Turn a posting into an evidence requirement
A project should answer a question that resembles the work in the posting. When a role asks for pricing analysis, a stock pitch may show financial interest but leave the main task unproved. When it asks for customer operations, a highly polished valuation model may be less relevant than a clean account-cohort analysis and a usable handoff. The first decision is not which tool to use. It is which contribution you want the employer to understand.
| Posting asks for | Useful evidence | What a polished file alone does not prove |
|---|---|---|
| Analyze performance and economic drivers | A reconciled result plus an operating interpretation | Causal explanations or forecast accuracy |
| Improve customer and renewal operations | An account-level cohort bridge and an actionable exception | That you managed real customers or saved a renewal |
| Prepare earnings materials | A source-defined profit bridge and a concise explanation | Permission to represent an issuer or access to private information |
| Research current and prospective investors | A clear audience brief and careful communication | Investment performance or real client-relationship experience |
| Use SQL and quantitative models | A reproducible query and tested calculation | SQL proficiency demonstrated only through a spreadsheet |
Take three responsibilities from the posting. Beside each, write the output the team would need, the skill it tests and one piece of work you can show. Use existing coursework, employment or volunteer work when it is relevant and safe to share. An additional project is optional; it should fill a real evidence gap rather than delay a prepared application.
The five examples below were checked on September 16, 2026. They are dated readings of exact employer pages, not a promise that a vacancy remains open. The project matches are our preparation suggestions, not employer assignments or endorsements. Keep the actual requisition as the source of application requirements.
Caterpillar pricing: show the decision behind a price change
Caterpillar’s Pricing internship, R0000380150, describes price recommendations using sales, competition and product-value information, with cross-functional analysis and pricing execution. Its Summer 2027 session runs May 17–August 6, onsite for forty hours weekly in Peoria, Mossville or Cary. The page lists October 12, 2026 as the apply-by date, a $22.50–$40.25 hourly range, resume plus unofficial transcripts, and no visa sponsorship.
The stated academic screen covers listed business or quantitative bachelor’s/master’s programs, a 3.0 major and cumulative GPA, and twelve completed semester hours; transfer students need twelve hours at a four-year university. Intermediate Excel appears under stronger-candidate attributes. Keep those requirements separate from the project: good analysis does not waive them.
Use the price–volume–mix project. First explain why contribution falls even though revenue rises. Then add a proposed action with a break-even calculation. The work sample demonstrates an arithmetic bridge and decision reasoning; it does not establish competitive price knowledge or real customer willingness to pay. Before submitting, adapt the memo’s business question to the role without inserting imaginary Caterpillar product data.
Sources[1]
Robinhood business analysis: a spreadsheet does not satisfy a SQL claim
Robinhood’s Summer 2027 Business Analyst posting describes performance analysis, pricing, segmentation and risk–return questions. It specifies strong SQL and quantitative modeling, a relevant bachelor’s degree with Spring/Summer 2028 graduation, and at least three office days weekly in Washington, DC. The twelve-week role lists $33 an hour for the location. The retrieved page does not give a fixed application deadline.
A price–volume–mix or partner-pilot analysis can provide a relevant decision problem, but the supplied spreadsheet is not evidence of SQL proficiency. To address that requirement, load the synthetic inputs into a local database, write the aggregation and reconciliation queries yourself, and retain checks of the result. Explain the unit of analysis before discussing a dashboard.
An application should identify the task you completed and the limits. Do not claim to have analyzed Robinhood customers, optimized its credit products or produced a real growth result. Likewise, questions about sponsorship on an application form do not establish that sponsorship is offered. Use the posting for the screen and your actual finished project for the evidence.
Sources[2]
Sprinto revenue operations: demonstrate a system someone can maintain
Sprinto’s Revenue & Operations Intern posting spans customer success, renewals, sales, finance and product. It names customer trends, retention, workflows, documentation and dashboards. Applicants may be pursuing or have recently completed a relevant degree; spreadsheet comfort and interest in tools such as SQL are stated. The page is labeled Bengaluru and remote, without a numeric pay figure or fixed application cutoff in the retrieved text.
Use the SaaS-retention project when the application stresses renewals and customer performance. Show the account-level bridge, including how a replacement subscription is treated and why total recurring revenue can rise while the original cohort shrinks. Use the sales-pipeline project when the emphasis is reporting or cross-team records. In either case, write a handoff that explains how the next analyst could repeat the work.
Do not translate “remote” into permission to work from every country or assume a U.S. compensation package. The practical question belongs to the employer. For the sample, the improvement you can honestly demonstrate is reproducibility: another reader can select the same population and reach the same result. It is not a measured retention increase at Sprinto.
Sources[3]
Amrize investor relations: explain a difficult quarter without spinning it
Amrize’s Summer 2027 Investor Relations internship lists earnings releases, scripts, presentations, quarterly Q&A and peer research. The Chicago program is ten weeks beginning June 1, 2027, with forty onsite hours a week and a published $20–$26 hourly range. It seeks enrolled undergraduates in finance, accounting, business, economics or a similar field. No explicit application cutoff was found in the retrieved body.
The earnings-analysis project is a direct preparation match for the described communication work. Produce the segment-level profit bridge and explain why reported and management-adjusted profit point in different directions. Give the reader the starting measure, adjustments and ending measure. The project should make the quarter easier to understand, not make its weak result disappear.
