It is hard to change a strategic perspective. We form our ideas on the world based on data and inferring causation and correlation. Acknowledging that an outcome is no longer viable means that either the circumstances changed or the logic wasn’t sound. Both are uncomfortable.
We’re going to dive into a lot of data in this post and I’m also going to outline my old perspective and how I got there.
An important disclaimer: Just because this post looks at a lot of data and there is a high probability that one path is correct, it does not mean that the other path is outright incorrect.
It simply means that there is a significantly lower probability that you will see the profit and victory your brand deserves going with the “loser” in the data.
Our conclusion, up front
This is a really long post. So, for the sake of time, here’s a TL;DR of the study we conducted.
We analyzed 2637 accounts, conducting a study to explore the effectiveness of Broad Match vs. Exact Match. Due to how closely tied Smart Bidding and Broad Match are, we also analyzed Maximize Conversions and Maximize Conversion Value (1334 accounts). Key findings include:
- Exact Match outperformed Broad Match in terms of CPC, CTR, CPA, ROAS, and conversion rate for the majority of accounts.
- Conversion-oriented metrics like CPA and ROAS favored Exact Match.
- Both conversion volume and click volume were better with Exact Match. Conversion value was flat between both match types.
- The data suggests not making drastic changes if Broad Match is already performing well but considering testing for potential benefits.
- Maximize Conversion Value performed better in terms of CPC, CTR, CPA, and ROAS for most accounts.
- Max Conversion Value had cheaper CPC, possibly due to bid caps and practical ROAS goals.
- CPA was generally better with Max Conversion Value, challenging the belief that higher CPA can lead to higher-value customers.
- The data also recommends using Max Conversion Value and determining conversion value based on customer value and channel conversion rates.
- Test your assumptions and don’t take conventional wisdom for granted.
- Keep evaluating your accounts and bidding strategies to optimize costs and performance.
- Only test Broad if you go in with protections in place and have budgeted for data acquisition.
My original point of view
I strongly supported Broad Match for a long time and would defend the match type in posts that attacked it. I did this for the following reasons:
- The pragmatist in me could see that match types as a mechanic were not really as powerful as they had been (or so I thought). Rather than fighting the current, it made more sense to just make the best with Broad Match.
- Broad Match would often provide Phrase and Exact Match “matched by” in the search terms report, so there was no reason to pay the perceived premium for Exact Match if we could get it with Broad Match.
- Broad Match was enhanced to include audiences that otherwise would not be included unless Smart Bidding was selected.
I strongly favored Max Conversion Value because it leans in to how ad channel algorithms function. However, I would often recommend Max Conversions because setting ROAS goals and customer values represented a struggle for lead generation accounts.
I hate DKI (dynamic keyword insertion) because the syntax ends up being weird and was a strong believer in pinning creative.
- DKI would force keywords into ads regardless of whether it would sound “correct”.
- DKI often gets paired with formulaic ads that don’t speak to the prospect in a meaningful way.
The Details of The Study
We wanted to make sure the data would be as clean as possible so set some pretty strict criteria for accounts we would include in the study.
We went through four different versions of the data and questioned the outcomes to make sure we could confidently stand behind the data.
Here are the considerations we factored in:
- Accounts had to have both things we were comparing (Broad and Exact, Max Conversions and Max Conversion Value).
- Accounts had to have at least 90 days of spend data at the start point of the analysis (we looked at Q1 of 2023).
- Accounts could be any vertical and any spend level. However, outliers (accounts spending more than $5 million per month and accounts that had periods of no spend) were excluded from the study.
- Data looks at the following: which thing had more accounts that did better with the mechanic in question, as well as what was the improvement over the other mechanic.
In the Broad vs. Exact Match study, we had 2637 accounts that met the criteria. These accounts come from all over the globe and vary in vertical and spend; 1402 accounts exceeded $10K per month. Additionally, 1235 accounts had less than $10K per month in spend.
When examining Max Conversions vs. Max Conversion Value, we had 1334 accounts that met the criteria. They were a mix of including and not including goals for tCPA and tROAS.
Our responsive search ad study investigation explored 13,671 ad accounts.
