When an automation or reporting project underdelivers, the usual cause is the data feeding into it. A Power BI dashboard connected to inconsistent data shows inconsistent numbers. An Excel automation built on duplicated records produces duplicated outputs. Switching to a different tool leaves both problems in place.
From a single messy spreadsheet to a multi-system data quality project, scoped and priced before any work begins.
Duplicate records identified and merged across your datasets, such as customers entered twice, products with several codes and contacts copied across systems. We also find how the duplicates got in, so the merge holds.
Dates in six different formats, phone numbers with and without country codes, postcodes in inconsistent cases. We standardise them so your data behaves predictably.
Data that lives in several systems and does not match up. We find the discrepancies, resolve the conflicts and set up a process to keep the systems in step from then on.
Blank fields that break your reports and formulas. We find what is missing, fill what can be inferred, flag what needs a person to check, and validate what remains.
Validation built into your data entry, so bad data cannot get into the system in the first place. This is the right fix for data quality problems that keep coming back.
Before any automation project, we audit and clean the source data. Automation built on bad data produces wrong outputs faster. We fix the foundation first.
Bad data builds up over years, a few records at a time. When we open a client's file, the problems nearly always trace back to one of five causes. Knowing which apply tells us what to fix in the data and what to fix in the process that feeds it. Our guide to what data cleansing is and the six steps involved covers the background in more detail.
Someone types a customer name, a date or a postcode by hand, and small differences creep in. We regularly find "United Kingdom", "UK" and "England" in the same country column, and 03/04/2025 meaning March in one row and April in the next. Each entry makes sense on its own. Together they break every SUMIF and pivot table built on them.
The CRM, the accounts package and the operations spreadsheet each hold their own copy of the customer. Staff copy details between them by hand or through a CSV export, and every copy drifts a little. By the time we see it, one client can have three records with three spellings and two addresses.
Nobody wrote down how a date, a product code or a company name should be entered, so each person follows their own habit. Telling staff to be more careful rarely helps. Dropdowns, required fields and validation at the point of entry do.
Records carried over from a previous system keep that system's conventions. Company numbers lose their leading zero when a CSV passes through Excel. Old status codes sit next to new ones. A field that meant one thing in 2018 means something else now, and nobody remembers the difference.
When everyone uses a dataset and nobody is responsible for it, it degrades. Optional fields stay blank. Customers who left in 2021 are still marked active. Duplicates get spotted and worked around, never merged. Part of our handover is agreeing who looks after each dataset once we have finished.
We use three tools. The choice depends on where your data lives, how big it is, and who will run the process after we hand it over. It is decided at the audit stage and written into the quote. Whichever we use, every cleaning step is written down so it can be run again when new data arrives.
Our default when the data lives in Excel or feeds a Power BI report, and your team will run the clean-up again themselves. Every step is recorded in the query, so next month you paste in the new export, press Refresh and get the same result. It handles trimming, splitting columns, fixing date formats, removing exact duplicates and merging lookup tables well. It slows down past a few hundred thousand rows and it is weak at fuzzy matching, so we move to Python when either comes up.
We use Python with pandas when a dataset runs to hundreds of thousands or millions of rows, when records need fuzzy matching (J Smith Ltd against John Smith Limited), or when the checks need real logic. Validating a VAT number against the HMRC mod-97 checksum, restoring the leading zero on an eight-digit company number, or checking postcodes against the UK format each take a few lines of Python and are awkward in Excel. Where the problem recurs, we hand over a script you run with one command, with written instructions.
If the data already sits in a database, such as SQL Server, PostgreSQL, MySQL or the database behind your line-of-business software, we clean it in SQL. Exporting to a spreadsheet and importing back adds a step where things go wrong. One SQL statement can fix a systematic error across millions of rows. We work on a backup or copy you send us, and hand back the scripts with before and after row counts, so your team or your software supplier can run them against the live database.
Most businesses know their data is messy without knowing what specifically is broken, which makes it hard to scope the work or judge a quote. Our free data health check reads a CSV in your browser and tells you what it finds.
It finds duplicate records and missing fields. It also finds the UK-specific problems that break systems quietly. These include VAT numbers that pass a format check but fail the HMRC mod-97 checksum, company numbers where Excel has stripped the leading zero, postcodes in a format no lookup will match, and columns holding both day-first and month-first dates.
Nothing is uploaded and there is no signup. It takes about a minute, and it gives you a far better brief for any cleansing work that follows.
Run a free data health checkEvery automation project we build starts with a data audit. Before we write any code or build any query, we look at the source data and decide whether it is clean enough to build on. In around half of all projects it is not, and the cleansing work happens first.
That is normal. A business that has run on manual data entry for several years will have inconsistencies built up over that time. Cleaning them once, with a structured approach, gives you a foundation that reporting and automation can rely on.
The tools we use are covered above, and the process on the right is the same whichever one the job needs.