Add three questions you would raise before drafting a real earnings document: which cost change is recurring, whether the segment definitions are consistent, and what evidence supports management’s explanation. You are demonstrating care with public-style financial information. You are not writing on Amrize’s behalf, forecasting its performance or claiming experience in its reporting process.
Sources[4]
D. E. Shaw investor relations: the same title, a different audience
D. E. Shaw’s London Summer 2027 IR internship describes work with current and prospective investors, including communication and investor-focused research. The ten-week program is expected between June and August. It accepts students enrolled in bachelor’s or postgraduate programs in any field and says previous finance experience is unnecessary; related clubs or internships are a plus, not a requirement. The retrieved page gives no numeric pay range or fixed closing date.
Do not automatically send the same earnings sample used for an issuer. This role’s audience is investment-manager clients rather than only public-company shareholders and analysts. An earnings bridge can show numeracy, but it does not fully demonstrate client-focused research or communication. Add a one-page audience brief: who is asking, what decision they face, what information is public and what an authorized colleague must answer.
Keep the brief factual and bounded. Do not invent fund returns, private investor holdings or a diligence conversation. A good sample can explain which questions remain unanswered and how you would organize a response. The stronger story is that you adapted your work to a different reader, not that every finance-facing internship calls for the same model.
Sources[5]
Write the evidence map before writing a cover letter
For one actual role, make a three-row table. The first column contains a responsibility, faithfully paraphrased from the posting. The second names a completed artifact. The third states what you personally did and what it proves. Add a separate note for a requirement you do not yet meet. This is not a way to explain away an essential condition; it is a way to prevent the application from implying that a project proves more than it does.
| Need | Completed evidence | Boundary |
|---|---|---|
| Reconcile business performance | Five-product price, volume and variable-cost bridge | Synthetic data; not a company operating result |
| Communicate a recommendation | Short memo with a condition that changes the decision | No claim a manager implemented it |
| Use SQL | Query file with aggregation and reconciliation checks, if completed | A spreadsheet alone does not establish this |
| Meet degree/location conditions | Applicant’s actual enrollment and availability | A work sample cannot replace an employer requirement |
For example, a commercial-analysis row might say: “Performance analysis → product-level contribution bridge → selected comparable products, separated new and exited items, and reconciled a $3,550 decline.” A second might say: “Recommendations → 180-word manager memo → identified which assumption needs investigation before a price action.” A third could say: “SQL → not yet demonstrated.” That last row is useful because it tells you exactly what remains, rather than allowing a dashboard screenshot to conceal it.
When a deadline is close, use evidence you already have. Do not create a rushed new project whose formulas you cannot defend. A concise account of relevant coursework can be better than a copied reference solution. An optional extension belongs after submission when it would improve your learning, unless the employer explicitly requires the additional material.
Describe the project without borrowing an employer’s results
Your application paragraph should connect the role’s task with a specific piece of finished work. Name the dataset’s synthetic status once, then explain your contribution and interpretation. Avoid a list of tool names without an analytical result. Avoid a result stated so grandly that it becomes a claim about a real company.
Original example: completed fictional commercial analysis
The example below is appropriate only after you have performed and understood the work. Change it to match your own implementation and contribution. A reviewer should be able to ask how a number was calculated, which rows were excluded and what would change the conclusion. You should be able to answer without relying on the reference memo’s wording.
Do not attach a large bundle unless the application requests it or provides a suitable portfolio field. A short project landing page or clearly named file can help the reviewer locate the brief, calculation and memo. Check the employer’s instructions for file type, size and permitted assistance instead of creating extra submission steps.
One meaningful extension is better than a cosmetic rewrite
Change one assumption that could matter to the decision. In the pipeline project, move the reporting cutoff and explain which snapshots become eligible. In retention, add an account with two subscriptions and verify that customer churn does not count a plan replacement. In the support project, change the evaluation time and show when a pending unanswered ticket becomes a breach.
For a financial project, add a sensitivity rather than more formatting. Delay acquisition benefits by one period, alter a variable cost, or change a service-start date using an explicit rule. Preserve the original result beside the revised one. Explain whether the decision changed and why. Your contribution is the extension and its analysis, not changing colors in a workbook.
Ask another person to inspect the result using the data dictionary and memo. Record a real correction or question they raised. Do not convert that one review into an exaggerated claim of industry validation. A small change with a clear reason gives you something authentic to discuss in an interview.
A finished sample is a short chain of verifiable decisions
Before putting the project in an application, check that the input files are present, the reporting population is stated, the arithmetic reconciles and the memo answers the original question. Confirm that the result is labeled with its units and period. Make missing information visible. Remove any sentence that claims employer access, a deployment or a measured business improvement you did not have.
Practice a two-minute explanation: the question, the first misleading answer, the correction and the decision. Then prepare for an interruption. Could you explain a different denominator? What happens if the missing data changes? What part of the work was yours? A sample you can revise in conversation is more convincing than one you can only present from beginning to end.
Return to the actual posting before applying. The dates, locations and requirements above belong to their own requisitions and may change. Use the official application route, retain the version you relied on, and submit with accurate evidence. The project is there to strengthen your explanation of your work, not to become another obstacle between you and a suitable application.
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.