New to this track? Our Salesforce Certified Platform Administrator exam prep is the recommended first step. See our certification path to understand where this certification fits. The Salesforce Certified CRM Analytics and Einstein Discovery Consultant exam weights Data Layer most heavily (23%), see the full breakdown, study tips, and practice questions below.
Free practice • No sign-up required
Independent study resource
Exam objectives and policies can change. Confirm current details with the certification issuer.
Check current objectives
Use the official exam guide as the source for current objectives and section weights.
Practice for learning
Use sample questions to review concepts and identify topics for further study.
Works on Any Device
Full practice experience on phone, tablet, or desktop
Free, No Sign-Up
Start practicing immediately, no account or email required
Salesforce Certified CRM Analytics and Einstein Discovery Consultant – Complete 2026 Guide
Covers the full CRM Analytics build: getting data in with dataflows and recipes, designing lenses and dashboards, row-level security, and using Einstein Discovery to create stories, evaluate models and write predictions back onto Salesforce records. The exam is 60 scored questions in 90 minutes, with a 65% pass mark, weighted towards Data Layer (23%) and Analytics Dashboard Implementation (19%).
Recommended Prerequisites
We recommend completing this certification first to prepare you better.
| Official exam name and code | Salesforce Certified CRM Analytics and Einstein Discovery Consultant (Analytics & Discovery) |
|---|---|
| Questions | 60 |
| Time allotted | 90 min |
| Passing score | 65% |
| Exam fee | $200 |
| Retake fee | $100 |
| Prerequisite certifications | Recommended preparation: Salesforce Certified Platform Administrator |
| Study guide release | Summer '26 |
Salesforce Certified CRM Analytics and Einstein Discovery Consultant Exam Weightage by Section
1. Admin/Configuration (17%)
17%2. Data Layer (23%)
23%3. Security (16%)
16%4. Analytics Dashboard Design (13%)
13%5. Analytics Dashboard Implementation (19%)
19%6. Einstein Discovery (12%)
12%Exam Topics
Last updated:
Who is the CRM Analytics Einstein Discovery Consultant exam for?
- Analysts building in CRM Analytics: You already make dashboards and now need the data layer: recipes, dataflows, security predicates and how predictions get back into Salesforce.
- Admins asked to own analytics: Reports and dashboards have hit their limit and CRM Analytics is the next step. This covers what that involves.
- Consultants on analytics projects: You scope and deliver these builds and need to know where the real constraints are, particularly around row-level security.
Exam Fees & Registration
Exam Fee
$200
One-time registration fee
Retake Fee
$100
If you need to retake the exam
Comparing certs? View all Salesforce exam fees in one place →
Certification Validity
Maintenance and renewal requirements vary. Check the issuer’s current credential page for requirements that apply to you.
How to Register
Check current Salesforce Certified CRM Analytics and Einstein Discovery Consultant credential and registration details with the issuer.
View official credential detailsOfficial references: Use the issuer’s current materials to confirm exam objectives, format, and fees before booking.
Official sources: Salesforce exam pricing·Independent exam-prep resource; details may change.
Exam Tips
- 1All four sections are 25%. Cover CRM Analytics Setup, Data Prep, Lens/Dashboards, and Einstein Discovery.
- 2Know CRM Analytics setup: org setup, data sources, and connections.
- 3Understand data preparation and datasets: data transformation, recipes, and datasets.
- 4Be ready for "how do you configure X?" questions.
Prerequisites
- •CRM Analytics experience
- •Salesforce Administrator
Focus Areas
- •CRM Analytics Setup
- •Data Preparation and Datasets
- •Lens, Dashboards, and Stories
- •Einstein Discovery
Study Strategy
Use CRM Analytics.
Configure data sources, create datasets, build dashboards, and use Einstein Discovery.
Align with the exam outline.
Exam Format and First-Attempt Readiness
The Salesforce Certified CRM Analytics and Einstein Discovery Consultant exam has 60 questions in 90 min, passing score 65%. Most Salesforce exams test scenario-based decisions. Focus on when to use each feature, not just terms.
- Study Data Layer and Analytics Dashboard Implementation first, together they carry 42% of the exam.
- Do timed question sets. Build pacing and confidence.
- Review why wrong answers are wrong. It improves scenario reasoning.
- Book the exam once your mock scores are steady at 75%+.
Is the CRM Analytics Einstein Discovery Consultant Exam Hard?
Is the CRM Analytics Einstein Discovery Consultant Exam Hard?