How does Broad Match compare to Exact Match?
We first wanted to look at overall performance and performance gains. It’s important to note that Optmyzr customers tend to be more advanced than the average advertiser, which means we are taking it as a given that the accounts on the whole will have healthy account structures.
We do not enforce a particular structure on our customers, so there will be a mix of all account structures in the data set. All comparisons are looking at how Broad compared to Exact within the same account.
For Cost Per Click (CPC):
- 56.55% of accounts performed better with EXACT, and the median percentage difference is 77.96%.
- 27.34% of accounts performed better with BROAD, and the median percentage difference is 36.96%.
For Click-Through Rate (CTR):
- 85.65% of accounts performed better with EXACT, and the median percentage difference is 84%.
- 13.88% of accounts performed better with BROAD, and the median percentage difference is 36%.
For Cost Per Action (CPA):
- 70.79% of accounts performed better with EXACT, and the median percentage difference is 100.71%.
- 27.48% of accounts performed better with BROAD, and the median percentage difference is 52.52%.
For Conversion Value/Cost:
- 64.12% of accounts performed better with EXACT, and the median percentage difference is 122.40%.
- 19.91% of accounts performed better with BROAD, and the median percentage difference is 79.87%.
For Return On Ad Spend (ROAS):
- 72.52% of accounts performed better with EXACT, and the median percentage difference is 113.47%.
- 26.47% of accounts performed better with BROAD, and the median percentage difference is 64.71%.
For Conversion Rate (CVR):
- 56.73% of accounts performed better with EXACT, and the median percentage difference is 68.63%.
- 22.72% of accounts performed better with BROAD, and the median percentage difference is 50.12%.
We can see that the majority of the accounts perform better with Exact Match, and the median percentage difference is also better for those users that performed better with Exact Match.
For accounts spending over $10,000:
There were a total of 1402 accounts.
- 76.03% of the accounts present had better ROAS with EXACT match. 22.54% had better ROAS with BROAD match. 1.43% had no difference.
- 74.61% of the accounts had better CPA with EXACT match. 24.54% had better CPA with BROAD match. 0.86% had no difference.
- 57.49% of the accounts had better CPC with EXACT match. 29.24% had better CPC with BROAD match. 13.27% had no difference.
- 88.23% of the accounts had better CTR with EXACT match. 11.34% had better CTR with BROAD match. 0.43% had no difference.
- 66.98% of the accounts had better Conversion Value/Cost with EXACT match. 16.98% had better Conversion Value/Cost with BROAD match. 16.05% had no difference.
- 57.20% of the accounts had better Conversion Rate with EXACT match. 17.76% had better ROAS with BROAD match. 25.04% had no difference.
For accounts spending less than $10,000:
There were a total of 1235 accounts.
- 69.07% of the accounts present had better ROAS with EXACT match. 30.36% had better ROAS with BROAD match. 0.57% had no difference.
- 67.21% of the accounts had better CPA with EXACT match. 30.04% had better CPA with BROAD match. 2.75% had no difference.
- 55.71% of the accounts had better CPC with EXACT match. 24.45% had better CPC with BROAD match. 19.84% had no difference.
- 83.00% of the accounts had better CTR with EXACT match. 16.44% had better CTR with BROAD match. 0.57% had no difference.
- 61.78% of the accounts had better Conversion Value/Cost with EXACT match. 23.00% had better Conversion Value/Cost with BROAD match. 15.22% had no difference.
- 56.36% of the accounts had better Conversion Rate with EXACT match. 27.53% had better ROAS with BROAD match. 16.11% had no difference.
The number of accounts using Exact Match wins irrespective of whether or not their spend is over $10,000. But we can see a slight drop in percentages of accounts that had better metrics with Exact Match for those who spend below $10,000.
Spend may not be the biggest factor at play here, but it does affect the numbers slightly.
Does the data translate over to the volume of conversions or other KPIs?
While we can’t show the average volume of the individual metrics (because of the amount of variables in each account), we can show which account had a higher percentage of the volume within the same account.
- 51.28% of the accounts performed better with EXACT, and the median percentage difference is 113.36%.
