We dive into the effectiveness and potential of broad match in paid search advertising. While the digital media landscape constantly introduces new products and features, it’s intriguing that one of the most significant improvements in 2022 comes from a longstanding tool: broad match.
We review Google’s claims and share the results of our own experiments conducted at Brainlabs. The findings reveal the considerable impact of broad match, with a median uplift of 19% and a remarkable 85% success rate in delivering positive results.
This document emphasizes the value of broad match, its ease of implementation, and its potential to enhance performance, making it a must-have tool in paid advertising.<\/p>\n\n\n\n
2022<\/p>\n
A Brainlabs review of broad match<\/p>\n
Andy Goodwin
Head of Paid Search, UK
at Brainlabs<\/p>\n\n
In the world of digital media \u2013 and in paid search in particular \u2013 it can feel like\nwe\u2019re always on high alert, waiting to jump on the next product announcement\nor new beta that will come along and help us to squeeze incremental\nperformance out of our paid marketing budget.<\/p>\n
So it\u2019s maybe somewhat counterintuitive that in 2022, the single biggest\nimprovement you can make to your paid search program comes from a product\nthat has been around since the inception of Google AdWords, over 2 decades\nago. Broad match is that product.<\/p>\n\n
From their internal studies, Google suggests that advertisers\nusing smart bidding who adopt broad match see an average\nconversion increase of 30%, at a similar CPA. The exact figure\nyou hear quoted may vary depending on the exact nature of\nthe test \u2013 for instance, if your KPI is conversion value rather\nthan conversions \u2013 but broadly speaking, the number you\nhear won\u2019t deviate too far from this 30% mark.<\/p>\n
As an agency, our role is to run our own experiments that\neither validate or reject these claims from media partners, so\nthat we can present the right solutions to our clients. At\nBrainlabs we\u2019ve invested heavily in this capability, and our\n\u201cHippocampus\u201d database allows us to track the results of all\ntests that have been run across the agency.<\/p>\n
From our extensive recent testing of broad match, our\nexperiments uncovered two major findings.<\/p>\n\n
We found convincing evidence corroborating Google\u2019s\nclaims. Whilst falling just short of the 30% figure, the median\nuplift we recorded was +19%, which is by no means a figure\nto be scoffed at.<\/p>\n<\/li>\n
Arguably more interestingly, our analysis reminded us just\nhow much can be hidden behind an average. We didn\u2019t just\nsee an impressive median uplift; we also saw broad match\ndeliver successful tests with remarkable consistency. Out of\nall 54 tests we have on record, broad match emerged as the\nwinning variant on 46 occasions. When you consider the\narray of external factors that can impact testing in digital\nmedia, an 85% success rate is almost unheard of.<\/p>\n<\/li>\n<\/ol>\n\n
The median uplift observed in\nbroad match tests; 19%<\/p>\n
54 Tests analysed 29 Advertisers 12 Verticals<\/p>\n
Chart annotations key<\/p>\n
Tests in which broad match was\nthe losing variant; 15%<\/p>\n\n
But let\u2019s take a step back from the nuances for a moment. Regardless of\nwhether the uplift is 19% or 30%, that\u2019s a number you can\u2019t afford to\nignore \u2013 especially for a feature that requires almost no effort to\nimplement. And if your budget won\u2019t stretch as far as a 19% increase, the\ngood news is that it doesn\u2019t need to: by enabling broad match and\nmaintaining your current investment, you can instead convert that\nadded-value into an improvement towards your CPA or ROAS goals.<\/p>\n
If there has ever been a silver bullet in paid advertising, broad match is it.<\/p>\n\n
Over the years, broad match has made a bit of a bad name\nfor itself – and it\u2019s easy to see why. It\u2019s not too long ago that\nthe presence of broad match in an account was a surefire\nindicator of a hands-off approach to management, leading to\nsuboptimal results, and ads being served to users who were\nreally not even close to converting with your product.<\/p>\n
As time has passed, the rise of broad match has been the\nstory of gradual & consistent technological progress. Whilst\nthere\u2019s a lot of noise around the topic nowadays, there\nprobably aren\u2019t too many outside of Google\u2019s engineering &\nleadership teams that could have anticipated just how big a\nhit it would become more than three years ago.<\/p>\n
If there ever was a turning point, it came in February 2021. In\none of their flagship announcements of the year, Google told\nthe industry:<\/p>\n\n
\nBroad match modifier will be\nsunsetted!<\/p>\n<\/blockquote>\n
\nPhrase match will be expanded\nand will replace BMM!<\/p>\n<\/blockquote>\n\n
And in the small print, right at the end of the update:<\/p>\n\n
\nAlso, broad match will begin accounting for a greater number of signals,\nsuch as landing page, other keywords in your ad group, and prior user\nsearch behaviour\u201d.<\/p>\n<\/blockquote>\n\n
This was quite an understated way to announce\nbroad match as a serious performance product, and\nit certainly wasn\u2019t the concerted push for adoption\nthat we\u2019ve become accustomed to with Google.<\/p>\n\n
What\u2019s actually been going on behind the scenes?<\/h2>\n\n
Encapsulated in that small print was the fact that there had\nbeen a fundamental algorithm shift towards semantic\nmatching, rather than syntactic matching. With this change,\nbroad match ceased being the fire-hose it once was, and\nbecame a genuinely viable tool to accurately expand your ad\ntargeting.<\/p>\n
The product improvements haven\u2019t stopped there, and today,\nGoogle\u2019s vision for broad match continues to reflect a\nslow-burn of updates and performance enhancements:<\/p>\n\n
