory
At Eight Oh Two, we strive to stay on the cutting edge of new tools and tech in the world of digital marketing. Over the past few months, we’ve created a unique testing strategy for incorporating new technologies and machine learning advancements into our campaigns. Namely, we began to use a new structure to make sure our broad match additions to campaigns did not inadvertently overshadow our existing best practices.
After six months of testing, the data confirmed that our new strategy was superior to the “instructions” that came with broad match from Google, which recommend mixing broad match into campaigns alongside phrase match and exact match keywords. By using our strategy instead of the suggested approach, we were able to reduce cannibalization of best practices keywords down to less than 1% of spend in as quickly as five weeks of delivery.
This blog provides a comprehensive guide to not only match types as they apply to keywords and search terms. We’ll also cover the ways they interact within a campaign built to perform to its maximum potential. To recap one of the defining elements of how search terms and keyword match types interact with each other, we’ve included below the previously identified hierarchy provided by Google in order to understand how they match search terms to corresponding keywords within our campaigns.
Here’s Google’s official hierarchy:
- 1st Priority: Exact match keywords that are identical to the search. For the search term “skydiving license,” the identical exact keyword “skydiving license” is prioritized over any other keyword.
- 2nd Priority: Phrase and broad match keywords that are identical to the search. For the search term “skydiving license,” the identical phrase keyword “skydiving license” is prioritized over the phrase keyword “skydiving.”
- 3rd Priority: AI-based keyword prioritization. The search term “skydiving certifications near me” could match to keywords in several ad groups. Some may be more relevant, such as “skydiving license,” while others may be less relevant, such as “skydiving courses for beginners.” In this case, only the most relevant broad match keywords from the most relevant ad groups are considered.
There’s a glaring point missing from the second priority here. The description states that priority is given to “phrase and broad match keywords that are identical to the search.” This raises the question, what happens when there is both a phrase and a broad match keyword present? Shouldn’t the phrase match keyword get the traffic first?
The basis for justification under Google’s official hierarchy is that when exact, phrase, and broad match keywords are all present in an ad group, traffic will naturally be delivered to phrase and exact matches first, with broad match keywords taking up the least spend. We tested this, and the results we saw were quite different.
You might ask yourself, “Why does it matter which keyword gets the traffic? Isn’t it all the same at the end of the day, within the same campaign?” The answer is that it matters because it’s important to put our money where we know we will generate the greatest return, especially within campaigns that don’t have unlimited funds. This is sometimes in broad match keywords, but not always. If we remove phrase match and exact match from the equation, and rely on broad match to do all of the work, we’re missing out on a few things.
How Do We Even The Playing Field? A New Hierarchy
Through testing, we determined that broad match keywords sit above phrase match keywords in the hierarchy. Additionally, phrase match search terms are just as (if not more) likely to deliver for broad match keywords than phrase match keywords when both are present in a campaign.
The table below highlights our initial approximated hierarchy within Google’s best practice recommendation, based on data from tests we’ve made at our agency.
| Rank | Keyword | Search Term |
| 1st | Exact | Exact |
| 2nd | Broad | Exact |
| 3rd | Phrase | Exact |
| 4th | Broad | Phrase |
| 5th | Phrase | Phrase |
| 6th | Broad | Broad |
The biggest problem with this structure is that, from a hierarchical standpoint, it pushes our broad match keywords to deliver phrase match search terms before our phrase match keywords. Our test results confirm this, as you can see below.
3 Months Under Google’s Recommended Best Practice (Before Period) | |||
| Search Term-Keyword | % of Spend | Avg. CPC | Avg ROAS |
| Exact-Exact | 38% | $1.32 | 3.6 |
| Phrase-Phrase | 27% | $1.64 | 2.7 |
| Exact-Broad | 18% | $1.52 | 2.3 |
| Phrase-Broad | 11% | $1.82 | 2.7 |
| Broad-Broad | 5% | $2.84 | 5.3 |
| Exact-Phrase | 1% | $1.30 | 0.7 |
This is what we saw at the search term and keyword level when we added broad match to our existing ad groups for a period of three months.
Historically, Google’s best practice was to incorporate a mix of phrase and exact keywords within campaigns. This was the case before broad match keywords were introduced, of course.
