Start with the commercial decision
A sales dashboard says the team has $330,000 of opportunities. A manager wants to know whether that supports next quarter's target. Before forecasting anything, you discover that one deal appears twice and another has no usable close date. This is the kind of bounded problem that makes revenue operations tangible: decide what the records mean, repair the reporting logic and explain the commercial consequence.
HubSpot's revenue-operations course treats the discipline as work across commercial functions, while Salesforce's data-quality learning material focuses on assessing and improving the information used in customer systems. These resources are useful preparation references. Neither certificate establishes internship eligibility or substitutes for an example of your own reasoning.
Search revenue operations intern, sales operations intern, go-to-market operations intern, CRM operations intern and deal operations intern. A role may sit in sales, marketing, finance or a central operations team. Do not infer its scope from the department alone. Read for the problem it owns: reporting, account assignment, systems administration, quoting, customer lifecycle processes or forecasting support.
Separate selling from operating the sales system
An internship that asks you to book meetings and contact prospects may be a sales-development role even when operations appears in the title. One that asks you to maintain reports, investigate records or document a handoff is a different assignment. Both can teach useful commercial skills, but prepare a work sample for the responsibilities actually listed.
Within operations, distinguish a reporting assignment from permission to change the system. Updating a spreadsheet prototype is not the same as altering a live CRM workflow, merging customer accounts or changing commission calculations. Ask which environment interns use and who approves changes. A good role should make ownership and review understandable, particularly when an error could affect a customer or employee.
Record the tools named in the posting without turning them into a shopping list of credentials. Spreadsheet fluency can support a small audit; SQL can help with larger datasets; a CRM training environment can demonstrate familiarity with objects and stages. The relevant question is whether you can explain the data model and the decision, not how many platform badges fit on your résumé.
Define the data before calculating pipeline
Use a fictional dataset with an opportunity ID, customer ID, amount, currency, stage, expected close date, owner and last-updated date. State whether one row means one opportunity, one product line or one activity. If the same opportunity legitimately has several product rows, summing by opportunity requires a different approach than simply deleting repeated IDs.
For the worked case, each row is supposed to represent one opportunity in US dollars. Deal A is $120,000 with a supplied weight of 50%; B is $80,000 at 25%; C is $50,000 at 80%; a second row repeats C; and D is $30,000 at 20% but lacks a close date. A, B and C have dates inside the target quarter. The repeated C row is an exact duplicate introduced during the fictional export.
Write these assumptions above the calculations. In real work, apparent duplicates can be legitimate renewals, expansions or regional records. A shared customer name is not enough evidence to merge them. Your sample should show how you would confirm identity and retain an audit trail rather than delete whatever looks similar and celebrate a cleaner table.
Reconcile the $330,000 headline
The raw opportunity amounts sum to $330,000. Removing the confirmed duplicate C gives $280,000 of unique opportunities. Excluding undated D from the target-quarter cohort leaves $250,000. That is a population decision, not proof that D has no value. Keep D in an unresolved queue and ask the owner for a supported date rather than forcing it into the quarter or quietly erasing it.
The supplied weighted amounts tell a parallel story. The raw total is $166,000: $60,000 for A, $20,000 for B, $40,000 for each C row and $6,000 for D. After deduplication it is $126,000. For the dated target-quarter cohort, it is $120,000. Show all three steps so the reader can reconcile the original headline to your result.
Do not call $120,000 an expected revenue forecast merely because each row has a percentage. The percentages were supplied assumptions in this exercise. To interpret them as probabilities, you would need evidence about how stages convert, the forecast horizon and whether the current deals resemble the historical population. Correlation between deals also matters for uncertainty around the total. A weighted pipeline can be a planning view without being a calibrated prediction.
Add checks that catch a wrong answer before presentation
Start with simple checks: IDs are present, amounts are numeric, currencies are consistent, stage values belong to an allowed list, and close dates are either valid or explicitly missing. Report a missing date separately from an out-of-quarter date. Then check whether a join duplicates opportunity rows, which can inflate amounts even when the original opportunity table has unique IDs.
Create an exceptions table that names the issue, affected record, proposed treatment and required approval. Include an unchanged original-data tab or file. Your corrected view should be reproducible from that input and documented rules. Do not overwrite the only copy of the source and leave a reviewer unable to see what changed.
For an extended project, add a snapshot date and compare two extracts. A deal moving from one quarter to another is not a newly created opportunity. A change in amount is not automatically a new sale. Explain changes with stable IDs and a small bridge between snapshots. You can demonstrate the concept with invented records; no customer database or paid software subscription is required.
Turn the audit into a decision memo
The memo should lead with the corrected question: the dated, deduplicated target-quarter pipeline is $250,000, with $120,000 of assumption-weighted pipeline and $30,000 of undated opportunities awaiting review. Explain the duplicate and date treatment in plain language. Then state what cannot yet be concluded, such as whether the sales target is achievable without validated conversion behavior and timing.
Recommend the next operational step. Assign the duplicate issue to the export or record owner and request a close-date review for D. A dashboard is not a remedy for a recurring identity problem. If the issue originates in a handoff between systems, document that handoff and propose a testable rule rather than merely adding a red cell to the report.
An accurate portfolio description might say that you built a fictional pipeline audit, reconciled five raw rows to three dated unique opportunities and documented exception handling. Do not describe it as improving a company's revenue forecast. In an interview, explain what would change if C were actually two legitimate product lines. The answer should begin by revisiting the row definition, not by defending your original cleanup rule.
Check whether the internship will teach the full workflow
Ask for a recent example of an intern project and who used its output. Clarify whether the role includes analysis and review, or mainly recurring data entry. Repetitive tasks can still teach the system if someone explains the decisions behind them; a vague promise of strategic exposure does not guarantee that explanation.
Ask which changes need approval, how access is provisioned and whether a sandbox is available. Also clarify whether forecast reporting, compensation operations and customer data are separate responsibilities. Do not request confidential exports as recruiting preparation. A synthetic model is enough to demonstrate the mechanics and your respect for access boundaries.
For a first application, a narrow auditable project is stronger than a sprawling dashboard with unexplained metrics. Make the data definition visible, show the reconciliation and give one sensible next action. Revenue operations rewards the ability to connect small record-level errors to the business question without pretending the cleaned number answers everything.
Choose one original practice task
Choose a task that matches a skill you want to demonstrate. Each original invented case has downloadable inputs, a separate answer key and a failure check. Build your own result, ask someone to reproduce it, then retain a dated revision. Describe it as independent work.