Freelance and independent data scientists bill companies, startups, nonprofits, and agencies for analysis, modeling, and data work. Typical invoices cover project phases such as data cleaning, exploratory analysis, model development, dashboard builds, and ongoing support or retainers. Many also bill for consulting calls, documentation, and cloud compute passed through to the client.
Create an invoice for payment collection
Upload logo
PDF will be generated with filename:
Invoice_1001_Client_2026-10-14.pdf
Most independent data scientists use one of three models. Hourly billing is common for open-ended or exploratory work, support, and ad hoc analysis. Fixed-fee project pricing works when the deliverable is clear, such as a churn model, a forecasting pipeline, or a customer segmentation report, and is usually split into phases with milestone payments. Monthly retainers cover a set number of hours or ongoing model monitoring and advisory work. Cloud compute, data licenses, and labeling costs are usually billed at cost or with a small markup, and are listed separately from labor. Rates rise with specialization (machine learning engineering, causal inference, NLP) and with the client's size and industry. Many people price a fixed project by estimating hours, adding a buffer of 15 to 30 percent for messy data and revisions, and then quoting the total.
| Service | Priced | Typical range |
|---|---|---|
| Data analysis and reporting | per hour | $75-150 |
| Machine learning model development | per hour | $100-200 |
| Dashboard build (BI tool) | flat fee | $2,000-8,000 |
| Data cleaning and pipeline setup | per hour | $85-160 |
| Strategy and advisory consulting | per hour | $150-250 |
| Monthly retainer for model monitoring and support | per month | $1,500-6,000 |
| A/B test design and analysis | flat fee | $1,500-5,000 |
| Training workshop or team session | per day | $1,500-3,500 |
Typical US ranges. Rates vary by region, experience, and job size.
| Description | Qty | Rate | Amount |
|---|---|---|---|
| Data cleaning and feature engineering (customer transaction data, 3 source tables) | 14 hours | $110.00 | $1,540.00 |
| Churn prediction model development and validation (gradient boosting, cross-validation) | 22 hours | $135.00 | $2,970.00 |
| Model documentation and results presentation to leadership team | 5 hours | $135.00 | $675.00 |
| Weekly check-in calls | 4 hours | $150.00 | $600.00 |
| Cloud compute for model training (client-approved, billed at cost) | 1 flat | $186.00 | $186.00 |
| Total | $5,971.00 | ||
Notes: Invoice covers Phase 2 of the churn modeling project per the signed statement of work dated March 3. Payment due within 30 days. Late balances accrue 1.5% interest per month. Code and model artifacts are delivered to the client repository upon payment in full.
Ties the charges to agreed scope so the client's accounts payable team can approve it without questions.
Many mid-size and large companies will not process an invoice without a PO number.
Separating data cleaning, modeling, and reporting shows where time went, especially when data quality was worse than expected.
Pass-through costs such as GPU time, API credits, or purchased datasets should be visible and match any receipts you can share.
Helps with retainers and hourly work, and lets the client match the invoice to their project timeline.
Stating that code, models, and reports transfer on full payment protects you if the client delays.
US clients typically need a W-9 to pay a contractor and to issue a 1099 at year end.
Net 30 is the norm with corporate clients, though startups and small businesses often agree to net 15. For fixed-fee projects, a deposit of 25 to 50 percent upfront is common, with the rest tied to milestones such as data delivery, a working model, and final handoff. Retainers are typically billed at the start of the month. Late fees of 1 to 1.5 percent per month are often written into the contract, though check your state's rules on interest. Payment is usually by ACH or wire for larger amounts, with cards accepted for smaller engagements. Large companies may pay on 45 or 60 day cycles after vendor onboarding, so confirm this before starting.
Poor data quality is the most common cause of overruns. Include a short paid discovery or data audit phase, and put a clause in the contract that significant data issues can change scope or timeline.
Define payments around a cleaned dataset, a baseline model, and a final handoff. Modeling work is uncertain, and milestones keep cash flowing even if accuracy targets shift.
Promising a specific accuracy or lift in exchange for payment is risky because it depends on data you do not control. Commit to a documented method and process, and report results honestly.
Get written approval before running expensive training jobs, and bill at cost with receipts or an agreed markup. Better still, have the client run work in their own cloud account.
Send the final invoice when you deliver, and make release of repositories, notebooks, and model files conditional on payment as set out in your contract.
Use hourly for exploratory work, unclear scope, or ongoing support. Use a fixed fee when the deliverable and data are well defined, and build in a buffer for data cleaning and revisions. Many people combine them with a paid discovery phase billed hourly followed by a fixed-fee build.
Only if your contract says so. Many consultants include a clause that delays caused by the client, such as late data delivery, can extend the schedule and trigger billing for idle or reserved time. Otherwise, build that risk into your estimate.
List them as separate line items from your labor, at cost or with an agreed markup, and attach receipts if requested. Get approval for large spending in advance. If the client owns the cloud account, you avoid the issue altogether.
As a US contractor you typically give the client a W-9, and they report payments to you on a 1099 if they exceed the IRS threshold in a year. You do not send the 1099 yourself. Check current thresholds and talk to a tax professional about your situation.
Refer to the acceptance criteria in your statement of work. If you defined deliverables, validation metrics, and a review period, you can invoice on delivery. Additional tuning beyond what was scoped can be billed as a change order.
Send a Data Scientist estimate for approval, then turn it into an invoice when the work is done.
Data Scientist estimate template