- 48.56% of the accounts performed better with BROAD, and the median percentage difference is 115.06%.
- 50.32% of the accounts performed better with EXACT, and the median percentage difference is 131.82%.
- 47.26% of the accounts performed better with BROAD, and the median percentage difference is 130.37%.
For Conversion Value:
- 52.09% of the accounts performed better with EXACT, and the median percentage difference is 158.10%.
- 47.29% of the accounts performed better with BROAD, and the median percentage difference is 161.27%.
- 49.30% of the accounts performed better with EXACT, and the median percentage difference is 99.37%.
- 50.36% of the accounts performed better with BROAD, and the median percentage difference is 104.31%.
- 51.28% of the accounts performed better with EXACT, and the median percentage difference is 113.16%.
- 48.56% of the accounts performed better with BROAD, and the median percentage difference is 115.06%.
- 47.11% of the accounts performed better with EXACT, and the median percentage difference is 111.38%.
- 52.89% of the accounts performed better with BROAD, and the median percentage difference is 103.46%.
Broad performs a hair better than exact in terms of cost and impressions. Exact performs in every other metric. However, the difference doesn’t seem to be too large. In terms of magnitude, Broad is better in every case except impressions and conversions.
Breaking down each metric and its respective findings
I was genuinely surprised that Broad Match lost to Exact in terms of auction price. There are a few reasons for this:
- An assumption is that Google would give Broad Match preferential treatment in the auction and therefore discounted rates. While this ended up being incorrect, it is worth noting that this category was one of the closer ones between Broad and Exact. As such, I’m not surprised that some advertisers will still see better CPCs on Broad than on Exact.
- Broad Match tends to have an assumption about it that it will be lower quality, so I thought the human element of bidding down would come into play.
- What I didn’t think about until the data came in was how many accounts would be on manual bidding vs. Smart Bidding. Ironically, the enhancements to Broad (e.g. improved audiences) may have made the algorithm bid more than it should have on Broad, while Exact picked up the cheaper rates. This is pure speculation and I would have no way of proving it, but it is an interesting idea.
- Average CPC tends to be higher for higher quality leads (or so we’ve been conditioned to believe).
- The revelation that Google had been raising the CPC floor by 5%-10% is just enough to bridge the gap between what savings we would expect from Broad vs. Exact. It’s possible if we had run this study a few years ago, the difference in CPC would have been much wider.
The big takeaway from this data point (especially looking at how close low and high spending accounts are) is that you can’t use Broad Match for discounted clicks anymore.
If you use it, you’re using it to gather data on what you should be investing in (and potentially which terms to add as negatives to your account).
CTR (Click-Through Rate)
I don’t think anyone was surprised to see Broad Match had a worse CTR than Exact Match. Broad Match by its very nature is going to expose itself to more queries and therefore be predisposed to lower CTR.
CPA (Cost Per Action)
This is another “not that surprised” category. However, there’s a bit more to dig into here than CTR.
One of the assumptions I and the Optmyzr data team made when we were going through the data is that anything conversion-oriented would be flawed. This was a big reason we only looked at performance in relation to individual accounts and aggregated those results.
However what I was surprised by is how Exact Match did 100% better than Broad when it was the winner, yet Broad Match did 50% better than Exact.
I have a few thoughts on why this might be:
- The sophistication of advertisers can mean they know to set more realistic CPA goals as well as budgets to help the campaigns achieve those goals. This likely contributes to why Broad Match advertisers who did well, saw the respectable average of 50% improvement over Exact.
- CPA is tied to which conversion actions are considered primary and secondary. While this data set looks at Q1 2023 (before the summer 2023 glitch where advertisers saw new conversion actions being created in their accounts in the migration away from UA), it still is in the sphere of influence. As advertisers were migrating to GA4, it is 100% possible that extra conversion actions could have been factored in.
- Because we looked at performance within the accounts, these potential errors/glitches would have been baked in and accounted for. This is more in reference to why the numbers aren’t completely one-sided.
ROAS (Return on Ad Spend)
Similarly to CPA, there is a certain degree of human error baked into anything conversion-related. However, unlike CPA, this metric is very one-sided favoring Exact (even in accounts with less than $10,000 in ad spend).
