Clean up messy business data before you automate anything
Your accounting system, your field software, and three spreadsheets each have a different version of the same customer. Every report and every automation you build inherits that argument.
Free, about four minutes, and read by Raquel herself.
What does messy data look like in a real business?
Picture a 25-person HVAC and plumbing company. A homeowner calls for a furnace repair. The dispatcher finds two customer records in ServiceTitan, one with the old address. The invoice goes out of QuickBooks under a third spelling of the name. The maintenance-plan list lives in a spreadsheet the office manager updates when she remembers.
Nobody did anything wrong. Each system was set up on a different day, by a different person, for a different job. Now nobody can answer two simple questions with confidence. How many active maintenance customers do we have? Which ones are due this month?
This is an illustration, not a client story. If your version involves different software, the work is the same.
How does the cleanup run?
- 01Access and inventory
- 02Map the fields and measure the mess
- 03Agree the rules, clean a copy
- 04Review with your team, then load
- 05Keep it clean, hand it over
All ten steps, in detail
- Kickoff and read-only access. You give Raquel export or read access to each system in scope. Nothing is edited at this stage.
- System inventory. A list of every place a customer, site, piece of equipment, or invoice record lives. That includes the spreadsheets people forget to mention.
- Field mapping. Which column in QuickBooks matches which field in ServiceTitan, and which fields exist in only one place.
- Duplicate and conflict profiling. You see a count of exact duplicates, likely duplicates, and records that disagree, before any fix is chosen.
- Which-record-wins rules. You and Raquel decide, field by field, which system is the source of truth. The phone number might come from the field app, and the billing address from accounting.
- Cleanup in a working copy. The merge happens in a copy of the data, never in the live system first.
- Sample review. Your office manager or dispatcher checks a sample of merged records against what they know to be true.
- Load back. Approved records go back into the system of record, or into one new master file if you chose that.
- Keep-it-clean guardrails. Entry standards, required fields, and a recurring duplicate check, so the mess does not quietly return.
- Handover. A walkthrough, the data dictionary, and a short note on what is now safe to automate.
What do you walk away with?
- A map of where every record type lives today
- A field map across your systems, in a spreadsheet you can keep
- A duplicate and conflict report with counts, before and after
- Which-record-wins rules, written down and signed off
- A cleaned record set, loaded where you chose
- A one-page data dictionary for each record type
- Entry standards your office staff can follow at the front desk
- A recurring duplicate check and who runs it
- A short next-steps note: which workflows can now be automated safely
What do you get, and what is this not?
What you get
- One trusted version of each customer, site, or account
- Rules your team agreed to, not rules imposed from outside
- A record of every merge decision, so nothing disappears without a trace
What this is not
- Not a move to new software. Your existing tools stay in place.
- Not an automation build. Cleanup comes first; automating a workflow is a separate sprint.
- Not a bookkeeping reconciliation. Your bookkeeper or accountant still reviews anything that touches the books.
- No promise that every old record can be rescued. Some will be marked unknown rather than guessed.
What does your team need to bring?
- An owner or manager who can settle "which one is right" when two records disagree
- Export or read access to each system in scope
- The person who enters these records every day, for about two review sessions
- A short list of the reports or automations you want to run once the data is clean
Five questions that set the price
- How many systems hold the same records?
- Roughly how many records are in scope?
- Is it customers only, or also sites, equipment, jobs, and invoices?
- Does the clean set load back into live software, or land in one new master file?
- How many conflicts can only a person settle, like two billing addresses for one account?
What does it cost?
- What moves it
- The number of systems, record volume, how many record types, and whether the result loads back into live software
- Payment
- 50% to start, 50% at handover
What if it isn't worth it?
The free diagnostic tells you whether cleanup is the right first step. And nothing loads into your live software until your team signs off on a sample.
Founding cohort: data cleaning is one of the offers included in the founding client program, at 25% off launch pricing.
Questions people ask before booking
Can't our software vendor just remove the duplicates for us?
Most tools can merge duplicates inside their own walls. In a firm like the HVAC example, the trouble is between systems. QuickBooks has no idea what ServiceTitan thinks. A vendor tool also cannot decide which address is correct. That call belongs to someone who knows your customers, and this engagement gives them a clear way to make it.
Will you change anything in our live accounting file?
Not without your sign-off. Every merge happens in a working copy first. Records only go back after your team reviews a sample. Posted transactions, payments, and journal entries are left alone. Your bookkeeper reviews anything financial before it loads.
How much of our office staff's time does this take?
Plan on a kickoff call, quick answers as questions come up, and two review sessions with the person who enters the records. Raquel does the profiling, mapping, and rule drafting herself. Your staff check the work instead of doing it.
What happens to the data once it is clean?
It stays clean only if entry habits change. That is why the guardrails and duplicate check are part of the deliverable. From there, you can automate a workflow that bad records were blocking, or build reporting on top of it. Raquel will point out which one makes sense, or say that neither is needed yet.
Start with one honest look at how scattered your records are.
Free, about four minutes, and read by Raquel herself.