Moderate, with a hard edge in two places. Row-level security through predicates and sharing inheritance is where candidates lose most marks, and Einstein Discovery model quality, bias checking and deployment is the part people have least hands-on experience with. The dashboard and dataset questions are fair.
- 60 questions in 105 minutes. Some questions show a predicate or a recipe step and ask what it does.
- Around 65% to pass.
- Security is the hard section. Security predicates, inherited sharing, and what each one can and cannot express.
- Einstein Discovery is half the exam's name. Story creation, model metrics, bias, and writing predictions back to records.
Pass Rate Guidance
Two things tell you whether you are ready. Can you write a security predicate for a requirement such as managers see their team's rows, and say when inherited sharing is the better answer? And can you look at a story's model and say whether it is good enough to deploy, and why? If either is shaky, that is where the marks are.
CRM Analytics Einstein Discovery Consultant Exam Format Explained
CRM Analytics Einstein Discovery Consultant Exam Format Explained
Proctored online or at a test centre. The shape of it:
Total Questions
60 scored questions
Time Limit
90 min
Passing Score
65%
Question Type
Multiple-choice & multiple-select
How Many Questions Are Scenario-Based?
Most questions are a requirement followed by four approaches, often with more than one that would work. The distinguishing factor is usually maintainability: a recipe over a dataflow, inherited sharing over a hand-written predicate, standard embedding over custom code.
Best Way to Pass on Your First Attempt
- Build a recipe, not just a dashboard. Load data, join it, aggregate it, output a dataset. The transformation questions become obvious once you have done this.
- Practise security predicates until the syntax is automatic. They are the most commonly failed topic and they are learnable in an afternoon.
- Create a story end to end. Pick a dataset, let Discovery build a model. Read the metrics, then deploy the prediction to an object. Doing it once covers a surprising amount of ground.
- Know when SAQL is actually needed. Most dashboard work does not require it, and the exam tests whether you know where the line is as much as it tests the syntax.
CRM Analytics & Einstein Discovery: Key Concepts
Datasets, Dataflows and Recipes
Everything in CRM Analytics queries a dataset, which is a columnar snapshot built on a schedule rather than a live view of Salesforce data. You get data in through two tools. Dataflows are the older JSON-defined approach, built from nodes such as sfdcDigest to pull an object, augment to join, flatten to resolve a role hierarchy, and sfdcRegister to output. Recipes are the newer visual tool, built on Data Prep, and they can do everything a dataflow can plus transformations that are painful in JSON: clustering, prediction, formula columns, appends and joins with a preview at every step. For new work, recipes are the expected answer, and the exam leans that way. Two dataflow nodes keep appearing in questions regardless: augment, because joining is so common, and flatten, because it is the only way to turn a hierarchy into something you can filter on.
Lenses, Dashboards and Bindings
A lens is a single exploration of one dataset. A dashboard assembles widgets, and the thing that makes dashboards hard is bindings: the expressions that let one widget drive another. Selection bindings take what the user has clicked and feed it into another query. Results bindings take the output of one query and use it in another, which is how you build a widget whose values depend on a calculation elsewhere. Pages let you split a dashboard into views without building separate dashboards. The design guidance the exam rewards is unexciting and consistent: filter at the dataset or query level rather than fetching everything and hiding it, limit the number of queries per dashboard because each is a round trip, and use faceting so related widgets filter together by default.
Einstein Discovery Stories and Models
A story is what Discovery produces when you point it at a dataset and name an outcome variable. It analyses the drivers of that outcome and produces insights in four flavours: what happened, why it happened, what will happen, and what to do about it. Behind the insights sits a model, and the exam expects you to be able to judge one. For a numeric outcome you look at R squared and mean absolute error; for a binary outcome you look at the confusion matrix and measures derived from it. The quality checks matter as much as the metrics. Discovery flags variables that leak the answer, such as a field that is only populated after the outcome is known, and it supports bias detection on sensitive variables so you can see whether the model treats groups differently. A model with strong metrics and a leaking variable is worse than useless, and questions test that you would reject it.
Deploying Predictions Back to Salesforce
A model only earns its keep when it reaches the people doing the work. Discovery deploys a prediction definition against a Salesforce object, which puts a prediction, the top drivers behind it, and suggested improvements onto the record, visible through the Einstein Predictions component. Predictions can be scored in real time as a user edits a record, or written in bulk to existing records. They can also be consumed in Flow, which is how you act on a score automatically rather than leaving it on screen. The points that get examined are the mapping step, where the model's variables must be matched to fields on the target object, and the distinction between scoring on demand and writing scores back to the records in bulk.
