Data cleansing services are the work of turning messy customer, vendor and product records into one clean copy you can load. Eicra Soft Limited, a Dhaka-based team serving international companies, works inside your own systems and prices the work per 1,000 records, and a quality assurance (QA) lead re-checks a sample of every batch before it goes back to you.
You can stop at any step, with no long-term contract. We start with a free sample: send 500 records and get them back cleaned, with a change log and a count of the duplicates, invalid emails and missing fields found. The full file is then profiled and priced per 1,000 records in writing, before any record is touched.
You get a written verdict — manual cleansing, automated rules with review, a pipeline instead or nothing yet. If the sample shows little to fix, you stop here.
Data cleansing services are one of our data engineering services and include six deliverables: an audit that counts what is wrong, deduplication on the match keys you agree, standardisation of names, addresses and dates, validation of emails, phones and postcodes, enrichment of missing fields from sources you approve, and a change log with a rejected-records list. You get a file you can load, not a report about one.
Microsoft’s Dataverse documentation states: “Duplicate records can creep into your data when you or others enter data manually or import data in bulk.” That is why every file is profiled first: the audit counts duplicates, invalid values and missing fields before any rule is written.
Every field is profiled before work starts: the duplicate rate, invalid emails, missing fields and the mix of formats. You get a written count of each, so the cleaned file can be checked against the same numbers and nobody has to take the result on trust.
Duplicates are found on the match keys you agree, such as email, phone, or company plus address, and merged into one surviving record. The merge rule is recorded in the change log with the record ID, so any merge can be traced and reversed later.
Names, job titles, company names, addresses, phone numbers and dates are put into one agreed format. Filters, mail merges and imports stop breaking on variants such as ABC ltd and Abc Limited, and the agreed formats are written down so new records can follow them.
Email syntax and domain checks, phone format checks and postcode-to-city checks run on every record. Records that fail are flagged in a rejected list for you to review, never silently deleted, so you decide what happens to each one before the file goes back.
Missing company, title or address fields are filled only from sources you approve and can name. Nothing is scraped and nothing is guessed, and every added value is marked in the change log, so your team can see which fields came from enrichment rather than the original record.
Every batch comes back with a row-level log of every change, the rejected-records list and a QA lead’s sign-off on a re-checked random sample. Each changed value carries the record ID, the rule and the reviewer, so any single change can be found and reversed.
CRM contacts and leads are deduplicated, emails validated and job titles standardised before the next campaign or territory split. Sales then works from one record per person, and marketing sends only to addresses that have passed syntax and domain checks rather than to lists nobody has verified.
Vendor and customer master records are merged and bank and tax fields validated, so payments reach one correct entity. Every merge is logged with the rule and the reviewer, which gives finance a written trail of which supplier record survived and why the others were set aside.
Product, stock-keeping unit (SKU) and supplier lists are standardised across ERP, e-commerce and spreadsheets so stock and orders match. One agreed format per field means the same product carries the same name in every system, and imports between those systems stop failing on format.
Employee and candidate records from human resources (HR) systems are deduplicated and addresses and IDs validated. Personal fields stay masked while reviewers work, so they see names and emails only for the fields being cleaned, and every working copy is deleted at sign-off.
Data cleansing changes a CRM file record by record. Duplicates are found on the match key you agree; names, companies and phone numbers are put into one format. Emails are validated, and anything that fails is listed rather than deleted. In the example below, three versions of one contact become one surviving record and one invalid email is set aside.
Rules agreed in writing
What the rule changes · its result
Match key · email + phone
Before · CRM export, 4 records
| Name | Company | Phone | |
|---|---|---|---|
| abc xyz | ABC ltd | abc@xxx.com.bd | 01711XXX142 |
| Abc Xyz | Abc Limited | ABC@XXX.COM.BD | +8801711XXX142 |
| A. Xyz | Abc | abc@xxx.com.bd | 01711 XXX142 |
| Xxx Yyy | Xyz Foods | xxx@xyz | 01819 XXX199 |
After · 1 surviving record
Formats · name case, company suffix, phone
One agreed format for every value.
Validation · email syntax and domain
Every email checked for syntax and a real domain.
Reviewer re-checks a sample · a person signs
A sample of every batch, before it goes back to you.
Click a rule, or a number below, to switch the result
Managed Ops cleans new and changed records every week or month, with a monthly quality report.
Illustrative example. Yellow marks the step where a person decides; nothing is deleted without being listed for you.
Our data cleansing services run in five steps, and each ends with a result you can stop on. A free 500-record sample comes first, then a profiling audit prices the full file per 1,000 records. A pilot batch is cleaned and signed off, production covers the whole file in batches, and Managed Ops handles the records that keep arriving.
| # | Step | Time | What you get |
|---|---|---|---|
| 01 | Scope | 2 days · free | Send 500 records; get them back cleaned with a change log and counts. |
| 02 | Diagnose | 3–5 days | Full file profiled; match keys and formats agreed; price per 1,000 quoted. |
| 03 | Pilot | 1 week | One batch of up to 10,000 records cleaned, logged and sample re-checked. |
| 04 | Production | In batches | Whole file in agreed batches, each with change log, rejected list, sign-off. |
| 05 | Managed Ops | Monthly · optional | New records cleaned weekly or monthly, with a monthly quality report. |
A free 500-record sample comes back in two working days, the profiling audit takes three to five days, and a first pilot batch of up to 10,000 records takes about a week. Each step closes with a written result, so you can stop there without paying for the next one.
