Import, Duplicates, and Data Quality

The mental model: before, during, and after the load

A data import does not begin when you click Start Import. It begins when you decide which records should enter, which object receives them, which fields are required, how existing records will be recognized, which rules can block the load, and how you will review results. On the exam, that complete view matters more than memorizing a menu.

Think in three moments. Before the load, clean the file, standardize formats, remove extra spaces, review duplicates, confirm owners, and prepare picklist values. During the load, choose the tool, object, operation, matching criteria, file, and field mapping. After the load, review success and error files, validate counts, inspect Salesforce data, and fix failed records.

Most distractors ignore one of those moments. One option may upload records but duplicate them. Another may update records without identifying the correct match. Another may choose a tool that cannot export. Another may suggest deactivating duplicate rules without understanding the cause. A good answer protects integrity, traceability, and cleanliness.

Practical rule: if the question mentions volume, export, delete, or automated loads, compare Data Loader; if it mentions a small or medium guided load for common objects, compare Data Import Wizard.

Data Import Wizard vs Data Loader

Data Import Wizard is an in-Salesforce tool for importing accounts, contacts, leads, solutions, campaign members, person accounts, and custom objects. Salesforce documents that it can import up to 50,000 records at a time and works with CSV files. It is a good option when the administrator needs a guided experience, a relatively limited load, and supported objects.

Data Loader is a client application for bulk import and export. It can insert, update, delete, and export records, works with CSV files or database connections, supports higher volume, and provides detailed success and error files. In exam scenarios, Data Loader often wins when you see export, delete, hard delete, scheduled or command-line style operation, very large volume, all objects, or a need to handle millions of records.

Do not choose by habit. If the question says "import 20,000 leads from a CSV and map fields with a guided tool", Data Import Wizard may be enough. If it says "export data", Data Import Wizard is not the tool. If it says "update 300,000 records", think Data Loader. If it says "avoid duplicates while importing leads by email", both tools can interact with duplicate rules, but you must review matching criteria and active rules.

Prepare data before importing

File preparation is an examinable part of the topic. Trailhead recommends reviewing data before import: meet system and custom validation requirements, remove duplicates, remove blank spaces, avoid unnecessary empty values, name columns in a way that maps to Salesforce fields, and standardize formats such as country, state, phone, date, and picklists.

A poorly prepared CSV can fail for reasons that are not the tool's fault. Picklists with disallowed values, owners that do not exist, incorrect date formats, empty required fields, invalid emails, copied IDs with mistakes, ambiguous column names, and invisible spaces create errors or dirty data. In exam questions, if the scenario mentions import errors, unmapped fields, or inconsistent values, the answer may be to review and clean the file before trying again.

Also think about relationships. If you are importing contacts related to accounts, you need to identify how the correct account will be found. If you are importing custom records with parents, you need IDs, External IDs, or the right load order. The exam rewards loading data in a way that respects dependencies, not just getting a file into the system.

External ID, Salesforce ID, and upsert

Salesforce ID is the unique internal identifier of a record. It works well for updating existing records when you have already exported or know those IDs. External ID is a custom field marked to hold identifiers from an external system. It is key when you need to relate or update records without relying on Salesforce IDs, especially in migrations or integrations.

Upsert combines update and insert: if Salesforce finds a record that matches the chosen identifier, it updates it; if not, it creates a new one. For Administrator, the word upsert often appears when data must stay synchronized with an external system or when a file contains both new and existing records. Without a reliable identifier, upsert can create duplicates or update the wrong records.

Data Import Wizard supports matching custom objects by name, Salesforce ID, or External ID, with considerations around case sensitivity and uniqueness. In questions, if you see "match records from another system", "unique value from ERP", "update existing and create new", or "avoid duplicates during recurring imports", External ID is usually a strong clue.

Matching rules and duplicate rules

Salesforce Duplicate Management relies on two rule types. Matching rules define how Salesforce identifies records that may be the same, using field criteria and matching algorithms. Duplicate rules define what happens when a match is found: allow, alert, report, or block depending on configuration. Trailhead summarizes the relationship clearly: matching rules find candidates; duplicate rules determine the action.