I was not expecting this to be true due to the perceived hesitation to adopt customer values and value-based bidding. I was expecting this to lead to reduced ROAS adoption.
If anything, this is a great testament to the value of ROAS and value-based bidding because Exact Match would be operating from a perceived point of weakness (lacking the enhancements of Broad Match).
CVR (Conversion Rate)
While this metric feels like CTR, it’s a little less obvious that Exact would win over Broad. There are a few reasons for that:
- Given how much audiences factor into Broad match, there’s an assumption that the conversion rates would have been closer. Additionally, since Exact match got more clicks/interactions than Broad on average, it’s reasonable to expect the conversion rate would be lower because of more leads in the pool.
- Conversion rates are very much dependent on the ad copy and the landing page. I would have expected both match types to struggle or be closer if ad copy/landing pages were a problem, however Exact clearly won.
Match-Type Action Plan
This is not the time to make drastic changes in accounts if things are working for you. If your account is currently running Broad Match and doing well, do not feel you need to pause those winning keywords.
However, if you’ve been considering “upgrading” to Broad, it’s worthwhile to take a pause and consider whether your account will benefit from the test.
If you do decide to test, make sure you pause your existing keywords and add the Broad Match variants manually. If you remove a keyword, you can’t get it back and you’ll likely want to have the ability to backtrack if you don’t like how broad behaved.
Optmyzr does not have a single “recommended” account structure as we see our customers succeed with different strategies. However, one fairly universal theme is that if you run match-type campaigns/ad groups you will likely get hit with impression share lost due to rank and budget.
Consider consolidating these so that you can have fewer but stronger ad groups and campaigns. Again, there is no conclusive “winning” structure. However, if you’re struggling with impression share, that’s a way to mitigate it.
Finally, there is no data to suggest (quite the opposite) that Performance Max is bad. I’d strongly recommend reallocating any paused Broad Match budget into Performance Max. Absolutely use the search themes in Performance Max to help focus those campaigns.
Which does better: Maximize Conversions or Maximize Conversion Value?
We did not include manual bidding in this analysis. However, it is worth noting that 12% of Optmyzr customers currently use manual bidding, while 66% use some form of Smart Bidding (Max Conversions or Max Conversion Value). We attribute this in large part to the heavy adoption of Performance Max, as well as the average size of Optmyzr customers (we tend to focus on $10,000 or higher monthly ad spend).
- 44.98% of accounts performed better with Maximize Conversion Value, and the median percentage difference is 64.30%.
- 36.73% of accounts performed better with Maximize Conversion, and the median percentage difference is 60.61%.
- 52.02% of accounts performed better with Maximize Conversion Value, and the median percentage difference is 62.43%.
- 46.48 accounts performed better with Maximize Conversion, and the median percentage difference is 51.15%.
- 52.55% of accounts performed better with Maximize Conversion Value, and the median percentage difference is 86.29%.
- 46.40% of accounts performed better with Maximize Conversion, and the median percentage difference is 81.04%.
- 60.19% of accounts performed better with Maximize Conversion Value, and the median percentage difference is 107%.
- 39.58% of accounts performed better with Maximize Conversion Value, and the median percentage difference is 91.31%.
When we compared all the accounts the majority performed better with Maximize Conversion Value and the median percentage gains were better as well.
The 645 accounts with over $10,000 spend in Search
- 46.67% of accounts had better CPC with Maximize Conversion Value, 36.9% had better CPC with Maximize Conversion
- 52.09% of accounts had better CTR with Maximize Conversion Value, and 45.74% had better CTR with Maximize Conversion
- 53.18% of accounts had better CPA with Maximize Conversion Value, 46% had better CPA with Maximize Conversion
- 63.26% of accounts had better ROAS with Maximize Conversion Value, 36.4% had better ROAS with Maximize Conversion
The 662 accounts with under $10,000 spend in Search campaigns
- 43.50% of accounts had better CPC with Maximize Conversion Value, 36.86% had better CPC with Maximize Conversion
- 52.57% of accounts had better CTR with Maximize Conversion Value, and 46.53% had better CTR with Maximize Conversion
- 51.81% of accounts had better CPA with Maximize Conversion Value, 46.83% had better CPA with Maximize Conversion
- 57.4% of accounts had better ROAS with Maximize Conversion Value, 42.45% had better ROAS with Maximize Conversion
Spend did not impact Max Conversion Value winning and there was very little change in performance looking at accounts that had over $10,000 vs. less than $10,000 in monthly ad spend.