Row-Level Security
This is the section that decides results. A dataset is a flat snapshot and does not inherit Salesforce record access by default, so when a dataset is registered it is visible in full to anyone who can see it. There are two ways to restrict it. A security predicate is a filter expression stored on the dataset and applied to every query, written against dataset columns and user attributes, for example 'OwnerId' == "$User.Id". Predicates can express hierarchies by combining a flattened role column with the running user's role. The alternative is inherited sharing, where the dataset is configured to respect the sharing of the underlying Salesforce object, which is simpler to maintain and the right answer when the requirement is just mirror Salesforce sharing. Inherited sharing has limits on the number of records it can handle, which is exactly the trade-off questions probe: use inheritance for straightforward sharing on moderate volumes. Write a predicate when the rule is custom or the volume is large.
A Story trains a predictive model against a dataset, surfaces narrative insights and key predictors, and once deployed can be embedded via a prediction field or Prompt Template so Salesforce records get real-time predicted outcomes.
How to Pass the CRM Analytics and Einstein Discovery Consultant Exam
This exam rewards having built something. The dashboard questions are forgiving; the security and model-quality questions are not.
Recipes Over Dataflows
Recipes are the current tool and the expected answer for new work. Know what they add: formula columns, clustering, prediction, visual previews.
Predicate Syntax
Practise writing predicates for real requirements, including owner-based and role-hierarchy cases. The most commonly failed topic and the most quickly fixed.
Inherited Sharing Trade-off
Simpler to maintain, limited by record volume. Know when to choose it over a predicate and be ready to justify either.
Judging a Model
R squared and mean absolute error for numeric outcomes, the confusion matrix for binary ones. Watch for leaking variables and check bias on sensitive fields.
Bindings
Selection bindings respond to clicks; results bindings use one query's output in another. Build one of each rather than reading about them.
Is the Salesforce Certified CRM Analytics and Einstein Discovery Consultant Exam Hard?
DifficultyModerate; plan about 4-8 weeks of focused prep.
Book When ReadyScore 75%+ on three timed mocks before scheduling.
Use Practice, Not DumpsOriginal practice questions build skill without risking your credential.
Exam format: 60 questions, 90 min, 65% passing score.
Salesforce Certified CRM Analytics and Einstein Discovery Consultant Practice Questions (With Explanations)
Test your knowledge with these 15 sample questions.
Question 1: A dataset must show each user only the records they own. Which approach expresses this most directly?
Key takeaway: Security predicates are applied to every query against the dataset and are designed for exactly this rule.
Show correct answer and explanation
Correct answer: C. A predicate comparing owner to the running user
Security predicates are applied to every query against the dataset and are designed for exactly this rule.
- A dashboard filter can be removed by the user and is not a security control.
- Hiding the dataset blocks legitimate access rather than filtering rows.
- Field-level security on the source object does not restrict rows in a registered dataset.
Question 2: A requirement is simply that the dataset should respect the sharing already configured on the Salesforce object, and record volume is moderate. What is the best choice?
Key takeaway: Inherited sharing mirrors Salesforce sharing without you maintaining a parallel rule set, which is why it suits this case.
Show correct answer and explanation
Correct answer: A. Inherited sharing on the dataset
Inherited sharing mirrors Salesforce sharing without you maintaining a parallel rule set, which is why it suits this case.
- App-level permissions control who sees the dataset, not which rows.
- Duplicating datasets per role multiplies maintenance and refresh cost.
- Replicating sharing in a predicate is more work to build and to keep correct.
Question 3: An Einstein Discovery model shows excellent accuracy, but one of its variables is only populated after the outcome has been decided. What should happen?
Key takeaway: A variable known only after the outcome inflates accuracy in training and is useless at prediction time.
Show correct answer and explanation
Correct answer: B. Remove that variable and retrain
A variable known only after the outcome inflates accuracy in training and is useless at prediction time.
- High accuracy caused by leakage will not survive contact with real data.
- Changing the field type does not change when the value becomes known.
- Documenting a broken model does not make it safe to deploy.
Question 4: Which dataflow node joins columns from one dataset onto another?
Key takeaway: Augment is the join node, bringing fields from a right dataset onto a left one on a matching key.
Show correct answer and explanation
Correct answer: A. the augment join node
Augment is the join node, bringing fields from a right dataset onto a left one on a matching key.
