Try it free first — before you pay for the full job.
Send 100 rows of your own file, free, and get them back cleaned within 24 hours. You see duplicates removed and fields fixed before you pay for the full job (Starter from $149).
Watch a DEMO: before / after cleanup walkthrough
DEMO — SAMPLE ONLY, NOT CLIENT WORK
This short DEMO walks through a before / after cleanup using fictional demo material only. It shows the 90% rule from this page: matches at or above 90% confidence merge; matches below 90% are flagged, not merged silently. It is not a client's file, and no result is claimed for any real job.
Before / after — SAMPLE data clean
SAMPLE — synthetic demo rows only, not a client's file. Use the toggle to switch between the messy file as received and the cleaned file as delivered. Actions follow this page's 90% rule: matches at or above 90% confidence merge; matches below 90% are flagged, not merged silently.
| Name | Phone | Note | |
|---|---|---|---|
| Jon Smith | jon.smith@example.com | (555) 010-1234 | — |
| John Smyth | jonsmith@example.com | 5550101234 | Same phone, different email format |
| Jon Smith | jon.smith@example.com | (555) 010-1234 | Exact duplicate — trailing space in email |
| Maria Lopez | maria@exampel.com | 555-010-9876 | Possibly misspelled domain |
| Sam Patel | sam.patel@example.com | 555.010.5555 | Inconsistent phone format |
SAMPLE only: the duplicate was removed, the 96% match merged, and the 82% match flagged — below the 90% line, so flagged, not silently changed.
How it works
Three steps. A human reviews every fuzzy-merge decision, and spot-checks 10% of enriched fields against the public source.
- 1Send a sample — 100 rows of your file, free, returned cleaned within 24 hours.
- 2Order — pick the tier that fits your row count and pay the deposit (or full price for Starter).
- 3Receive — cleaned file, change log, confidence flags, and a 1-page “what to do next” note.
Prices
The clock starts only when payment/deposit, a file that opens, and the completed intake form are all received. If the file does not open, the row count does not match the quote, or the job asks for login-walled data, work stops and the job is re-scoped before starting.
Starter Clean
Up to 1,000 rows. Dedupe, standardisation, formatting fixes, bounce/invalid flags. 100% upfront. 2 business days.
Standard Clean + Enrich
1,000–5,000 rows. Everything in Starter, plus enrichment from public sources. 50% upfront, 50% on delivery. 2–3 business days.
Pro Job
5,000–20,000 rows, or custom scrape + clean. Quoted after your free 100-row sample. 50% upfront, 50% on delivery. Typically 3–7 business days.
Tiers compared — rows, days, and what is included
Every figure in this table is repeated from the Prices section above. This service does not state a revision-round count in its terms; the follow-up included with every job is the one bounce re-run within 14 days shown in the last column.
| Tier | Rows / records | Delivery (business days) | Payment | What is included | Included follow-up |
|---|---|---|---|---|---|
| Starter Clean — $149 | Up to 1,000 rows | 2 | 100% upfront | Dedupe (exact + fuzzy), standardisation, formatting fixes, bounce/invalid flags, change log, match-confidence flags, 1-page next-steps note | One bounce re-run within 14 days |
| Standard Clean + Enrich — $349 | 1,000–5,000 rows | 2–3 | 50% upfront, 50% on delivery | Everything in Starter, plus enrichment from public sources: missing company, title, website, and company-domain fields filled where found, each labelled with confidence and source | One bounce re-run within 14 days |
| Pro Job — $749–$1,500 | 5,000–20,000 rows, or a custom scrape + clean | Typically 3–7 (quoted after the free 100-row sample) | 50% upfront, 50% on delivery | Everything in Standard, scoped to your file after the free sample shows what the job needs | One bounce re-run within 14 days |
Scope boundaries — read before ordering
- Row-count boundaries: Starter covers up to 1,000 rows; Standard covers 1,000–5,000 rows; Pro covers 5,000–20,000 rows or a custom scrape + clean. The exact row count in your file (excluding the header row) must match the quote — if it does not, the job is re-scoped before starting.
- What makes a job the larger, quoted kind of job: a row count above the Standard range, or a custom scrape + clean as part of the job. Enrichment — filling missing company, title, website, and company-domain fields from public sources — is a Standard and Pro feature, not part of Starter.
- The audit trail on every delivery: the delivered file carries cleaned fields plus match_confidence, action_taken (kept / merged / removed / flagged), and source for every enriched field, and the change log shows exactly what was removed, merged, or fixed — so every change can be traced back to a stated action.