This is a classic exam distinction. If the question says Salesforce must identify duplicates using email, phone, name, address, or fuzzy combinations, think matching rule. If it says users must receive a warning or creation must be blocked when a duplicate exists, think duplicate rule. If it says an import fails with a duplicate message, review active duplicate rules and their matching rules.

Do not deactivate rules as the first answer unless the scenario justifies it. Salesforce warns that deactivating duplicate management can create duplicates and confusion. A more mature answer is to review the rule, understand the criteria, clean the file, adjust data, change allow or block behavior when appropriate, or import in a way that respects the intended match.

Validations, errors, and result files

Imports can fail because of validation rules, required fields, formats, permissions, duplicate rules, ownership, missing relationships, or automations that run on save. The fact that a file looks correct in Excel does not mean Salesforce will accept it. The administrator must review error messages and fix the cause, not repeat the same load and hope for a different result.

Data Import Wizard and Data Loader generate results that show which records loaded and which failed. Data Loader in particular produces detailed success and error files. On the exam, if the question asks what to do after a partial load, look for options such as review error file, correct failed rows, reload only failed records, validate counts, and inspect imported records.

Testing with a few records also matters. Salesforce recommends using a small CSV first to confirm the file is prepared correctly. For important loads, a sandbox or sample test avoids large mistakes. If the question involves a big migration, critical data, or many dependencies, an answer that includes planning, template, test, validation, and error review is usually stronger than loading directly into production.

Data quality as ongoing work

Data quality is not only the day of the import. It includes preventing bad data, detecting problems, correcting them, and maintaining standards. Required fields, validation rules, duplicate rules, matching rules, controlled picklists, help text, page layouts, Flow, reports, and dashboards can all help users capture consistent information.

An administrator should distinguish reactive cleanup from prevention. If duplicates already exist, you can use reports, merge, duplicate management tools, or cleanup processes. If you want to prevent them from returning, you need rules, validations, training, and better data-entry design. If users type "USA", "U.S.", "United States", and "Estados Unidos", the solution may be standardizing values and using picklists where appropriate.

In certification questions, data quality mixes with adoption. Bad data reduces trust in reports, forecasts, automations, and Agentforce. A dashboard does not fix inconsistent data; it only displays it. If the root problem is capture, fix the entry process. If the problem is historical, clean and deduplicate. If the problem is integration, agree on identifiers and synchronization rules.

Common mistakes when studying imports and duplicates

The first mistake is memorizing "Wizard small, Loader large" without reading the operation. Data Import Wizard imports, but it does not export. Data Loader imports and exports, and it supports operations the Wizard does not cover. The second mistake is ignoring the object: not every tool supports every object in the same way.

The third mistake is confusing matching rule with duplicate rule. One detects matches; the other decides whether to alert or block. The fourth mistake is treating External ID like any text field. External ID matters because it represents a reliable identifier from another system and should be unique when used for safe matching.

The fifth mistake is not reviewing results. After importing, you should compare counts, review errors, validate created or updated records, and reload only what failed. The sixth mistake is thinking deactivating rules is a cleanup strategy. Sometimes it is a controlled temporary step, but as an exam answer it is usually safer to adjust data, review criteria, and protect quality.

Quick decision checklist:

Is the need import, update, upsert, export, or delete?
How many records are involved, and which objects are in scope?
How are existing records identified: Salesforce ID, External ID, email, or name?
Which duplicate rules, validations, and permissions can block the load?
How will errors, counts, and quality be reviewed after import?

Official resources and practice

To practice, create a small CSV of leads with repeated emails, trailing spaces, countries written in several ways, and one empty required field. Import it first as a test, review mapping, observe errors, and correct the file. Then imagine those leads come from an external system: define an External ID, decide whether to use insert, update, or upsert, and explain which rules would prevent duplicates.

BlueForce is not affiliated with Salesforce. This guide is original editorial content based on study experience, public objectives, and official resources.