- sfdcDigest extracts data from a Salesforce object.
- sfdcRegister writes the result out as a dataset.
- Flatten resolves a hierarchy into a column of ancestor values.
Question 5: A dashboard must show a chart whose values depend on the result of a separate query rather than on a user click. What is needed?
Key takeaway: Results bindings feed the output of one query into another, independent of user interaction.
Show correct answer and explanation
Correct answer: A. A results binding
Results bindings feed the output of one query into another, independent of user interaction.
- Predicates are a security mechanism.
- A new dataset does not create the dependency between widgets.
- Selection bindings respond to what the user has clicked.
Question 6: For new data preparation work in CRM Analytics, which tool is the expected choice?
Key takeaway: Recipes are the current tool and cover everything dataflows do plus transformations that are difficult in JSON.
Show correct answer and explanation
Correct answer: D. A recipe in Data Prep
Recipes are the current tool and cover everything dataflows do plus transformations that are difficult in JSON.
- Report snapshots are a core Salesforce feature, unrelated to CRM Analytics datasets.
- Apex is not how CRM Analytics datasets are built.
- Dataflows remain supported but are no longer the recommended starting point.
Question 7: A deployed prediction should trigger an automated follow-up task when the score falls below a threshold. How is this best done?
Key takeaway: Predictions can be consumed in Flow, which is how a score drives automated action.
Show correct answer and explanation
Correct answer: C. Consume the prediction in Flow
Predictions can be consumed in Flow, which is how a score drives automated action.
- Manual action does not satisfy an automation requirement.
- Dashboards display information and cannot create records.
- Changing the model does not create the follow-up task.
Question 8: Which dataflow node is required to filter on a role hierarchy within a dataset?
Key takeaway: Flatten converts a hierarchy into a multi-value column of ancestors, which a predicate or filter can then use.
Show correct answer and explanation
Correct answer: A. flatten
Flatten converts a hierarchy into a multi-value column of ancestors, which a predicate or filter can then use.
- Augment joins datasets but does not resolve hierarchy paths.
- dim2mea converts a dimension to a measure.
- Slice selects or removes fields.
Question 9: Which metrics are appropriate for judging a model that predicts a numeric outcome such as deal amount?
Key takeaway: Numeric outcomes are evaluated with regression measures such as R squared and mean absolute error.
Show correct answer and explanation
Correct answer: C. R squared and mean absolute error
Numeric outcomes are evaluated with regression measures such as R squared and mean absolute error.
- Precision and recall are classification measures.
- Confusion matrices describe classification outcomes.
- Insight count says nothing about model quality.
Question 10: Why does a CRM Analytics dashboard get slower as more widgets are added?
Key takeaway: Query count is the main driver of dashboard performance, which is why consolidating and faceting helps.
Show correct answer and explanation
Correct answer: B. Each query is a separate round trip against the dataset
Query count is the main driver of dashboard performance, which is why consolidating and faceting helps.
- Rendering is not the bottleneck; querying is.
- Datasets are fixed snapshots and do not grow when widgets are added.
- Predicate evaluation is not the cause of the scaling problem described.
Question 11: A sensitive variable such as gender is present in a story's dataset. What should be done before deploying?
Key takeaway: Discovery supports bias checking on sensitive variables, and checking is the responsible step before deployment.
Show correct answer and explanation
Correct answer: C. Run bias detection across the groups
Discovery supports bias checking on sensitive variables, and checking is the responsible step before deployment.
- Models reproduce patterns in their training data, including unfair ones.
- Deleting the dataset discards work that may be legitimate once checked.
- Changing the field type does not address the fairness question.
Question 12: What is the relationship between a dataset and live Salesforce data?
Key takeaway: Datasets are built and refreshed by dataflows or recipes, so they reflect data as of the last run.
Show correct answer and explanation
Correct answer: B. A dataset is a snapshot refreshed on a schedule
Datasets are built and refreshed by dataflows or recipes, so they reflect data as of the last run.
- Report types are a core reporting concept.
- Real-time querying is not how CRM Analytics datasets work.
- There is no live view; the data is copied in.
Question 13: A prediction definition is being deployed to the Opportunity object. Which step is essential?
Key takeaway: Variable to field mapping is what lets the model score live records on the target object.
Show correct answer and explanation
Correct answer: B. Mapping model variables to object fields
Variable to field mapping is what lets the model score live records on the target object.
- Bindings are a dashboard construct.
- Predicates secure dataset rows and are not part of prediction mapping.