- Work stops and is re-scoped before starting if the file does not open, the row count does not match the quote, or the job asks for login-walled, paywalled, or terms-prohibited data.
- Merge rule: when two rows look like the same person or company, you choose at intake which row should survive — the newest, the most complete, or ask-me-per-match. The default is most complete, and nothing merges below 90% confidence without your decision.
After you order
You receive a confirmation stating the tier, the row count on record, the date your turnaround clock started, and — for Standard and Pro — the balance due on delivery. The clock starts only when payment or deposit, a file that opens, and the completed intake form are all received. On delivery you receive the cleaned file, the change log, the confidence flags, and the 1-page next-steps note described on this page.
The 90% rule
No merge happens automatically below 90% confidence. Low-confidence matches are flagged for your decision, not merged silently. This rule is never waived to hit a deadline.
Every delivered row carries match_confidence + action_taken (kept / merged / removed / flagged) + source for every enriched field.
What a delivery actually contains
Not just a cleaned file. The change log shows exactly what was removed, merged, or fixed, and the next-steps note tells you what to do with the flagged rows. Format below is a SAMPLE — demo figures only, from a fictional demo dataset, not a client's file.
| Delivered item | What it shows |
|---|---|
| cleaned file | Cleaned fields, plus match_confidence, action_taken, and source on every enriched field. |
| change log | Rows removed as exact duplicates, rows merged (fuzzy, ≥90% confidence), rows flagged for client decision (<90%), emails and phones standardised, enriched fields filled, bounce/invalid flags. |
| next-steps note (1 page) | Review the flagged rows and confirm or reject each merge. Re-import the cleaned file, not the original, so duplicates do not return. Suppress the bounce/invalid flags before mailing. One bounce re-run is included within 14 days — send the bounce report and those rows are re-run once, free. |
SAMPLE — demo dataset only: in the demo file, two spellings of the same person (different email formats, same phone) merged into one record at 96% confidence; a misspelled domain was flagged at 82% — below the 90% line, so flagged, not silently changed; and an exact duplicate (a trailing space in the email) was removed.
What's included
Dedupe (exact + fuzzy), standardisation, formatting fixes, bounce/invalid flags, enrichment from public sources (Standard/Pro), change log, match-confidence flags, 1-page next-steps note, and one bounce re-run within 14 days.
What's NOT included
Purchased lead lists sold as a product. Cold outreach sent on your behalf. “Verified” claims for data that was not verified. CRM migration unless scoped separately. Login-walled, paywalled, or terms-prohibited scraping. Personal mobiles from non-public sources. Any promise of 100% deliverability.
How to judge a data-cleaning vendor — a checklist
Use these questions with any vendor, including this one. The answers in the right-hand column are this page's own stated terms.
| Ask this | Why it matters | DataScrubbed's stated answer |
|---|---|---|
| Can I see the work on my own file before paying for the full job? | A sample on your own data shows real duplicates removed, not a polished demo. | Yes — a free 100-row sample on your file, returned cleaned within 24 hours. It is the only free item; there is no free full job. |
| What is your rule for uncertain matches? | Silent merges destroy good records; a stated confidence threshold can be checked. | No merge happens automatically below 90% confidence. Lower-confidence matches are flagged for your decision, never merged silently, and the rule is never waived to hit a deadline. |
| Will I get a record of what changed? | Without a change log and per-row actions, a cleaned file cannot be audited. | Yes — a change log, plus match_confidence, action_taken, and source on the delivered file, and a 1-page next-steps note. |
| Does a human check the judgement calls? | Fuzzy merges and enriched fields are where automated runs go wrong quietly. | A human reviews every fuzzy-merge decision and spot-checks 10% of enriched fields against the public source. |
| What happens to my file afterwards? | Your list should not become someone else's asset. | Your file is deleted within 30 days of delivery unless a retention agreement exists in writing. Your list is never retained, resold, or used for another client. |
| What will you not do? | A vendor with no stated exclusions will attempt anything, including data you should not hold. | No login-walled, paywalled, or terms-prohibited scraping; no personal mobiles from non-public sources; no purchased lead lists sold as a product; no outreach sent on your behalf; no promise of 100% deliverability. |
Category comparison — done-for-you clean vs the alternatives
Category-level only: no vendor is named here, and no other vendor's prices are quoted.