Breaking Down Each Metric
The biggest surprise for me was that Max Conversion Value had the better (cheaper) CPC because it runs counter to what we know of how the algorithm bids. Traditionally we’d expect the algorithm to bid more aggressively for a lead that would have a higher probability of meeting the objective (conversion value goal).
That Max Conversion Value had the cheaper CPC implies the following:
- Additional automation layering to enforce bid caps.
- The ROAS goals were more practical than I tend to give folks credit for, so the algorithm didn’t spike bids as much as they might have otherwise. This speaks to the data source and the higher probability that Optmyzr customers will manage their accounts at a higher level.
- Bid floors are leveling the playing field so those who are using value-based bidding are getting access to a “smarter” algorithm.
The main takeaway here is that advertisers should not default to thinking cheaper is inherently worse, however getting discounts on clicks is much more about giving data to the algorithm than having a perfect quality score.
I was not terribly surprised that CTR would be better with conversion value because if an advertiser takes the time to put in conversion values, they likely will put more effort into message mapping creative.
That said, both were close, which implies that it’s more on the human running the campaigns as opposed to the bidding strategy directly influencing the CTR.
To be honest, I was expecting Max Conversion Value to have a worse CPA because we’ve been trained to believe that CPA can be higher to get higher value customers. However that it had the cheaper CPA overall is more of a wake-up call than anything not to get complacent on CPAs.
It is worth noting higher spending accounts did slightly better with CPA than lower spending accounts (but ultimately it was negligible).
If you’re struggling with your CPA, consider whether you’re asking your budget to do too many things or if the campaign can get enough clicks in the day to lead to conversions. Both those mechanics can influence CPAs being artificially high.
It should not come as a surprise that the majority of Max Conversion Value campaigns did better than Max Conversions on ROAS. What is interesting is that there were accounts that saw better ROAS using Max Conversions.
I have a few theories on this:
- Some brands are not allowed to use conventional conversions and it’s possible that in those accounts max conversions can do better than ROAS simply because users will represent more than one conversion (and the advertiser intends this).
- Max Conversions might have been in older campaigns which would be predisposed to do well.
It’s important to note that we did not include conversion rate in the data because it was essentially the same.
Bidding Strategy Action Plan
There is no good reason not to use Max Conversion Value. Hiding behind a lack of clear customer value is just giving your competitors the chance to overtake you.
When determining your conversion value the best way to do it is to consider your average customer value against the conversion rate of each channel. If you’re unsure what the average would be, you can start with a minimum SQL (sales-qualified lead) or minimum subscription price. While this won’t be perfectly accurate, it will give you a place to start.
My new outlook
The biggest takeaway from looking at the data is not taking anything for granted. Just because we’re told something is true, it’s important to test and prove whether it’s viable in our accounts before committing to it or discarding it.
Additionally, given that the conventional wisdom—that Exact Match and Max Conversion Values are more expensive because they provide more value—didn’t play out at scale, it’s worth doing a deep dive into your accounts if they are driving up your costs.
Consider being more aggressive with negatives and exclusions, as well as owning whether you have the budget to go after desired transactional traffic or if you’d be better served leveraging your budgets on cheaper networks (Microsoft) or top of the funnel (Performance Max, social, video).
We’re very grateful to our customers for allowing us to enable them on the path to profit and victory and it means a lot to get to continue to empower them through automation and freedom of structure.
And if you aren’t an Optmyzr customer, but need help with running more profitable campaigns, sign up for our 14-day free trial today to give our tools a try.
Thousands of advertisers — from small agencies to big brands — around the world use Optmyzr to manage over $5 billion in ad spend every year. You will also get the resources you need to get started and more. Our team will also be on hand to answer questions and provide any support we can.