BidGrow

Upwork proposal templates for data analysts

Updated

Clients hiring a data analyst on Upwork usually check whether you understand the business question, the data source, and the output they need. They are not only hiring “Excel,” “SQL,” or “Power BI.” They want someone who can turn messy numbers into a decision, a dashboard, a report, or a clean answer they can trust.

Template 1: fully written Power BI dashboard proposal

Power BI sales dashboard
Hi Jordan — I read the brief and saw that you need a Power BI sales dashboard from Shopify exports and ad spend data, with weekly refreshes and clear views by product, channel, and region.

I can help build this in a way that is useful for sales and marketing, not just visually clean. I would start by confirming the grain of each table: order line, customer, product, campaign, and date. Then I’d map the joins, check for duplicate orders or missing campaign tags, and agree on metric definitions before building the dashboard.

For the first version, I’d suggest these pages:
- Executive summary: revenue, orders, AOV, gross margin, ad spend, ROAS, and week-over-week change
- Product view: best and worst sellers, margin by SKU, and trend by category
- Channel view: paid search, paid social, email, organic, and direct
- Regional view: revenue and margin by country or state
- Data QA page: missing values, unmatched campaign names, and refresh status

A few questions before I estimate the final timeline:
1. Is Shopify the source of truth for revenue, or do refunds/adjustments live somewhere else?
2. Do you want ROAS based on order date, ad click date, or reporting date?
3. Should the dashboard refresh from exported files, Google Sheets, or a database connection?

Relevant work: I’ve built Power BI dashboards for sales reporting, marketing performance, and recurring KPI packs. I can share a dashboard sample that shows my layout, DAX style, and QA notes.

If helpful, I can start with a short metric-definition sheet first so we avoid rebuilding visuals after the dashboard is already designed.

This is a complete example for a common dashboard brief. Replace the tools, data sources, and metric list if the job post is different.

Template 2: SQL analysis or one-time reporting task

SQL analysis task
Hi {{client_name}} — I saw that you need {{specific_analysis_from_post}}, especially {{detail_that_proves_you_read_the_post}}.

I can help with the SQL analysis and make the output easy to check. My first step would be to confirm the grain of the tables involved, such as {{table_1_grain}} and {{table_2_grain}}, then check the join keys, date filters, and any exclusions before writing the final query.

For this type of task, I’d usually deliver:
- A clean SQL query with comments
- A short results table or export in {{preferred_format}}
- Notes on assumptions, filters, and edge cases
- A quick QA check, such as row counts before and after joins, duplicate checks, and null checks

A few questions that will help me get this right:
1. Which table is the source of truth for {{main_metric}}?
2. Should {{metric_name}} be counted by {{date_option_1}} or {{date_option_2}}?
3. Do you want the final answer only, or the reusable query as well?

Relevant proof: {{one_sentence_about_similar_sql_work}}. I can link a sanitized SQL sample or a short write-up that shows how I document joins, filters, and metric logic.

If you send the schema or a sample extract, I can confirm the approach and estimate the time before starting.

Use this when the client asks for SQL, BigQuery, PostgreSQL, MySQL, Snowflake, Redshift, or warehouse analysis.

Template 3: ongoing reporting and analytics support

Ongoing KPI reporting
Hi {{client_name}} — I read that you’re looking for ongoing help with {{reporting_area}}, including {{specific_recurring_task_from_post}}.

This sounds like a good fit if you need someone who can maintain the reporting process, not just create one dashboard and disappear. For recurring analytics work, I focus on keeping metric definitions consistent, catching data issues early, and explaining changes in plain English for the people using the reports.

I can help with tasks like:
- Weekly KPI packs for {{stakeholder_team}}
- Variance analysis when {{metric}} moves up or down
- QA on ETL or imported outputs before reports go out
- Cohort, retention, or funnel refreshes
- Ad-hoc SQL requests from sales, marketing, finance, or operations
- Dashboard cleanup when old metrics no longer match the current business

To keep the work organized, I’d suggest a simple intake process: request, data source, metric definition, due date, output format, and decision owner. That helps avoid vague requests like “pull performance” when the real question is about margin, channel mix, churn, or conversion.