Not every dirty file needs the same data cleansing fix. A few thousand records with judgement calls need reviewers. A hundred thousand records with repeatable errors need rules and a reviewed sample. Records that get dirty again every month need a pipeline from our AI-ready data services, not a one-off clean. Five questions show which one fits, before anyone quotes a price.
1. How many records need cleaning?
2. What is mostly wrong with them?
3. Where do the records live?
4. How often do new records arrive?
5. Do the records contain personal data?
5,000 to 100,000 records with duplicates and format errors: rules do the bulk, reviewers handle the exceptions, and a 10,000-record pilot batch shows the real rate.
How the verdict is decided: daily arrivals from several systems → a pipeline · under 5,000 records, or judgement calls → manual cleansing by reviewers · over 100,000 records with repeatable errors → automated rules with a reviewed sample · everything in between → reviewers plus rules, piloted on one batch.
A data cleansing company is judged on the counts it can show before and after, not on adjectives. We profile every file first and price per 1,000 records. Our data scrubbing services deliver a change log, a rejected-records list and a signed sample check on every batch, which supports the accuracy principle of the General Data Protection Regulation (GDPR).
The Government Data Quality Framework states: “Data quality is more than just data cleaning.” That is why every batch comes back with counts before and after, a rejected list and a reviewer’s sign-off, not just a cleaned file.
Sending records for data cleansing is safe when access, transfer and deletion are settled in writing. As a Bangladesh-based company, we sign a non-disclosure agreement (NDA) and a data processing agreement (DPA) before any file moves. We work inside your CRM and delete every working copy at sign-off; the rules stay your intellectual property (IP).
Three agreements are signed before any file moves, and access is set up in the order shown below. Each agreement names what we may see, where it may be processed and when our access ends, and certifications are listed only when they are actually held.
Five controls apply to every engagement and are written into the contract rather than promised in a proposal. They cover where the work happens, how personal fields are handled, how every change is logged, who owns the result and how a batch that misses its sample check is reworked.
Before you pay for data cleansing, you can see the result first. You get three measured commitments, a free 500-record sample cleaned before any quote, and a signed sample check on every batch. Client case studies follow as clients agree to be named.
For a free 500-record sample, returned cleaned with a change log and counts before any quote.
Re-checked on a random sample and signed off by the QA lead, with a rejected-records list.
Records: the price is written after the profiling audit, before any record is touched.
Buyers ask twelve questions about data cleansing services before they send a file. They ask what the services cost, what they are, why they matter, manual versus automated work and how a project starts. They also ask about cleaning inside their CRM, ownership, storage, protection, scraping, what follows a sample and when a one-off clean is not enough.
More answersData cleansing services start with a free 500-record sample, then a profiling audit that prices the full file per 1,000 records in writing and is credited to the pilot. The pilot batch is a fixed $1,200, production starts from $3,600 and is quoted after the pilot, and Managed Ops is priced monthly and optional.
Data cleansing services find and fix duplicate, invalid and incomplete records so every team, data analytics report and data visualization dashboard works from one correct copy. They cover profiling, deduplication on agreed keys, standard formats, validation, enrichment from approved sources, and a change log with a rejected-records list.
Data cleansing matters wherever a record is typed in more than once, because campaigns and payments go wrong when one customer or vendor exists in several versions. A cleaned file gives each customer one record, each payment one correct entity and each campaign a list of addresses that passed validation, with a change log showing what was fixed.
Manual cleansing uses reviewers working record by record, while automated cleansing runs rules on the whole file. Manual suits small files and judgement calls; automated rules, written by an engineer in Structured Query Language (SQL) or OpenRefine, suit large files, with reviewers checking exceptions. Most files over 100,000 records need both.
A project starts with a free 500-record sample, and every later step is priced in writing before any record is touched, so you can stop after any step with no long-term contract. Each step ends with a written result, and the steps run in this order.
Yes. Where your CRM, such as HubSpot, Salesforce or Dynamics 365, has native merge and dedupe tools we use them, so the change history stays in your system. Where it does not, we clean an export in a shared workbook you control and load the result back through the CRM's own import; nothing is copied to our systems.
You do. The clean file, change log, rejected list and rules are yours under the contract, and the work happens inside your CRM or a workbook you control. Our team signs a data processing agreement before any personal field is seen, our access is removed at sign-off, and every working copy is deleted then.
Your data stays in your own CRM or a shared drive you own; the work happens there and nothing is copied to our systems. Every working copy is deleted at sign-off, in writing, and international transfers use standard contractual clauses or the instrument your jurisdiction requires before any personal data moves.
Customer data is protected by contract first. A mutual NDA is signed before any file, export or credential is shared, and a DPA under Article 28(3) of the GDPR covers any personal data. Personal fields are masked wherever a rule does not need them, and reviewers see names and emails only for the fields being cleaned.
No. Third-party contact lists and scraping are outside the service. Missing fields are filled only from sources you approve and can name, nothing is guessed, and every added value is marked in the change log, so you can see exactly which fields came from enrichment and which came from your original records.
After you send 500 records, they come back cleaned within two working days with a change log and counts of the duplicates, invalid emails and missing fields found. If the sample shows little to fix, you stop there; otherwise the full file is profiled and priced per 1,000 records in writing before any record is touched.
A one-off clean is not enough when records get dirty again every month or arrive daily from several systems. Those files need a pipeline instead, or Managed Ops, which cleans new and changed records every week or month, sends a monthly quality report with duplicate and invalid rates, and updates the rules as sources change.
Send 500 records for a free cleaned sample before any quote.
Request A QuoteSend a Free SampleYour request is subject to our Terms and Privacy Policy.
