Questions about data cleaning and file organization

Well Kept is a digital housekeeping and data stewardship service. We clean up and maintain contact lists, CRMs, spreadsheets, trackers, shared drives, and file archives so the data in them is accurate, consistent, and easy to use.

Businesses and individuals whose contact lists, spreadsheets, or shared drives have become hard to trust or hard to navigate. You do not need a data team. If your records live in a spreadsheet, a CRM, or a shared drive, we can work with them.

Well Kept is based in Asheville, North Carolina. The walkthrough and the work happen remotely, so we work with clients outside the region too.

For now, just Mary Greene, a business strategist, data architect, and analyst. She spent years in agency sales and operations managing large datasets, implementing new software platforms, and migrating people and their data from one system to the next.

Data cleaning, also called data cleansing, is the process of finding and fixing data that is wrong, badly formatted, duplicated, or incomplete. In practice that means correcting typos and formats, merging duplicate records, filling gaps where the right answer can be confirmed, and removing records that no longer belong.

Data quality is usually judged on five measures:

Validity: entries follow the rules for that field, like a real date in a date column.

Accuracy: entries match the real world, like a current email address.

Completeness: the fields you need are filled in.

Consistency: the same fact matches across your systems, so your CRM and your email list agree.

Uniformity: the same kind of value is written the same way everywhere, like one phone number format.

Messy data shows up downstream: in a report that does not add up, a bounced email, a package sent to the wrong address, or a wrong answer from an AI tool. Cleaning it pays off in a few specific ways:

Better decisions, because reports and forecasts rest on accurate rows. Time back, because your team stops fixing the same errors by hand. Fewer costly mistakes, like resent mail or two people contacting the same client. Systems that share data without breaking. Less risk, because old copies of sensitive information get cleared out. And more accurate results from AI and automation tools.

Data cleaning fixes or removes data that is wrong or does not belong. Data transformation converts data from one format or structure into another, for example reshaping a spreadsheet so it can be imported into a new CRM. Most cleanups involve a little of both.

Data enrichment means making records more complete by adding related information from other sources. When we clean a contact list, that usually means finding company websites and LinkedIn profiles from public sources. We flag everything we assume or infer so you can hand-check as desired.

Data hygiene is the everyday side of data stewardship: the habits and rules that keep data clean after a cleanup, like naming conventions, required fields, and regular checks for duplicates.

It depends on how fast new data comes in. A list that grows through web forms, imports, and several people typing into it collects new errors every week. That is why we offer a monthly tidy after the first deep clean, so small messes are caught before they pile up.

Yes. Clean, consistently structured records export and import cleanly, so you can move from one CRM, email platform, or storage service to another without rebuilding your data from scratch. We think of this as platform sovereignty: your data belongs to you, not to the subscription it happens to live in.

Regulators treat this as a real concern. The FTC has noted that data portability can reduce the lock-in effect that keeps customers tied to one provider, and European privacy law gives people the right to receive their data in a structured, commonly used, machine-readable format.

When the meaning is held in the structure, fewer instructions are needed. A new hire can find the right proposal without asking for a link. An AI assistant can answer from the right files instead of five versions of the same one.

AI tools also follow your permissions. Microsoft says Copilot only surfaces files a person already has permission to view, and Google says Gemini has the same access to Workspace data you do. Clear folders and access rules shape what those tools can see and use.

You cannot protect what you cannot find. When every folder has a clear audience, it is easy to see who can open what, remove stray copies of sensitive files, and close access when someone leaves. Google's shared drive guidance recommends limited-access folders for sensitive content inside a drive.

Calmer. You stop hunting for files, stop wondering which list is current, and stop being the only person who knows where things are. Holiday cards, client follow-ups, and tax season become a matter of opening the right list or folder.

We work in the tools most small teams already use. Here is the current list, grouped by type:

Yes. We merge duplicate contacts and companies, standardize formats like names, phone numbers, and states, fill in missing company websites and LinkedIn profiles from public sources, and flag anything we cannot confirm. The same applies to Zoho CRM and Pipedrive.

If it can be read as a spreadsheet or a filing structure, we can help. Most tools can export to CSV or Excel, and most file systems are folders and file names underneath. Bring it to the walkthrough and we will tell you.