A few questions:
1. What reports are already in place, and which ones are trusted?
2. Which metrics cause the most disagreement internally?
3. How often do you need refreshes: daily, weekly, monthly, or on request?
4. Who is the audience: executives, managers, clients, or an internal ops team?

Relevant proof: {{link_or_description_of_reporting_sample}}. I can also share a sample metric dictionary or reporting QA checklist if that would be useful.

Best for hourly or long-term jobs where the client needs repeatable reporting, dashboard maintenance, and ad-hoc analysis.

What clients screen for first

  • Business understanding. Do you know what the metric is for? A sales dashboard, churn report, pricing analysis, and GA4 cleanup all need different thinking.
  • Source-of-truth judgment. Clients want to know you will ask which system is correct when Shopify, Stripe, HubSpot, GA4, and spreadsheets disagree.
  • Data handling. Mention joins, grain, duplicates, date logic, missing values, and QA when they matter. This signals that you will not build a pretty report on broken data.
  • Clear output. Say what they will receive: dashboard, SQL query, Excel model, KPI pack, metric dictionary, cleaned dataset, or written findings.
  • Communication with non-technical people. Many data analyst clients are founders, marketers, operators, or finance leads. Show that you can explain the result without hiding behind technical terms.

Opening line: bad, better, best

First-line examples for data analyst proposals
TypeOpening lineWhy it works or fails
BadI am an experienced data analyst with strong skills in Excel, SQL, and Power BI.It says nothing about the client’s job. Many proposals can say the same thing.
BetterI saw you need a Power BI dashboard for sales reporting from Shopify and ad data.It proves you read the post and names the output and sources.
BestI saw you need a Power BI sales dashboard from Shopify and ad spend data, and the tricky part is likely matching order date, campaign date, refunds, and channel names before calculating ROAS.It shows you understand the hidden analyst work behind the dashboard. It also gives the client confidence that you will check definitions before building visuals.

What to customize before sending

  • The business question. Replace generic wording with the actual question: “Why did paid conversion drop?” “Which products drive margin?” “What is our retention by cohort?”
  • The grain. If relevant, mention whether the data is by order, order line, user, session, event, invoice, ticket, subscription, or campaign. This is often where reporting errors start.
  • The join logic. If the job involves multiple sources, name the likely join points: customer ID, email, SKU, campaign ID, order ID, date, region, or account.
  • The refresh cadence. A one-time analysis is different from a dashboard that refreshes every Monday before a leadership meeting.
  • The audience. A CEO summary, finance model, client-facing report, and analyst handoff need different levels of detail.
  • The tool stack. Mention only the tools that match the post: Excel, Google Sheets, SQL, Power BI, Tableau, Looker Studio, GA4, Python, R, dbt, or a warehouse.

Metric-definition checklist to include when useful

For dashboard and recurring reporting jobs, offering a metric-definition step can make your proposal stand out. It shows you know that “revenue,” “active user,” “lead,” and “conversion” can mean different things across teams.

Simple metric-definition checklist
ItemQuestion to answer
Metric nameWhat exact metric are we reporting?
FormulaHow is it calculated?
Source of truthWhich system or table wins if sources disagree?
Date logicIs it based on created date, paid date, shipped date, event date, or reporting date?
GrainIs each row a user, order, order line, session, event, invoice, or campaign?
FiltersWhat should be excluded, such as test users, refunds, internal traffic, or canceled orders?
OwnerWho approves the definition before the dashboard or report is built?

Proof to include in a data analyst proposal

Match your proof to the client’s need
Client needBest proof to showHow to describe it
Power BI or Tableau dashboardA public or sanitized dashboard screenshot, short video, or portfolio page“Here is a sample dashboard showing layout, drilldowns, DAX measures, and QA notes. The data is anonymized.”
SQL analysisA sanitized query, GitHub gist, screenshot, or short case write-up“This sample shows how I document joins, date filters, and row-count checks before returning results.”
Excel or Google Sheets reportingA sample workbook with formulas, pivots, data validation, and clear tabs“This is a reporting workbook with raw data, cleaned data, calculations, and an executive summary tab.”
GA4 or marketing reportingA Looker Studio sample, channel report, or metric dictionary“This shows how I separate source/medium logic, campaign naming issues, and conversion definitions.”
Ongoing analyst supportA recurring KPI pack, issue log, or report calendar“This shows how I manage weekly reporting, variance notes, and data quality checks.”

