In the first article, we learned something that surprises many Power Platform developers:
Dataverse Search doesn’t search your tables directly.
Instead, it searches something called a search index.
That naturally raises the next question:
If users aren’t searching the database, where do the search results actually come from?
The answer lies in one of the most important pieces of Dataverse Search the indexing engine.
What Exactly Is a Search Index?
A search index is a highly optimised structure designed specifically for searching.
Think of it as a giant lookup dictionary.
Instead of storing complete Dataverse records, it stores searchable information extracted from your tables.
Rather than reading every Account, Contact or Case record during every search, Dataverse simply consults this index.
That’s why results often appear in milliseconds.
When Is the Index Built?
This is one of the biggest misconceptions.
Many people imagine that every search starts by building an index.
It doesn’t.
The index already exists before the user searches.
Whenever searchable data changes, Dataverse automatically updates the search index in the background.
That means:
- A new Account is created.
- An existing Contact is updated.
- A Case title changes.
- A Product is renamed.
The indexing service detects these changes and refreshes the searchable information.
By the time a user performs a search, the heavy work has already been done.
Why Not Search SQL Directly?
Let’s imagine your environment contains:
- 12 million records
- 180 tables
- Thousands of users
Now imagine every user typing:
John
If Dataverse queried SQL directly for every search request, it would need to inspect enormous amounts of data repeatedly.
Instead, Dataverse searches an index specifically designed for fast text retrieval.
The database remains responsible for storing business data, while the search index is responsible for finding it quickly.
This separation is one of the reasons Dataverse Search scales so effectively.
What Happens Behind the Scenes?
A simplified flow looks like this:
Notice that indexing happens before users perform searches.
Searching becomes incredibly fast because the expensive processing has already taken place.
Does Every Column Get Indexed?
No.
Only supported, searchable columns are included in the search index.
This is intentional.
Indexing every column in every table would consume unnecessary storage and reduce efficiency.
Microsoft allows administrators to choose which tables participate in Dataverse Search and which columns should contribute meaningful search results.
We’ll explore those configuration options in a later article.
Why Developers Should Care
Understanding indexing helps explain many common questions:
- Why doesn’t a recently updated record appear immediately?
- Why are some columns searchable while others aren’t?
- Why is Dataverse Search significantly faster than Quick Find?
- Why can search span multiple tables simultaneously?
These behaviours aren’t random they are direct consequences of how the search index is designed.
Once you understand indexing, many of Dataverse Search’s behaviours suddenly make perfect sense.
My Trial 