A few months ago Bryant Garvin wrote an interesting post arguing how phrase match keywords have negligible use in Google AdWords campaigns, and how exact and modified broad match keywords should instead provide sufficient opportunity for most Google AdWords advertisers to maximise value from their campaigns. The following reasons were cited for the apparent lack of need to use phrase match keywords:

  1. PPC marketing is becoming increasingly complex, with phrase match keywords adding extra (unnecessary) complexity to campaign management, requiring additional time and effort
  2. Any search query which could potentially be matched to a phrase match keyword could instead be matched to a modified broad match keyword, making phrase match obsolete

Since the first reason (additional complexity) isn't really a valid excuse to largely write-off any PPC strategy, let's explore the second, more intricate, reason in more detail.

The theory argued by Bryant is that any relevant search query matched to a modified broad match keyword and discovered in the search query report can simply be added as an exact match keyword, with modified broad match simply being used as a generator of new exact match keywords. Due to phrase match being more restrictive than modified broad match, any search query which could have been matched to a phrase match keyword would have instead been captured by modified broad match keywords, rendering phrase match obsolete.

However, while I agree with the use of broad match as a generator of new keywords, and agree that modified broad match has numerous benefits, excluding phrase match from Google AdWords campaigns has one major disadvantage.

Price Discrimination

What if I were to take away location bid adjustments from your Google AdWords campaigns? Or remove your ability to apply different bids for mobile users? What if I were to delete your time of day and day of week bid settings from your campaigns? How would you feel?

Similarly, removing phrase match provides you with one less opportunity to differentiate and discriminate between different types of search queries, and apply bids based on relative performance insights to maximise overall value.

Phrase match essentially provides you with an opportunity to discriminate between searches which show varying levels of profitability. If you instead grouped those variations in performance into one modified broad match keyword, any statistical variation between different keywords is averaged into the modified broad match keyword, preventing you from discriminating between keywords to improve overall campaign performance.

For example, consider the two scenarios, one which used phrase match keywords alongside exact and modified broad (scenario 1), and another which neglects phrase match and uses only exact and modified broad match (scenario 2).

Scenario 1 (using exact, phrase, and modified broad match keywords)

In this scenario, 400 clicks are distributed to 4 different keywords. The exact match [california hotels] achieves an average CPA of $25.00, while the phrase match "california hotels" performs considerably better ($12.50 CPA) than the phrase match "hotel california" ($50.00). The modified broad match keyword has a CPA of $50.00.

Scenario 2 (using exact, phrase, and modified broad match keywords)

However, in a scenario where phrase match is not used, searches which would otherwise have been matched to phrase match keywords must instead be matched to the modified broad match keywords. The clicks and conversions from the two phrase match keywords have simply been added to the modified broad match keyword.

Both scenarios result in exactly the same total number of clicks (400), and exactly the same total number of conversions (16). Both scenarios also achieve exactly the same average CPA of $25.00. If you used only exact and broad match keywords in your campaigns as Bryant suggests (scenario 2), you would likely set similar bids for your exact and modified broad match keywords due to their similarity in profitability ($25.00 CPA).

However, by allowing the search query traffic to distribute to phrase match keywords (scenario 1), you are suddenly able to spot significant differences between keywords of different word orders. One of your phrase match keywords ("california hotels") is performing exceptionally well, achieving an average CPA of $12.50, while another of your phrase match keywords ("hotel california") is performing relatively poorly, achieving an average CPA of $50.00. Simply adjust your bids based on the relative performance of each keyword, and the result is an increase in total conversions (from 16 to 19) and a reduction in average CPA (from $25.00 to $21.00).

By using phrase match keywords, you can hunt for significant differences in keyword performance, and optimise bids accordingly. Without phrase match keywords, any significant variation in keyword performance is averaged out, forcing you to work with an average bid. The result is, well, average campaign performance.

By removing phrase match, you are essentially removing a valuable signal of variation, and removing and excellent opportunity to better optimise your campaigns. It's like taking away the ability to apply bid adjustments based on time of day, day of week, or location, or like taking away the ability to send mobile searchers to a mobile-friendly landing page.

Phrase Match Is Not Dead

Although Bryant does point out that some rare cases for phrase match may still exist for some campaigns (for example to distinguish irrelevant searches such as 'how do I store apples' from relevant searches such as 'apple store'), phrase match has considerably more value that simply being relegated to occasional status.

Approximately 20-25% of searches made every day on Google have never been made before, making it impossible to target a high percentage of long-tail search query traffic using exact match keywords. Since phrase match is less restrictive than exact match, phrase match allows you to extract a greater percentage of the long-tail from your broad match traffic. Sending search query traffic sent to a larger number of ad groups allows you to show tailored ad messages for a greater percentage of search traffic, which would not be possible to the same extent if all phrase match traffic were to be sent to broad match keywords.

In one lead-generation Google AdWords account I'm currently running for a client, which contains exact, phrase, and modified broad keyword variations in equal proportion, 49.7% of search queries were matched to phrase match keywords in an average month. In another account (ecommerce), 25.5% of search queries matched to phrase match keywords. And in two other Google AdWords accounts (homeware and building products), 23.5% and 35.8% of search queries were phrase matched. All 4 Google AdWords account had exact match negatives in the phrase match ad groups, and phrase match negatives in the modified broad match ad groups, making it impossible for any cannibalisation to have occured due to higher ad rank causing a less restrictive keyword to show in favour of a more restrictive keyword.

What this means is that at least 20% of Google AdWords search query traffic in a typical Google AdWords account could not have been picked up by exact match keywords, and would simply have been matched to modified broad match keywords if it wasn't for phrase match. That's 20% of search query traffic that can be measured separately, given strategic bids, shown tailored ads, and taken to specific landing pages.

That's a lot of opportunity to forsake by ditching the phrase match.