\n
- \n
Utilising state-of-the-art natural language recognition models<\/p>\n<\/li>\n
- \n
Continual improvements to data quality and infrastructure<\/p>\n<\/li>\n
- \n
Simplified management of broad match<\/p>\n<\/li>\n
- \n
Updates to the keyword prioritisation process<\/p>\n<\/li>\n
- \n
Building a joint optimisation process across creative, bidding &\nkeyword targeting<\/p>\n<\/li>\n<\/ol>\n\n
Of course, nowadays Google are anything but subtle in the\nway they talk about broad match with their agencies and\nadvertisers. Recommendations tabs in Google accounts\nacross the World are littered with the call to arms of \u201cupgrade\nyour existing keywords to broad match\u201d, as one of the most\nconsistently highly ranked opportunities.<\/p>\n
Despite the bolder approach that Google are now taking in\ntheir comms, the steady stream of product changes and\nrefinements will continue, with it being likely that many will go\nunnoticed for the most part.<\/p>\n\n
When can Broad match go wrong?<\/h2>\n\n
You already know this by now, but auction-time bidding is table stakes in paid search. However, this is never more true than when\ncombined with broad match. If you aren\u2019t using this Google Ads feature to tailor your bids in real-time, then for sure broad match will\nsee you spending a lot of money across a lot of users who each have very little interest in your product right now.<\/p>\n
Flipping that logic on its head though, it\u2019s also crucial to appreciate the value that some<\/em> of those more tenuous-looking queries bring\nto the table, when combined with smart bidding. As humans, we can very easily dip into a search query report, find a handful of\nquestionable examples & overreact: \u201cthese queries will almost never convert for us, we need to exclude them!\u201d But as long as there is\na non-zero<\/em> probability of those users converting, then by setting an appropriately low bid you are well justified in serving an ad.<\/p>\n\n
\n\n
\n \nQuery<\/th>\n Conversion likelihood<\/th>\n Bid<\/th>\n<\/tr>\n<\/thead>\n \n \u201cbuy sofa\u201d<\/td>\n 1.00%<\/td>\n \u00a32.00<\/td>\n<\/tr>\n \n \u201cpopular sofa colours\u201d<\/td>\n 0.50%<\/td>\n \u00a31.00<\/td>\n<\/tr>\n \n \u201chow to plan for moving house\u201d<\/td>\n 0.02%<\/td>\n \u00a30.04<\/td>\n<\/tr>\n \n \u201creturn sofa\u201d<\/td>\n 0.00%<\/td>\n \u2013<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n Example: preemptively excluding \u201chow to plan for moving house\u201d causes us to miss out on the 1 in 5,000 users who would go on\nconvert after searching that term – even though we\u2019re already apportioning such a small amount of our budget to these queries.<\/p>\n\n
To be clear, we\u2019re not advocating an entirely hands-off approach to search query reports, and you will need to use some element of\nhuman judgement. But if performance is your goal, then you really should be trying to give the algorithm as much freedom to optimise as\nyou\u2019re comfortable with.<\/p>\n
Another common pitfall is to start focussing on the wrong metrics. The dynamics of your paid search account won\u2019t look the same after\nadopting broad match – and that\u2019s okay, because the numbers you care about will be going in the right direction. All else being equal, a\nmove to broad match should typically see your campaign metrics do something like the below:<\/p>\n\n
Clicks<\/strong> Impr<\/strong> CTR<\/strong> CPC<\/strong> Cost<\/strong> Conv<\/strong> CvR<\/strong> CPA<\/strong> Value<\/strong> AOV<\/strong> ROAS<\/strong><\/p>\n\n
It\u2019s true that some of these numbers are in the red, and you can already hear the reaction from the less assured marketers in the room –\n\u201cWhy are my CTR & conversion rate going down?! broad match must be driving less qualified traffic!\u201d But the point is, you\u2019ve delivered\nincremental value and conversions with that additional traffic – and because your bid strategy has paid less for the clicks, there\u2019s not been\nany adverse impact on your CPA & ROAS.<\/p>\n
Finally, there\u2019s an increasing emphasis on being able to steer Google\u2019s automation in the right direction. Remember: an algorithm will\noptimise single-mindedly towards achieving a conversion goal. So if you haven\u2019t developed a conversion goal that\u2019s sophisticated enough\nto distinguish your most and least valuable customers, then you should be prepared to see an increase in low-quality customers.<\/p>\n
Depending on your business, that might mean MQLs that never turn into sales, or it might mean one-off shoppers who never come back\nfor repeat purchases. Sharing first-party conversion data with Google is the solution to offset this trade-off.<\/p>\n\n
In conclusion<\/h2>\n\n
As a matured, world-weary PPC marketer, you\u2019ve every right to\nhave a healthy level of skepticism about broad match. Try to\novercome that, and bear in mind the steady progress that has\nbeen made in the field of machine learning.<\/p>\n
Whilst you\u2019re going through that mindset shift, make sure you\nstay focused on the objectives of your marketing – and ideally,\nthink about how you can advance these objectives to better\ntarget your most valuable customers.<\/p>\n
But most of all, make sure you take action – don\u2019t let that 19%\nuplift pass you by. It\u2019s no exaggeration to say that broad match\nis at a stage where we\u2019d consider it a fundamental component\nof your search activity. If it isn\u2019t on your roadmap, assess how\nyou can bring it to the forefront – and get testing today<\/em> .<\/p>\n\n
Find out more<\/h2>\n\n
APAC<\/strong><\/p>\n
Gyn Ang
APAC SVP
gyn.ang@brainlabsdigital.com<\/p>\n\nEMEA<\/strong><\/p>\n
Leo Jennings
UK SVP, GROWTH
leo@brainlabsdigital.com<\/p>\nRachelle Risner
VP, NEW BUSINESS
rachelle.risner@brainlabsdigital.com<\/p>\n\nThe Americas<\/strong><\/p>\n\n\n
<\/div>