While we are not recommending a resistance to change, there is virtue to the idea that “if it ain’t broke, don’t fix it.” Phrase match keywords are proven to deliver phrase match search terms that generate 21% lower CPCs than phrase match search terms from broad match keywords.
This confirms the hypothesis that if we move all of our phrase traffic to broad match keywords, we will likely see that drop in quality. Our new structure for campaigns safeguards against this. But even more importantly, our results confirmed that broad match can be an essential element to a well-structured search campaign.
Broad match keywords spent a lot over the course of our entire test—spending 31% of the total budget over the course of six months, the second most of all three keyword types. But there was a clear decline in efficiency over time. By the third month of testing Google’s recommended structure, this was what the search term keyword hierarchy looked like:
| Month 3 (Performance after 3 Months of Google’s Best Practice | |||
| Search Term-Keyword | % of Spend | Avg. CPC | Avg ROAS |
| Exact-Exact | 30% | $1.47 | 4.6 |
| Exact-Broad | 27% | $1.61 | 1.0 |
| Phrase-Broad | 18% | $2.06 | 2.7 |
| Phrase-Phrase | 17% | $1.76 | 1.3 |
| Broad-Broad | 8% | $3.48 | 1.9 |
| Exact-Phrase | 0.3% | $1.12 | 3.5 |
In the before period, we saw a significantly different spend allocation and growth pattern:
- Broad Match Growth: Broad match spend started low but grew sharply over the three months, from $552 in month 1 to $3,388 in month 3. This increase took a larger share of the budget, from 11% in month 1 to 53% in month 3.
- Total vs. Average Spend: In the before period, total spend fluctuated around the average. Month 1 was 19% below average, while months 2 and 3 went above average by 12% and 8%, respectively. However, the incremental increases were more modest than in the after period, when broad match really took off.
- Exact and Phrase Match Cannibalization: Broad match spending in the before period appeared to cannibalize phrase match spend. The broad allocation rose consistently over time, while the phrase percentage fell from 36% to 18%. Exact match saw a less dramatic decline in allocation but still dropped as broad rose.
| 3 Months Under Google’s Recommended Best Practice (Before Period) | |||
| Match Type | Month 1% of Total Spend | Month 2% of Total Spend | Month 3% of Total Spend |
| Broad | 11% | 32% | 53% |
| Exact | 52% | 37% | 30% |
| Phrase | 36% | 31% | 18% |
| Monthly Spend vs Average | -19% | 12% | 8% |
We made the pivot to the new structure at Month 3, and the results by Month 5 (September) displayed a stronger, more efficient spend hierarchy than what existed before, confirming our theories. As you can see, there’s a significant growth in size from “-broad” result volume between the two periods.
| Month 5 (Performance after 2 Months of Our New Best Practice) | |||
| Search Term-Keyword | % of Spend | Avg. CPC | Avg. ROAS |
| Exact-Exact | 46% | $1.97 | 2.6 |
| Phrase-Phrase | 24% | $2.24 | 2.2 |
| Phrase-Broad | 23% | $3.06 | 1.8 |
| Broad-Broad | 7% | $3.57 | 3.1 |
| Exact-Phrase | 0.6% | $2.00 | 0 |
| Exact-Broad | 0.5% | $2.65 | 0 |
The most distinct shift is that by Month 5, we have exact match search terms almost entirely isolated to exact match keywords, with a small volume coming from phrase and broad.
We also have a clear priority on phrase match search terms coming from phrase match keywords. While we continue to see a lot of phrase match search terms coming from our broad match keywords, we know that these new phrase match search terms are distinct from those delivered by phrase match keywords.
The difference is indicated by the distinctions in CPC and ROAS between phrase search terms coming from each type of keyword. At the aggregate level, we can see improvements in efficiency, visualized below with GEI (Growth Efficiency Index, a metric we used here) alongside month-over-month spend growth.
| Month over Month Spend Growth vs. GEI (Growth Efficiency index) | ||||
| Metric | Month 2 (Before) | Month 3 (Before) | Month 4 (After) | Month 5 (After) |
| Spend Growth (%) | 39% | -0.6% | 36% | 53% |
| Change in ROAS (%) | -29% | -16% | 2% | -7% |
| GEI | -0.16 | -1.06 | 1.04 | 0.60 |
These are a lot of metrics and numbers, but the real questions are: What happens to each keyword search term level result? And, what are these campaigns actually delivering? Below, we dig in at this more granular level and highlight some of the most valuable situations we encountered over the course of our test.