- A dataflow is not required to deploy an existing model.
Question 14: When is writing SAQL by hand the appropriate choice?
Key takeaway: Most dashboard work needs no SAQL, so hand-written queries are for the cases the builder cannot express.
Show correct answer and explanation
Correct answer: D. When the builder cannot express it
Most dashboard work needs no SAQL, so hand-written queries are for the cases the builder cannot express.
- Routine widgets are better built with the standard tooling and are easier to maintain.
- SAQL queries data; recipes prepare it.
- Hand-written SAQL is not automatically faster.
Question 15: A manager must see rows belonging to everyone beneath them in the role hierarchy. Which combination achieves this?
Key takeaway: Flattening produces the ancestor path, and the predicate then compares it against the running user.
Show correct answer and explanation
Correct answer: D. A flattened role column plus a predicate
Flattening produces the ancestor path, and the predicate then compares it against the running user.
- Field-level security does not filter rows.
- Dashboard filters are not a security control.
- Building a dashboard per manager does not scale and leaves the data unsecured.
Get the Full Question Bank
Most candidates book the exam after scoring 75%+ on full mocks.
If you’re planning to test this quarter, aim to complete full mocks at least 10–14 days before your exam date.
Contact us for more information on our full 60-question mock exams and a personalized study plan. Use the contact form.
Request Mock Exams & Study PlanSalesforce Certified CRM Analytics and Einstein Discovery Consultant Exam FAQs
- What is covered on the Salesforce Certified CRM Analytics and Einstein Discovery Consultant exam?
- The Salesforce Certified CRM Analytics and Einstein Discovery Consultant exam (formerly Salesforce Certified CRM Analytics & Einstein Discovery Consultant) covers section-wise weightage as shown above. Use the exam topics and practice questions on this page to align your study with the official outline.
- What is CRM Analytics (formerly Tableau CRM)?
- CRM Analytics is Salesforce's analytics platform that provides AI-powered insights, dashboards, and data visualization directly within Salesforce. It includes Einstein Discovery for predictive analytics.
- Do I need Tableau experience for CRM Analytics Consultant?
- No, CRM Analytics is different from Tableau. However, understanding data visualization, analytics concepts, and having Salesforce Administrator knowledge is helpful.
- Are there free practice questions for the Salesforce Certified CRM Analytics and Einstein Discovery Consultant exam?
- Yes. This page includes free sample practice questions with explanations. Use them alongside the current official documentation.
- How do I prepare for the Salesforce Certified CRM Analytics and Einstein Discovery Consultant certification?
- Use the exam tips, prerequisites, and study strategy on this Salesforce Certified CRM Analytics and Einstein Discovery Consultant study guide (formerly Salesforce Certified CRM Analytics & Einstein Discovery Consultant). Focus first on the highest-weighted sections, then take the sample practice questions. Schedule the exam when you consistently score well on practice tests.
Next Consultant Certifications
Where analytics work usually leads next:
Honest Cert Breakdown
Difficulty: Moderate
3/5Tableau CRM / Einstein Analytics platform. Dashboard design, SAQL queries, and dataflow configuration. Easier with hands-on app experience.
Salary Range
$90,000–$125,000 / year
US average for certified professionals • Updated Summer '26 • Source: Salesforce Talent Ecosystem
Is it worth it?
Good add-on for analytics-focused admins and consultants. Standalone value is moderate, strongest when paired with a functional cert.
Weightings by Section
What Each Domain Covers
Data Integration
Dataflows: transformations (augment, computeExpression, flatten, filter, sfdcDigest, sfdcRegister), scheduling, output datasets. Recipes: Data Prep UI, node types (input, output, formula, bucket, join, filter, aggregate), recipe scheduling. External data upload: CSV connector, SFTP. Connected datasets: live Salesforce data without dataflow. When to use dataflows vs recipes.
Dashboards & Lenses
Dashboard designer: widget types (chart, table, number, filter, date), binding types (static, column, measure). SAQL in dashboards: inline queries. Lenses: exploratory analysis on a single dataset. Dashboard JSON editor for advanced customisation. Faceting: how widgets filter each other. Global filters and local filters. Mobile dashboard considerations.
Analytics Projects & Solutions
Analytics templates: pre-built app templates for specific use cases (Sales Analytics, Service Analytics, CRM Analytics for Financial Services). Customising templates. Creating apps from scratch vs from templates. Analytics Studio navigation: apps, datasets, dashboards, lenses, recipes. Embedded analytics in Salesforce pages using Wave dashboard component or Analytics tab.
