Most advertisers think Google Ads works like this:

Exact match keywords enter one auction.
Phrase match keywords enter another.
Broad match keywords enter a third.

But, that is not what actually happens!

There is only one auction for each search. Before that auction starts, Google has to decide which single keyword from your account is allowed to participate. If you have overlapping Exact, Phrase, and Broad keywords, Google runs an internal selection process to choose the one it believes will perform best for that query.

Understanding that selection process is the key to succeeding in Google Ads.

Let’s take a look at how it works!

The search happens

Let’s use this as an example. A user searches:

“virtual counselor for schools”

You have three keywords:

• [outsourced school counselor]
• “outsourced school counselor”
• outsourced counselor

All three can be considered a match. But only one can enter the auction.

Google now evaluates all eligible keywords in your account and asks:

Which one will generate the highest Ad Rank for this search?

How Google compares match types

Google calculates a predicted Ad Rank for each eligible keyword based on:

• Expected click-through rate
• Expected conversion rate or value
• Ad relevance
• Landing page quality
• Your bid or Smart Bidding target

The keyword with the highest predicted Ad Rank is selected. The others are discarded for that search.

This means match type alone does not decide which keyword runs. Performance prediction does.

Where Exact match gets priority

Now hold on. It get’s a little complicated. Exact match does get one special rule.

If the user’s search query is identical or almost identical to an Exact match keyword, that Exact match keyword is selected automatically. Phrase and Broad are not allowed to compete in that situation.

So if you have:

[outsourced school counselor]

And the user searches:

“outsourced school counselor”

That Exact keyword enters the auction.

But this rule only applies to literal or near literal matches. It does not apply to meaning based matches.

When all match types compete

For queries that are related in meaning but not identical in text, all eligible match types are allowed to compete. That includes Exact, Phrase, Broad, and even Performance Max search themes.

In these cases, Google relies entirely on predicted performance.

This is where the differences between match types matter.

Why Broad often wins

Broad match is allowed to use more user level signals when predicting performance. These include:

• Past search behavior
• Browsing activity
• Location patterns
• Device signals
• Similar users’ conversion history

Exact and Phrase are more tied to the query itself.

So for a user who looks highly likely to convert, Broad often receives a higher predicted conversion rate. That increases its Ad Rank and allows it to be selected more often for expanded or meaning based queries.

What this means for reporting

Because different match types are selected for different kinds of users, performance data can look skewed.

Broad often appears to have:

• Higher conversion rates
• Lower CPA
• Better ROAS

Exact often looks less efficient even though it is still capturing the core demand.

This is not because Broad is targeting better queries. It is because Google is routing different types of users to different match types based on predicted behavior.

Why structure matters

When Exact, Phrase, and Broad all live in the same campaign with the same Smart Bidding goal, you give Google full control over how traffic is distributed.

If you want more visibility and control, the best practice is to:

• Separate Broad into its own campaigns
• Give it its own budget and bidding target
• Use Exact and Phrase as your demand capture layer

This forces each match type to prove its value instead of being blended into one optimization pool.

My Advice

Google Ads today is less about picking the perfect keyword and more about knowing how Google decides which of your keywords gets shown. Exact, Phrase, and Broad are not really separate levers anymore. They are just different signals feeding into the same selection system that tries to predict what will work best for each search. When you look at it that way, a lot of the strange things you see in your account start to feel much more logical.