We fix formatting, merge duplicate records, and correct errors like misspelled email domains. Then we research public sources to fill in missing details, such as company websites and LinkedIn profiles.

For anything we cannot confirm through research, we make a smart, risk-aware inference and flag it for you to check or remove as appropriate.

One column, one kind of value. We clean up entries, standardize formats, write clear headers, and fix the structure that makes formulas break. Think of it like degunking the gears of a machine.

That is one of the main reasons to clean it. AI and automation tools learn from what you give them. Clean, consistent records and clearly named files make them more accurate and easier to use, for AI agents and for the people on your team.

Files take more planning than records. During intake we agree on a brief: the folder structure and the file naming rules that match how you use your files. Nothing moves until you approve it. Then we rename and sort every file to follow the brief, and flag anything we cannot place.

The structure itself follows a few well-tested rules. Google recommends giving each shared drive one clear purpose and setting access to match it. The UK Data Service advises keeping folder structures simple, predictable, and documented, with working files, finished files, and archives kept apart.

A file naming convention is a rule for how every file gets named, so anyone can tell what a file is before opening it and files sort in a useful order. One common pattern is Date_Who_What_Version, for example 2026-02-20_Acme_Proposal_v2. We agree on the convention with you during intake.

University data management guides, including Harvard Medical School's, recommend short names built from letters, numbers, dashes, and underscores, with no spaces or special characters.

Dates written as year, month, day, like 2026-02-20, sort in true time order in any folder. They are also unambiguous. The international standard for dates, ISO 8601, uses this order because 01/05/26 could mean January 5 or May 1 depending on who reads it.

Final, latest, new, and copy. They stop being true the moment a newer version arrives. The UK Data Service recommends against them. A date and a version number, like v2 or v03, say the same thing and stay accurate.

The short version: one row per record, one column per kind of fact, and one fact per cell. A widely cited paper on the topic, Data Organization in Spreadsheets by Karl Broman and Kara Woo, adds a few more rules: use one header row, write dates as YYYY-MM-DD, do not leave meaningful cells empty, do not use color to carry information, and be consistent in how you write every value.

Structured this way, a spreadsheet can be sorted, filtered, imported into another tool, and read correctly by an AI assistant.

A short companion sheet that explains each column: what it holds, what format it uses, and what the allowed values are. Broman and Woo recommend keeping one. It is how a new team member, or an AI tool, learns that Status means the stage of a deal and not whether a client paid.

Only the people who need them. Security standards call this the principle of least privilege: NIST defines it as limiting each person's access to the minimum needed for their work. In practice, that means folders grouped by audience, like a team drive, a leadership drive, and a client-facing drive, so confidential files are not buried inside folders everyone can open.

Every deep clean includes intake and covers one system: one contact list, one shared drive, or one set of spreadsheets.

Small is $600, for up to 16,000 cells (about 2,000 contacts with 8 columns each) or up to 500 files. Medium is $2,000, for up to 60,000 cells (about 7,500 contacts) or up to 1,500 files. Anything larger or more complex is quoted after the walkthrough.

Sizes and prices are a starting point. If your project does not fit cleanly into one of these, we will quote it custom after the walkthrough.

A cell is one field in one record. A list of 2,000 contacts with 8 columns each, such as name, company, email, and phone, is 16,000 cells.

Upkeep after a deep clean, for $200 a month. New mess arrives every week through imports, form fills, and new hires saving files their own way. The monthly tidy catches it before it piles up.

A $1,500 option for jobs too tangled to size up front or for ongoing, flexible relationships.

1. Walkthrough: free, 20 minutes, on a screen share. We see the system, count what is there, and quote a price.

2. Intake: we talk through your priorities and the problems you want solved. For files, we agree on the folder structure and naming rules before anything moves.

3. Deep clean: we do the work. You get the system back clean, plus a one-page guide for keeping it that way.

4. Monthly tidy: optional. We handle new mess each month before it piles up.

Yes. It is 20 minutes on a screen share, and you leave with a quote. Book a walkthrough.

An NDA is signed before we look at anything. We use read-only access wherever your system allows it, our access is removed the day the job ends, and you get a written handoff listing every change we made.

We use AI to work faster, but only after we have agreed with you on security measures and on the structure your data should follow. The judgment about what matters, and what gets cleaned and how, stays human.