Salesforce CRM Analytics and Einstein Discovery Consultant

Nothing works until the right people are licensed and permitted. You enable the platform with licenses and permission sets, work within data sync, dataflow and recipe limits, and move assets between sandboxes and production.

• Licenses and permission sets
• Enabling the platform
• Limits that shape a design
• Packaging an app as a template
• Moving assets between environments
• Checking a deployment

You build A setup and deployment plan that moves an analytics app from sandbox to production.

Data must arrive before it can be analyzed. You use Data Manager to connect Salesforce and outside sources, choose which objects and fields to sync, and set refresh schedules that suit the business.

• Connections and data sync
• Choosing objects and fields
• Incremental and full sync
• Scheduling refreshes
• Monitoring runs and fixing failures
• Bringing in outside data

You build A scheduled data sync of three objects with a monitoring routine.

Raw data rarely matches a question. You build dataflows that extract, join, filter and compute new fields, register datasets, and read the dataflow’s history to find why a run failed.

• Extracting Salesforce data into a dataflow
• Augmenting and joining datasets
• Filtering and computing new fields
• Flattening hierarchies
• Registering a dataset
• Reading a failed run

You build A dataflow that joins accounts, opportunities and users into one dataset.

Recipes offer a visual way to prepare data. You build one to clean and combine sources, compare it with a dataflow for the same task, and choose which to use for each kind of problem.

• Building a recipe step by step
• Joins, appends and calculated fields
• Cleaning values and handling nulls
• Recipes compared with dataflows
• Scheduling and running a recipe
• Choosing between the two

You build A recipe that cleans and merges two sources, with a note on why it was chosen over a dataflow.

A dataset is easier to use when it reads well. You manage extended metadata to set labels, formats and defaults for fields, so people exploring the data see names and values that make sense to them.

• Dimensions, measures and dates
• Extended metadata for field labels and formats
• Default values and sort order
• Hiding fields people should not see
• Documenting a dataset

You build A dataset with clear field labels, formats and documentation.

People must see only the numbers they are entitled to. You apply user, group and profile settings, restrict rows with security predicates, use sharing inheritance where it fits, and share apps at the right level.

• Users, groups and profiles
• Row-level security with predicates
• Sharing inheritance from Salesforce
• Sharing apps at viewer, editor and manager level
• Testing security as another user
• Troubleshooting why numbers differ between users

You build A security design for a dataset, tested with three sample users.

A dashboard begins with a requirement, not a chart. You scope and validate what is needed, prioritize it with stakeholders, apply UX principles, and decide when a templated app already does most of the job.

• Gathering and validating requirements
• Prioritizing what goes on the page
• UX principles for analytics
• Wireframing before building
• When a templated app fits
• Reviewing a design with users

You build A dashboard requirements document and wireframe for one team.

Widgets turn a dataset into something readable. You configure queries and charts, choose the query type for the job, and format widgets so numbers are clear and consistent.

• Query types and when to use each
• Charts, tables and number widgets
• Measures, groupings and filters
• Formatting and consistency
• Using SAQL when the builder is not enough
• Testing a widget against the source

You build A dashboard with five widgets built on one dataset, each tested against source data.

Dashboards become useful when they respond. You build interactions between widgets, use compare tables to show change, and apply time series so people can see trends and period-over-period movement.

• Faceting and selection between widgets
• Using bindings to drive queries
• Compare tables and calculated columns
• Time series and period comparisons
• Moving between dashboards

You build An interactive dashboard with a compare table and a time series comparison.

A dashboard that is slow will not be used. You find and fix performance problems with Dashboard Inspector, then place dashboards on Lightning pages so people meet analytics where they already work.

• Reading Dashboard Inspector
• Reducing query cost
• Reducing the number of widgets and steps
• Adding a dashboard to a Lightning page
• Filtering by the record being viewed

You build A tuned dashboard placed on a record page, with before-and-after load times.

Einstein Discovery explains why things happen and predicts what comes next. You prepare a dataset for it, choose between numeric, binary and multi-class prediction, and build a story that produces a model.

• Choosing an outcome variable
• Numeric, binary and multi-class prediction
• Preparing a dataset for modeling
• Creating a story
• Reading what the story says
• Choosing which variables to include

You build A story and model predicting a business outcome from a prepared dataset.

A model is only as useful as it is trusted and used. You read the Model Card, improve the data when results are weak, and show predictions on Lightning pages so that decision makers can act on them.

• Reading the Model Card
• Spotting leakage and bias in a model
• Improving a model by improving the data
• Deploying a model
• Showing predictions on Lightning pages
• Monitoring a deployed model

You build A deployed model with predictions on a record page and a short model assessment.

Answer one question from start to finish. You bring in data, prepare and secure it, design the dashboard, add a prediction, and present the result to the people who asked.

• Framing the business question
• Preparing and securing the data
• Building the dashboard and model
• Placing results in Salesforce
• Presenting to stakeholders

You build An analytics solution, from data sync to dashboard and prediction, for one question.