Do not link a private client file unless you have permission. If you cannot show the original work, create a safe sample with fake data. The proof does not need to be flashy. It needs to show that you can structure data, check it, and explain the answer.

Sample answers to common screening questions

How would you approach building a Power BI dashboard from multiple data sources?

Answer

I’d first confirm the business questions and metric definitions, then inspect each source for grain, keys, dates, duplicates, and missing values. After that I’d build the data model, define measures, create the main views, and add a QA page so the client can see refresh status, unmatched records, and key checks.

How do you make sure your SQL results are accurate?

Answer

I check row counts before and after joins, test for duplicates, inspect nulls in key fields, confirm date filters, and compare totals to the source system when possible. I also comment the query so the client can see assumptions and reuse it later.

Mistakes that get data analyst proposals skipped

  • Listing tools instead of solving the job. “I know SQL, Excel, Power BI, Tableau, and Python” is weak unless you connect those tools to the client’s report or analysis.
  • Ignoring metric definitions. If the client asks for revenue, ROAS, churn, conversion, retention, or LTV, do not assume the formula is obvious.
  • Skipping data quality. Clients worry about wrong dashboards. Mention how you check duplicates, nulls, joins, and source totals when the job calls for it.
  • Overcomplicating a simple job. A founder asking for a clean spreadsheet may not need a warehouse model, Python pipeline, and three dashboards.
  • Sending irrelevant portfolio links. A beautiful dashboard is not useful proof if the job is SQL cleanup. Match the proof to the task.
  • Not asking about access and refresh. For dashboards, clarify whether the data comes from exports, live connectors, Google Sheets, a database, or manual uploads.
  • Sounding like a data scientist when the job is analyst work. If the client needs KPI reporting, variance analysis, or dashboard cleanup, do not lead with machine learning unless they asked for it.

A simple structure you can reuse

  1. 1

    Start with the job detail

    Name the dashboard, report, analysis, data source, or metric from the post. This is the fastest way to prove you read it.

  2. 2

    State the likely data issue

    Mention grain, joins, source-of-truth, refresh cadence, or metric definitions if they are relevant. Keep it brief.

  3. 3

    Describe the deliverable

    Say exactly what they will get: dashboard pages, SQL query, cleaned workbook, report pack, written findings, or QA notes.

  4. 4

    Show matched proof

    Link one relevant sample and explain what the client should look at. Do not dump a portfolio with no context.

  5. 5

    Ask two or three useful questions

    Good analyst questions reduce risk. Ask about source systems, definitions, audience, deadline, and refresh needs.

Where BidGrow can help

If you apply to many analytics jobs, BidGrow can help you sort the feed and draft from your own templates. It reads new Upwork jobs, scores them with rules you can see, and drafts proposals and screening-question answers for your review. You still review everything and press send yourself. BidGrow does not submit proposals, does not log in to Upwork as you, and is not affiliated with or endorsed by Upwork.

Questions

What should a data analyst proposal on Upwork include?

Include a first line that references the exact job, a short plan, the likely data checks, the deliverable, one relevant proof link, and a few useful questions about metrics, sources, or refresh needs.

Should I mention SQL, Excel, Power BI, or Tableau first?

Mention the tool the client asked for, but do not stop there. Connect it to the business question, data source, metric definitions, and final output.

How long should a data analyst proposal be?

Keep it long enough to show you understood the job. For simple spreadsheet cleanup, a few short paragraphs may be enough. For dashboards or SQL analysis, include your approach to grain, joins, QA, and metric definitions.

Can I use the same data analyst proposal template for every job?

No. Use a template for structure, then customize the opening line, data source, metric, deliverable, proof, and questions for each job post.

Related