Analytics Ecosystem & Admin
Platform overview: Analytics Studio, Data Manager, Data Integration Studio. Dataset limits and row counts. Scheduled dataflow and recipe jobs: run order, failure handling, dependency. Analytics notifications: alert users when metric thresholds are crossed. Analytics home page, navigation, and user experience configuration.
Security & Row-Level Security
Dataset-level security: private vs shared datasets. Row-level security (RLS): predicate logic, security predicates on datasets to filter records per user. Sharing inheritance from Salesforce: replicating record-level security in analytics datasets using role hierarchy sharing. User attributes: dynamic predicates based on user profile fields. Dataset sharing with groups and roles.
SAQL & Recipes
SAQL (Salesforce Analytics Query Language): q() function, groupby, foreach, order, limit, offset, cogroup for joins. SAQL in lenses vs in dashboard queries. Recipes formula nodes: date functions, string manipulation, conditional logic, bucket expressions. Common recipe transformations for data cleansing and enrichment.
8-Week Study Plan
Working Through Scenario Questions
- 1.Dataflows vs recipes: For complex transformations requiring custom JSON logic. Use a dataflow. For visual, maintainable ETL without coding. Use a recipe. When both could work, the exam typically favours recipes as the modern recommended approach.
- 2.RLS predicates must be applied to every dataset: Row-level security does not cascade. If a user joins two datasets in a dashboard query, the predicate must be applied to both datasets separately. Forgetting this is a common exam mistake.
- 3.Faceting vs filtering: Faceting allows widgets to filter other widgets on the same dashboard when a user makes a selection. It is enabled per-widget in the dashboard designer. Global filters apply to all widgets simultaneously. Know which to use for a given requirement.
- 4.Connected datasets for live data: If the exam describes a need for real-time or very fresh data without a dataflow/recipe delay, connected datasets (live query to Salesforce) is the answer, at the cost of slower query performance than indexed datasets.
Practice-Test Benchmark
Aim for 75%+ on practice exams before scheduling. CRM Analytics is very tool-specific, knowing the exact transformation nodes in a dataflow (sfdcDigest, augment, computeExpression, flatten) and SAQL syntax is tested directly. There is no shortcut other than hands-on practice in the tool.
10 Concepts Worth Reviewing
- Dataflow transformation nodes: sfdcDigest, augment, computeExpression, flatten, filter, sfdcRegister
- Recipes vs dataflows: when to use each, how they differ in architecture
- SAQL: q(), groupby, foreach, cogroup, order, limit
- Dashboard widgets: chart, table, number, filter, date, and configuration options
- Binding types: static, column, measure, how they link widgets to queries
- Faceting: enabling between widgets, how user selection propagates
- Row-level security predicates: syntax, user attributes, testing RLS
- Sharing inheritance: replicating Salesforce record access in analytics datasets
- Connected datasets: live query to Salesforce, performance trade-offs
- Analytics templates: structure, customisation, app creation from templates
Questions Candidates Ask
- What is Salesforce CRM Analytics?
- CRM Analytics (formerly Tableau CRM, formerly Einstein Analytics) is Salesforce's native business intelligence and analytics platform. It allows users to connect Salesforce and external data, build interactive dashboards, and surface AI-powered insights within the Salesforce UI. The CRM Analytics and Einstein Discovery Consultant certification validates skills in designing, building, and securing analytics solutions. The exam has 60 questions, 105-minute time limit, 65% passing score, and a $200 fee.
- What is the difference between a dataflow and a recipe in CRM Analytics?
- Dataflows are the original ETL mechanism in CRM Analytics, they use JSON-based transformation nodes (augment, filter, computeExpression, flatten, sfdcDigest, sfdcRegister) and run on a scheduled job. Recipes are the newer visual ETL builder, a drag-and-drop Data Prep UI with node types for joining, filtering, aggregating, and transforming data. Recipes are easier to build and maintain for most use cases. Dataflows offer more control for complex transformations and are still used in many legacy implementations.
- What is SAQL in CRM Analytics?
- SAQL (Salesforce Analytics Query Language) is the query language used by CRM Analytics to query datasets. It resembles SQL but is designed for CRM Analytics datasets rather than relational databases. Key functions: q() wraps the dataset reference, groupby groups results, foreach iterates over groups, cogroup joins two datasets. SAQL is used in lens and dashboard queries when the visual designer is insufficient for complex calculations.
