BigQuery Views for the PDE Exam: Standard, Materialized, Authorized

BigQuery views show up on the Professional Data Engineer exam in a very predictable way. You get a scenario, and you have to pick between a standard view, a materialized view, or an authorized view based on cost, freshness, and access control trade-offs. The wording on the question is often subtle, so knowing exactly when each one wins is the difference between a guess and a guaranteed point.

In this post I want to walk through the three view types the way I think about them when I am teaching the Professional Data Engineer curriculum, plus a quick note on saved queries because those occasionally appear in distractor answers and trip people up.

Standard views: virtual tables that recompute every time

A standard view is the default and most common type of view in BigQuery. It is a virtual table that represents the result of a query. The key thing to internalize is that a standard view does not actually store any data. Every time you query the view, the underlying SQL runs again and the results are computed on the fly.

That has two big consequences for exam scenarios:

• Standard views always reflect real-time data. Since the query runs each time, you never have a staleness problem. The view always shows whatever the underlying tables contain right now.
• Standard views can get expensive if accessed frequently. Recomputing the same query over and over burns slot time and scans the same bytes repeatedly.

So when should you pick a standard view on the exam? Look for these signals in the question stem:

• The query is accessed infrequently
• The query computation size is small
• Storage costs for caching are described as too high or unnecessary
• Real-time freshness is required and the team does not want to manage refresh schedules

If the scenario mentions a dashboard that runs a query against a multi-terabyte table every minute, a standard view is the wrong answer. If it mentions an analyst who pulls a small summary once a week, the standard view is correct.

Materialized views: pre-computed and cached

Materialized views flip the trade-off. Instead of recomputing every time, BigQuery pre-computes the query result and caches it. When you query the materialized view, you read the cached result instead of rescanning the underlying tables. BigQuery also keeps the cache fresh automatically as the base table changes, within the refresh tolerance you configure.

For the Professional Data Engineer exam, the pattern is straightforward:

• Query is accessed frequently and the same aggregation or filter runs over and over
• Query computation size is high and rescanning the base table is expensive
• Storage costs for caching the result are low relative to the compute savings
• Real-time data reflection is not strictly required, or the small refresh lag is acceptable

The economic argument is the one Google likes to test. If you have a heavy aggregation that thousands of dashboard users hit every hour, recomputing it from scratch every time is wasteful. The materialized view pays a small storage cost so that everyone reading from it pays a tiny scan cost instead of a giant one. That is the cost-effective answer for frequently accessed, expensive queries on large datasets.

Watch for distractor answers that suggest scheduled queries writing into a destination table. That works, but a materialized view is the cleaner, lower-maintenance choice when the goal is just to cache an aggregation result. If the question asks for the most cost-effective solution for a frequently accessed query over a large dataset, the materialized view is what you want.

Authorized views: security boundary, not a performance feature

Authorized views are less common in real-world usage, but they show up on the exam often enough that you need to recognize them. An authorized view is a security mechanism. It lets users query the results of a specific query without granting them access to the underlying tables.

The mental model I use is this. You have a sensitive table that holds customer records, and you want to expose only a few aggregated columns to a partner team. You do not want to give them SELECT on the raw table because that would expose sensitive fields. Instead, you write a view that selects only the safe columns, put it in a separate dataset, and authorize that view to read from the source dataset. The partner team gets access to the view's dataset but not the source dataset. They can query the view, see the filtered results, and never touch the underlying data.

On the exam, the giveaway phrases are:

• "Restrict access to sensitive data while sharing only the results of a query"
• "Different teams need different levels of access"
• "Control data visibility in a multi-user environment"
• "Share specific query results without exposing the underlying tables"

If you see those signals, authorized view is the answer. Do not pick column-level security or row-level access policies unless the question specifically asks about column or row filtering on the same table. Authorized views are about exposing a query result to a different audience entirely.

Saved queries are not views

One quick note. Saved queries sometimes show up as a distractor answer. A saved query is just a predefined query you can save in the BigQuery console, share with teammates, and re-run on demand. It still requires full computation every time it runs, and it has no security or caching properties. If a question is asking about access control or performance optimization, saved queries are almost never the right answer. They are a productivity feature, not a data engineering primitive.

How to attack a view question on exam day

When I see a view scenario on a practice exam, I run through three questions in order:

• Is the question about security or data exposure? If yes, the answer is an authorized view.
• Is the question about performance and cost on frequently accessed queries? If yes, materialized view.
• Is the question about real-time freshness, low access frequency, or small computation? If yes, standard view.

That ordering catches almost every variant of this question family. Security signals override everything else, then cost-versus-freshness trade-offs decide between the two performance options.

My Professional Data Engineer course covers BigQuery views alongside the rest of the BigQuery storage, query optimization, and access control topics you need for exam day.