Exact Match Keywords Under Our Structure
Exact match performs under our structure as it would under any other structure; all we do is maximize it. Best practice for years now has been to use exact match keywords alongside phrase match keywords. Our data confirmed that there is no reason to break from tradition and separate our phrase match keywords from exact. We found that for a keyword like “Levi’s Denim Jeans,” exact match keywords delivered $300 in spend and $1,000 in revenue over the course of our test.
The phrase match keywords delivering exact matches hardly make up enough spend to be significant. For context, all exact match search terms coming from phrase match keywords spent $110 over the course of the experiment, without generating any conversions. This is understandable given the insignificant spend here.
However, when broad match keywords deliver exact match search terms unfettered, they just do a slightly poorer job of allocating spend to the right keywords. In the initial three months without optimized structuring, broad match keywords drove up spend for exact match terms that could be used to drive innovation through broad-phrase, or broad-broad traffic.
In the before period, when broad match was allowed to infringe on exact, it delivered $4,000 in spend and $10,000 in revenue on exact match search terms coming from broad match keywords, for a ROAS of 2. Meanwhile, exact match dropped $10,000 in spend with $30,000 in revenue for a ROAS of 3. What we are considering here isn’t the strength of broad match keywords at delivering exact search terms; it’s the opportunity cost. Because we already have exact match delivery covered, we don’t need broad match keywords here.
At the core, most of the exact search terms broad match keywords will deliver are ones exact match keywords can better optimize spending for. And if they are exact match variants, we see terms like “jeans denim Levi’s” for keywords like “Levi’s denim jeans”—which is neither significant enough to focus on nor helpful from a revenue generation perspective. Post-restructuring, we decreased our total broad-exact spending down to $12, instead of $4,000. Not only did we decrease the exact match search term volume coming from broad match keywords, we also increased our investment in other available tools.
Phrase Match Keyword Performance
Now let’s look at the performance of different kinds of phrase match search terms, coming from both broad and phrase match keywords. For a given keyword like Levi’s Jeans, phrase match search terms delivered by broad match keywords resulted in $700 in spend and $1,400 in revenue, generating a ROAS of 2. Phrase match keywords only delivered $30 in spend, but generated $200 in revenue, a stronger ROAS of 6.7. This indicates that while broad does have the ability to generate meaningful phrase match keywords, it isn’t going to have the same level of targeted specificity that makes phrase match keywords valuable to our campaigns. That is why it is better to let them work together, instead of letting broad do the job of phrase.
Levi’s Jeans delivered a phrase-phrase match for Levi’s Bootcut Jeans, and broad delivered a broad phrase for Levi’s Bootcut Jeans. Because of our structure, broad was informed not to deliver any more traffic for this specific search term, since phrase is capable of delivering for it. This allowed broad to get more creative and deliver for “Levi’s Flared Jeans Size 32,” which secured conversions for the term and delivered additional impressions for traffic not covered under the best practice ad group.
Broad Match
Finally, we come to the promise of machine learning and the keywords of the future! A lot of the first steps in this new structure are meant to ensure broad has the ability to focus on the one thing it can do that other keywords cannot: identifying new terms and keywords that phrase just isn’t brave enough to go for.
But this purpose gets obscured when broad match keywords attempt to do the same thing exact match keywords do better. As we illustrated above, phrase match suffers when broad match tries to “cheat off their work.” Before we added our scripts and restructured our campaigns, Google’s best practice broad match additions to existing campaigns spent a measly $600 on broad match search terms, and failed to break even from a ROAS standpoint, with a ROAS of only 0.5.
That’s where our restructuring and broad match redirection comes in. After our restructuring, these search terms spent $1,200 and delivered $4,400 in revenue, for a ROAS of 3.7! Results like “Levi’s Cone Mills Skinny Jeans” delivering for a broad match keyword as simple as “Levi’s Jeans” (spend of $5, revenue of $1,400), or “Levi’s Sawtooth Western Top” for “Levi’s Shirt” (spend of $5.65, revenue of $369) showcase the ability of broad match to take a general keyword down to the highly specific product level. This is especially helpful for search terms that may not be covered by a standardized campaign that doesn’t have flawless granularity down to the SKU level.
