Keyword match types still matter (phrase & exact match are not obsolete)

Keyword match types are a powerful tool for managing ad spend, helping advertisers tailor their ad campaigns to the most relevant audience, driving the right traffic and ultimately leading to higher conversion rates.

With the advent of more advanced targeting options, some marketers have begun to question whether keyword match types still matter.

This article will break down the historical background of match types and why they still play an important role in paid search campaigns today.

Brief background on how we got to our current state

When I started in this industry, Yahoo! was the dominant search engine. It only had two match types (standard and extended), while Google had what we currently have (exact, phrase, and broad; although they also modified broad match for seven years).

When Bing (because I still refuse to call it Microsoft Advertising) completely separated from Yahoo! In the mid/late 2000s it had the same setup as Google.

Yahoo! would eventually switch to the three-match types when it went from Overture to Panama (yes, I’m old), and then again when it launched Gemini (god willing, that’s never coming back!).

Up until the last few years, it was always emphasized that exact match was most accurate for the query, followed by phrase, then modified broad match (if it existed) and broad (which was a kind of crapshoot).

But as things evolved, close-match variants as a standalone feature and wide-match mods went the way of Old Yeller.

Additionally, around 2018, exact match became much looser and began to feel like a modified combination of phrase match and broad match. Needless to say, the industry masses did not take this information well.

Because Google might be the first to use the term “Keywordless AI‘ in February 2023, marketers are questioning the validity of match types moving into the future.

People didn’t take well to match type changes (I saw this at a Google event after the announcement)

For years, broad match keywords have been condemned by large and sophisticated operations or used sparingly, often due to keywords’ lower Quality Score.

Advertisers almost always used exact matches and phrase matches, duplicating the keywords in both match types and placing the exact highest bid and then the phrase.

Some would also use broad but bid the lowest (to minimize risk) mainly to gain insights from the search query reports and create robust lists of negative keywords (I still do this today).

Pro tip that is still relevant today: Never use dynamic keyword insertion (DKI) in ad groups with broad match keywords.

I should note that WebRanking’s James Svoboda blew me away at SMX West 2018 with a hybrid match type that combines broad match modified matches with phrase matches in a single keyword. Unfortunately this is no longer possible.

Keep in mind that this history ignores “keywordless” searches – Shopping (formerly PLAs), Dynamic Search Ads, Local Service Ads, or Local Search; most shopper marketing platforms or niche/unique search engines like Yelp. Not to mention that it’s ahead of the questionable Performance Max.

If we’re leaning toward keywordless search, why are match types important?

I’m glad you asked me that question because I’ve worn my tinfoil hat with it for years.

My only easily proven claim is that the big search (a new term that I hope will catch on) is trying to remove our scrutiny by breaking away from the traditional keyword approach in order to make more money. I realize the band-aid hasn’t come off (yet), and we still have some degree of control over keyword-focused search.

Therefore, it is both relevant and important to focus on match types.

Some direct and simple answers (although this does not yet apply to all) are:

  • Shopping does not apply to all advertisers.
  • Not all advertisers have YouTube assets (and don’t want the engines to create the videos for them because they’re a bit crumbly).
  • Performance Max is expanded. It can be manipulated, but still doesn’t necessarily apply to all advertisers.
  • Not all advertisers want to rotate images or placements (for various reasons).
  • Many advertisers want to be able to control spend and generate reports based on where their ads appear.

The truth is that many advertisers, for various reasons, only want to show up for certain keywords in search and not much else. And “keywordless” efforts really don’t show that.

No real way to tell where your Pmax ads are showingAuthor’s screenshot, March 2023

There’s no real way to tell where your Performance Max ads will appear.

So, why are phrase and exact match still important?

Despite the neutering changes made to them by great search, phrase and exact match still have their power.

In keyword-based searching, exact match keywords still have the most relevance (and therefore Quality Score) to a search query.

This results in a more cost-effective cost-per-click (CPC) – with phrase matching close behind.

Exact is cheaper since QS is actually used more often as a phraseAuthor’s screenshot, March 2023

“Exact” is cheaper because the Quality Score actually includes it more often than a phrase.

The Single Keyword Ad Groups (SKAG) model is largely a dead bid model; The need and use of curated and tightly knit ad groups is still very important.

Typically this can only be achieved via Phrase and Exact Match as Broad Match is more or less a game of Russian Roulette (although I can’t explain why, it’s definitely a riskier gamble with Broad Match in Bing than in Google).

The next need for these match types, and why they’re so important, is one that’s often overlooked: budget cannibalization.

In its simplest terms, budget cannibalization means you have a single pool of money for anyone to draw from with little or no restrictions. So instead of everyone getting an equal share of the funds, whoever takes it fastest gets the most.

Keywordless search bids on a user query are relevant to the site – not a specific keyword that you specifically want to pay for.

This essentially means that a high volume search query can steal the budget of a medium/low volume search query, preventing one advertiser from showing up for both.

Basically, think of keywordless and to some extent broad match as a mix of your branded and non-branded or high volume and low volume combined into a single ad group. Someone will lose out.

So while you can bid on “anything” with a keywordless search campaign, chances are that the off-brand and/or high-volume searches will account for the majority of spend.

Other potential searches you could show for (long tail, brand, medium/low volume, etc.) don’t get a decent budget to work with (or no budget at all).

Important NOTE: Some of this is or will soon be driven with campaign-level negative ads in Performance Max (already applicable to Shopping).

So if you want to ensure your important keywords (e.g. brand, high volume/higher conversion, etc.), a keyword-based search engine consisting of phrases and/or exact matches remains 100% required.

Making it a standalone campaign (I still love doing match-type isolation at the ad group level as well) ensures that specific keywords or searches don’t have to fight to get trigger funds.

You will have a separate stand alone budget for this. (And before you ask, no, a shared campaign budget isn’t going to help you in this scenario, no matter how much you think.)

take that away

When all is said and done, here’s what should really be taken away from this article:

  • The big search is pushing hard towards a world of search advertising without keywords.
  • Keywordless search, despite optimization, lacks control and transparency, leading to cannibalization.
  • The current broad match search isn’t much better than “keywordless” search.
  • Ad groups composed of phrases and exact matches are the only way to ensure you’re bidding on your intended search query and to minimize the lack of transparency.
  • Finally, because I haven’t mentioned it anywhere before: The most important match type of all is negative match.

More resources:

Featured image: Just dance/Shutterstock