You send a sample or the full file. We profile every column: blank rates, duplicate counts, the mix of formats, and values that fail validation. You get a short written list of what we found, with counts, usually within 48 hours. If you have already run our free data health check, send us the results and we start from there.
The quote lists each problem we will fix, the rule we will apply to it, the tool we will use and the delivery date. The price is fixed. If we find something outside the agreed scope once work starts, we tell you what it is and what it would cost before doing it, and you decide.
We apply the agreed rules to a copy of your data, never the original. Anything the rules cannot settle on their own, such as two records that might or might not be the same customer, goes into an exceptions list for you to decide. We do not guess.
We reconcile the output against the input: record counts in and out, totals that should not have changed, and a sample checked by hand against your business rules. You see the numbers, for example that 5,212 records became 4,870 and which rule removed each of the 342.
You get a written note of every rule applied and, where the problem recurs, the Power Query or Python step to run it again yourself. We walk your team through it. After handover we stay available for questions or problems with anything we built.
Most work falls into one of two shapes. Both are fixed price. For how these prices compare with the wider market, see what data cleansing costs a UK business in 2026.
One export from one system, fixed once. Typical jobs are deduplicating a customer list, standardising dates and postcodes, or fixing company and VAT numbers before an import.
Data from two or more systems reconciled into one, or a Power Query or Python process you run yourself each time new data arrives.
Excel (xlsx, xls and xlsm), CSV, Google Sheets, database backups or exports from SQL Server, PostgreSQL and MySQL, and exports from Xero, Sage, Salesforce, HubSpot and most other business software. If your system can export it, we can work with it.
We ask for the minimum fields the job needs. Deduplicating customers needs names, addresses and emails, and it does not need purchase history or bank details. The work is done in the UK on an encrypted machine, and your data is not uploaded to third-party cloud tools. We delete it 30 days after you sign off delivery. Under UK GDPR you remain the controller and we act as your processor, and we are happy to sign your data processing agreement before you send anything.
Everything is done remotely, for businesses across the UK. We do not need logins to your systems. You export the data, we clean the copy, and you import the result or pass our scripts to whoever manages the system. If direct access would make a job faster or cheaper, we say so in the quote and you decide.
After handover we stay available for questions or problems with anything we built. If a supplier changes an export format or a step stops working, you contact the people who wrote it.
Health and beauty · Owner-managed business
Every project gets a fixed quote before any work starts. The price covers the audit, the cleansing, validation, documentation and handover. You do not pay by the hour, and the figure does not move unless you ask us to add something.
A single file or spreadsheet clean-up starts from £99. That is one export from one system, fixed once. A multi-source database, or a repeatable pipeline you run yourself each time new data arrives, starts from £500.
The quote comes down when there is one source in one format, when you can state the rules up front (which field wins in a conflict, which records to drop, what counts as a duplicate), when the data is structured with one record per row, and when you have run our free data health check first so the audit starts from known problems.
Our guide to what data cleansing costs a UK business in 2026, linked below, compares these prices with the wider market and explains when a job is small enough to do in-house with Power Query.
For a figure without a call, the build estimator itemises one in two minutes.
Nothing is paid up front. You watch the finished thing work with your own data, and only then do you pay. No deposit and no retainer. How the process works →
A single file or spreadsheet clean-up starts from £99. A multi-source database, or a repeatable process you run yourself, starts from £500. The price depends on how many sources there are, how much fuzzy matching and manual review the data needs, and whether you want a one-off fix or something you can rerun. We give a fixed quote before any work starts, and it does not change unless you add to the scope.
A single dataset with straightforward problems usually takes 2 to 5 working days from the agreed quote. Jobs involving several systems or a lot of manual review take longer, and the quote states the delivery date. The audit that comes before the quote is usually back within 48 hours of us receiving the data.
We ask for only the fields the job needs. The work is done in the UK on an encrypted machine, and your data is not uploaded to third-party cloud tools. We delete it 30 days after you sign off delivery. You remain the controller under UK GDPR and we act as your processor. We are happy to sign your data processing agreement before you send anything.
Usually not. You export the data, we clean it, and you import the result or pass our scripts to whoever runs the system. Most jobs never need a login. If direct access would make a job faster or cheaper, for example a database too large to export, we say so in the quote and you decide.
We take an export from each system, match records across them, and flag the conflicts, such as a customer whose address differs between the CRM and the accounts package. You tell us which system wins for each field, or we agree a rule together, and you get one reconciled dataset back. Multi-source work starts from £500.
Every project is a one-off build, and we do not sell retainers. Where the same problem will come back each month, we build a Power Query or Python step your team runs themselves, and show them how. After handover we stay available for questions or problems with anything we built. If an export changes format or a step stops working, you get in touch and we help.
Book a free 30-minute call. We will look at your data, tell you what the quality issues are, and give you a fixed price to fix them before any work begins.
Book a free 30-minute call →Or call 07961 593334
From £99 · 2 to 5 working days · Fixed price, paid when it works