| DataScrubbed (this service) | A freelancer | An agency | Cleaning software | |
|---|---|---|---|---|
| Who does the work | DataScrubbed — you send a file and receive a cleaned file back | An individual you find, brief, and manage yourself | A managed team you brief through the agency | You — the software is a tool you run yourself |
| What you receive | Cleaned file, change log, confidence flags, and a 1-page next-steps note | Whatever deliverables you agree with that individual | Whatever deliverables the engagement is scoped to | Whatever output the tool produces from the settings you choose |
| Judgement on uncertain matches | Stated 90% rule; below-threshold matches flagged for your decision; human review of fuzzy merges | Ask what rule the individual follows before you hand over the file | Ask what rule the team follows and who reviews merges | Ask what the tool does with matches below the threshold you set |
| Before you commit | Free 100-row sample on your own file first | Ask for a sample on your own file before the full job | Ask what pilot or sample is offered and what it commits you to | Ask whether a trial lets you test it on your own data |
Data handling
Only data you lawfully hold is processed. Your file is deleted within 30 days of delivery unless a retention agreement exists in writing. Your list is never retained, resold, or used for another client. Sources' terms and robots rules are respected; if a source prohibits scraping, its API is used or that field is declined.
Maximum 3 concurrent jobs. If capacity is full, your job joins a waitlist and you are told the start date before work begins — turnaround is never silently extended.
Refund / cancel terms
Refunds under $50 are automatic; refunds of $50 or more are escalated to the business owner (decision within 3 business days). Cancel before work starts for a full refund minus any platform fee already charged. Once work starts, the fee for work completed is retained.
Who this is for
Solo/small sales teams, recruiting firms, real-estate teams, marketing agencies, and SaaS SDR teams with 1,000–20,000-row lists they bought, scraped badly, or inherited.
Who this is NOT for
Enterprises needing SOC 2 / DPA / on-premise handling; anyone asking for scraped personal mobiles, purchased consumer-data resale, or login-walled data; mortgage lenders.
The full SAMPLE — before/after pair and change log (demo dataset only)
SAMPLE — built from a fictional demo dataset. It is not a client's file, it shows no real client's results, and no deliverability or accuracy rate is claimed for it. The toggle above shows a shortened version; the complete sample from the sample deliverables is set out here.
SAMPLE — Before: the fictional demo file “Demo Realty Contacts” (illustrative rows)
- Row 3: “Jon Smith, jon.smith@demorealty.example, 555-010-1234, Demo Realty” and Row 41: “John Smith, jonsmith@demorealty.example, (555) 010-1234, Demo Realty LLC” — the same person, two spellings, two email formats.
- Row 17: “Acme Corp” with no contact name and a misspelled domain, “acmecorp.exampel”.
- Row 58: a duplicate of Row 12 with a trailing space in the email address.
SAMPLE — After: what the cleaned file looks like
- Rows 3 + 41 merged into one record, confidence 96%, surviving email jon.smith@demorealty.example, phone standardised to +1-555-010-1234.
- Row 17's domain corrected to acmecorp.example and flagged “confidence 82% — verify before mailing” — below the 90% auto-merge line, so it is flagged, not silently changed.
- Row 58 removed as an exact duplicate.
- Delivered columns in a real job: cleaned fields plus match_confidence, action_taken (kept / merged / removed / flagged), and source for every enriched field.
SAMPLE — Change log excerpt (format used on every real delivery; figures are demo figures)
| Change-log line | Demo figure |
|---|---|
| Rows received | 1,000 |
| Rows removed as exact duplicates | 87 |
| Rows merged (fuzzy, at or above 90% confidence) | 64 |
| Rows flagged for client decision (below 90% confidence) | 23 |
| Emails standardised | 412 |
| Phones standardised | 388 |
| Enriched fields filled (Standard/Pro): company / title / website | 190 / 142 / 176 |
| Bounce/invalid flags | 31 |
The “what to do next” note that accompanies every real delivery reads, in template form: (1) review the flagged rows (confidence below 90%) and confirm or reject each merge; (2) re-import the cleaned file, not the original, so duplicates do not return; (3) if you mail this list, suppress the bounce/invalid flags first; (4) one bounce re-run is included within 14 days of delivery — send the bounce report and those rows are re-run once, free.
Contact DataScrubbed
Use this form to start by email. Prefer plain email? The mailto links elsewhere on this page still work as before.
Get started
Questions or ready to start? Email DataScrubbed and tell us what you need — or send your free 100-row sample.
hello@datascrubbed.co