- What is row-level security in CRM Analytics?
- Row-level security (RLS) in CRM Analytics controls which rows of a dataset a user can see when they view a dashboard or lens. It is implemented via security predicates, filter expressions evaluated per user based on their attributes (role, profile, user fields). For example, a predicate like 'Region' == "$User.Region__c" would limit each user to rows matching their assigned region. Without RLS, all users with dataset access see all rows regardless of their Salesforce record-level access.
- How long should I study for the CRM Analytics exam?
- Plan for 8–10 weeks with 10–12 hours per week. CRM Analytics requires hands-on experience, many exam questions test specific configuration steps in Analytics Studio that you cannot learn from reading alone. Use a free CRM Analytics developer org (available via Trailhead Playground) to build real dataflows, recipes, and dashboards. The Data Integration section (23%) is the most heavily weighted and requires practical dataflow/recipe knowledge.
Topic Difficulty Heatmap
Which topics are a gimme vs which ones trap confident candidates. Use this to prioritise your final-week revision. These are our own topic groupings for study planning, not the official exam sections, those and their weightings are in the credential's exam guide.
| Topic | Difficulty | Study Tip |
|---|---|---|
| CRM Analytics Setup | Moderate | Dataset creation and data prep. Know the difference between recipe and lens. |
| Data Preparation and Datasets | Hard | SAQL and dataflow, syntax and transformation order are common failure points. |
| Lens, Dashboards, and Stories | Trap ⚠ | Lens vs dashboard vs story, and when to use Explorer, exam tests these distinctions. |
| Einstein Discovery | Moderate | Story creation and prediction, outcome vs predictor and model interpretation. |
Difficulty based on analysis of common candidate errors across each exam section.
Try It Yourself
Free CRM Analytics practice questions covering dataflows, recipes, SAQL, and dashboard design.
Start Free Practice QuestionsRelated: Tableau Data Analyst study guide
At a Glance: What CRM Analytics Consultant Covers
- Dashboard design: Building lenses and dashboards with steps (query), widgets (charts, tables, numbers), filters, and bindings. Understanding the CRM Analytics dashboard designer, how steps connect to widgets, and how to use faceting for cross-filter interactions.
- Data integration and preparation: Dataflows for connecting Salesforce data to analytics datasets, Recipes for more visual data preparation, and the augment/append/join transformation types. Row-level security (predicates) for controlling which data users can see in dashboards.
- SAQL and Einstein Discovery: Reading and writing SAQL queries for calculated fields and custom step logic. Setting up Einstein Discovery models, understanding prediction scores and key influencer output, and writing back predictions to Salesforce records.
Where the Exam Weight Falls
Dashboards + Data Integration + Security = 68%. Row-level security and dataflow design are the most technically complex areas.
Scenario Strategy: How to Approach CRM Analytics Consultant Questions
Questions describe an analytics requirement and ask which dashboard component, dataflow transformation, security approach, or SAQL expression addresses it. Trace the question from the data requirement back to the CRM Analytics component that delivers it.
- For security questions: row-level security in CRM Analytics uses security predicates: SAQL conditions applied to every dataset query that restrict which rows a user can see. Predicates reference user attributes (user ID, role, profile). For dynamic security based on user hierarchy. Use the row-level security dataset with a join in the predicate.
- For dataflow vs. recipe questions: Dataflows are JSON-configured ETL processes (more flexible, code-like configuration). Recipes are visually configured in a UI (more accessible for non-developers). For complex multi-dataset joins, Dataflows may be required; for standard preparation tasks, Recipes are preferred as best practice.
- For SAQL questions: read the question’s output requirement first. SAQL group by determines the grain; foreach determines which measures to compute; order by and limit control the result set. When a step needs a calculated field (e.g., win rate = closed won / total), it requires a custom SAQL step, not a standard step type.
Readiness Check Before You Book
75%+ on 3 timed full mocks before booking
Build at least 3 complete CRM Analytics dashboards with dataflows, row-level security, and custom SAQL steps before booking. The SAQL and security predicate questions are very difficult to answer correctly without hands-on experience, theoretical study alone is insufficient.
3 Concepts That Fail Most CRM Analytics Candidates
These are not the hardest topics; they are the ones where candidates are most confidently wrong. Learn the distinction early.