Broad match was also able to understand elements not covered by the keywords themselves, like delivering for “Levi’s Ultra Wide Skateboarding Jeans” on a “Levi’s Jeans” keyword ($15 in spend, $190 in revenue).
All of this offers a great summary of the changes we saw at a keyword and search term level, but what’s important to a lot of marketers are the macro-level benefits from this new campaign structure.
To further justify the positive benefits of this strategy, we can return to GEI as an indicator of success in scaling up our campaigns. From month 3 to month 5, we generated a GEI of .9 overall, confirming that we saw strong scalability of this strategy.
To summarize, after 3 months we saw the following results:
| After: Months 4-6 Under Our Best Practice | |||
| Search Term-Keyword | % of Spend | Avg. CPC | Avg. ROAS |
| Exact-Exact | 47% | $1.82 | 2.1 |
| Phrase-Phrase | 23% | $2.08 | 2.1 |
| Phrase-Broad | 22% | $2.80 | 2.1 |
| Broad-Broad | 6% | $3.10 | 2.8 |
| Exact-Phrase | 0.5% | $1.69 | 0 |
| Exact-Broad | 0.3% | $2.31 | 0 |
Why Our Strategy Works
Broad match keywords have shown a distinct impact on campaign performance when compared to best practice campaigns. At an aggregate level, broad match keywords tend to have a higher CPC, averaging 70% more than their best practice counterparts. This increased expenditure often comes at the expense of overall ROAS, which is typically 14% lower for broad match campaigns. These findings highlight the need for careful consideration when integrating broad match keywords into advertising strategies.
Moreover, broad match keywords can significantly affect the allocation of budget across different keyword types. They can consume nearly as much budget as best practice campaigns; if not properly managed, this can lead to a detrimental decrease in spend for more precise keyword types, particularly phrase match keywords. Our data analysis revealed a substantial 35% drop in phrase match spend when broad match keywords were present in the same ad group, underscoring the importance of structuring campaigns to mitigate such impacts.
Given these insights, marketers must approach the use of broad match keywords with caution. While they can offer wider reach, the associated costs and potential negative effects on other keyword strategies cannot be overlooked. Effective campaign management practices, such as the implementation of negative keywords and a balanced keyword mix, are essential to ensure that the benefits of broad match do not come at the expense of overall efficiency and effectiveness.
Our unique campaign structure allows us to improve upon Google’s hierarchy:
- Exact match keywords do exact match things well enough that we don’t need phrase or broad match keywords to deliver exact match search terms at all.
- Phrase match keywords need to be able to focus on phrase match search terms to the best of their abilities, without being suffocated by broad match infringement.
- Broad match is at its best when we know it is using Google’s AI to deliver creative search terms through phrase match delivery distinct from our best practice campaign.
- Broad match search terms will never be prioritized by broad match keywords, regardless of what structure we use. They are the bottom rung on the ladder here. With our new structure, we can scale their delivery over time by narrowing the scope of what broad match can steal from best practice campaigns, forcing innovation. With this approach, we saw a 93% increase in broad match search term spending and a GEI of .8 for broad match.
What Broad Match Can Teach Us
Our new strategy resulted in significant spend growth of 93% over a short period of time when looking at the first three months of delivery pre-implementation, compared to the three months after our new structure. We also saw revenue grow 35% in that same period of time. This is a huge improvement over the results we saw when implementing the recommended Google best practice of broad match addition to existing campaigns. In that instance, we saw spending grow only 38% and did not see any growth in revenue. Revenue actually decreased slightly—by 18%—as broad match keywords continued to cannibalize phrase and exact counterparts. This data confirms the hypothesis we posited when first implementing this new structure. Now, you can leverage these learnings and utilize them yourself!
Even without implementing explicit automations, there are a number of things you can learn from this data to improve your Google ads campaigns. Use tools such as spend segmentation to control your investment in new tactics. Finally, our new hierarchy will help you further understand the behavior of different Google keywords and search terms. Want to see how broad match can strengthen your campaigns? We’d be happy to dig in and discuss strategies. Reach out to us using our contact form or connect with us on LinkedIn.