1. Dataflows vs Recipes: Two Different Data Preparation Tools
Dataflows are the legacy CRM Analytics data preparation tool that runs on a schedule and outputs datasets. Recipes are the newer drag-and-drop interface for transforming data into datasets. New implementations should use Recipes. Candidates use Dataflow JSON syntax for Recipes or vice versa. Know that Recipes replaced Dataflows for most use cases; Dataflows are still used for specific advanced transformations.
2. XMD (Extended Metadata): Formatting Is Not Stored in the Dataset
XMD controls display properties: field labels, formatting (currency, percent), colours, and display order in lenses and dashboards. It is separate from the dataset schema. Candidates modify dataset schemas to change display formatting, the exam expects XMD modifications for presentation changes. Changing a field’s label in XMD does not change the underlying field name in SAQL.
3. SAQL vs Equals Filters: When to Use Each in Step Configuration
Equals Filters (the UI configuration) handle simple value comparisons in dashboard steps. SAQL (Salesforce Analytics Query Language) is needed for complex logic: date arithmetic, string manipulation, conditional aggregation. Candidates write SAQL for simple equality filters, the exam expects Equals Filters for those and SAQL only when the logic cannot be expressed in the UI filter interface.
Common Questions
- What is the Salesforce CRM Analytics Consultant exam format?
- The CRM Analytics and Einstein Discovery Consultant exam has 60 multiple-choice questions, a 105-minute time limit, a 65% passing score, and a $200 fee ($100 retake). It tests implementation of CRM Analytics (formerly Tableau CRM / Einstein Analytics): dashboards, lenses, dataflows, recipes, SAQL queries, and Einstein Discovery predictive analytics.
- What are the highest-weight CRM Analytics Consultant exam sections?
- Building and Customising Dashboards (28%) and Data Integration (22%) together account for 50% of the exam. Creating dashboards with steps and widgets, writing SAQL queries for custom calculations, configuring dataflows and recipes to prepare data, and setting up row-level security are the most heavily tested areas.
- What is SAQL and how important is it for this exam?
- SAQL (Salesforce Analytics Query Language) is CRM Analytics’ proprietary query language for dashboard steps. It is similar to SQL but has unique CRM Analytics syntax for groupby, order, limit, and custom calculations. The exam includes questions where you must read SAQL and identify what it returns, or identify the correct SAQL for a given data requirement.
- What is Einstein Discovery and what does the exam test about it?
- Einstein Discovery is CRM Analytics’ automated machine learning feature that finds patterns in data and generates predictions and recommendations. The exam tests how to set up a story (now called a model), interpret key influencers, understand prediction scores, and embed Einstein Discovery predictions back into Salesforce records using writeback.
- What concepts do most CRM Analytics candidates get wrong?
- The most commonly misunderstood topics for the CRM Analytics exam are: (1) Dataflows vs Recipes: Two Different Data Preparation Tools; (2) XMD (Extended Metadata): Formatting Is Not Stored in the Dataset; (3) SAQL vs Equals Filters: When to Use Each in Step Configuration. Candidates are most confidently wrong on these. Learn the distinctions early to avoid losing marks on questions you expect to get right.
- Why do most Crm Analytics candidates fail questions about Dataflows vs Recipes?
- Dataflows are the legacy CRM Analytics data preparation tool that runs on a schedule and outputs datasets. Recipes are the newer drag-and-drop interface for transforming data into datasets. New implementations should use Recipes. Candidates use Dataflow JSON syntax for Recipes or vice versa. Know that Recipes replaced Dataflows for most use cases; Dataflows are still used for specific advanced...
- Why do most Crm Analytics candidates fail questions about XMD (Extended Metadata)?
- XMD controls display properties: field labels, formatting (currency, percent), colours, and display order in lenses and dashboards. It is separate from the dataset schema. Candidates modify dataset schemas to change display formatting, the exam expects XMD modifications for presentation changes. Changing a field's label in XMD does not change the underlying field name in SAQL.
- Why do most Crm Analytics candidates fail questions about SAQL vs Equals Filters?
- Equals Filters (the UI configuration) handle simple value comparisons in dashboard steps. SAQL (Salesforce Analytics Query Language) is needed for complex logic: date arithmetic, string manipulation, conditional aggregation. Candidates write SAQL for simple equality filters, the exam expects Equals Filters for those and SAQL only when the logic cannot be expressed in the UI filter interface.
More Exam-Prep Guides
Start CRM Analytics Prep
After this exam, consider Sales Cloud Consultant or Service Cloud Consultant next.
Share or cite this study guide
Free to link from a forum post, study repo or article. Copy the Markdown and paste it anywhere.