Data Quality Insights
Tutorials, best practices, and real-world guides — all privacy-first.
Featured
Data Validation in Google Sheets: Dropdowns, Rules, and Protected Ranges
The fastest way to prevent bad data in a shared sheet: Add a dropdown list to any column by selecting the column → Data → Data validation → Criteria: List of items → enter your valid values. Collaborators can only enter values from the list.
Latest articles
- Jun 14, 2026
Practical How-To Guides
UTF-8 Characters Look Broken in Excel (é, ’): The Encoding Fix
The diagnosis: When you see `é` instead of `é`, or `’` instead of `'`, Excel is reading a UTF-8 encoded file as if it were Windows-1252 (Latin-1) encoded. Each multi-byte UTF-8 character is being interpreted as separate Latin-1 characters — producing garbled output.
- Jun 14, 2026
Practical How-To Guides
How to Convert JSON to Excel Without Coding (3 Methods)
The fastest method for a simple JSON array: Use Power Query in Excel — Data → Get Data → From File → From JSON → select your file → Load to worksheet. This works natively in Excel 365 and Excel 2016+ without any add-ins.
- Jun 14, 2026
Practical How-To Guides
How to Find Duplicates Across Two CSV Files (4 Methods)
The fastest approach for non-technical users: Open both CSVs in Excel or Google Sheets, combine them into one sheet, then run VLOOKUP (or COUNTIF) to find records that appear in both files. For fuzzy matches across files — "Jon Smith" in file 1 matching "John Smith" in file 2 — you need a fuzzy…
- Jun 14, 2026
Practical How-To Guides
What Does This Excel Formula Do? Decoding Nested IFs, VLOOKUP, and INDEX/MATCH
The fastest approach for any formula you don't understand: Paste it into Sohovi's Excel Formula Explainer — it gives you a plain-English breakdown of what each part does.
- Jun 14, 2026
Practical How-To Guides
CSV to JSON: Arrays, Objects, and Which Format APIs Actually Expect
The core confusion: Converting CSV to JSON seems simple, but JSON has multiple valid structures for tabular data. Most APIs expect one specific format, and sending the wrong one causes a 400 error even though your data is correct. Here's exactly which format to use and how to produce it.
- Jun 14, 2026
Practical How-To Guides
Conditional Formatting Recipes for Spotting Bad Data Instantly
Conditional formatting turns a spreadsheet into a live quality audit. Instead of writing formulas column by column, you define a rule once and every problem glows red. Here are the 6 most useful recipes for data quality work.
- Jun 14, 2026
Comparisons
Informatica Data Quality Alternatives for Small and Mid-Size Businesses
The core problem: Informatica Data Quality (IDQ) is one of the most capable data quality platforms on the market. It's also priced for Fortune 500 companies with dedicated data governance teams. Most SMBs get sticker shock at the quote and start googling. Here are the realistic alternatives.
- Jun 14, 2026
Practical How-To Guides
Test Data vs Production Data: Why You Should Never Test with Real PII
The short answer: Using real customer data in development, testing, or staging environments creates privacy risk (data breaches in less-secure environments), legal risk (GDPR requires appropriate security for personal data at all times), and compliance risk (many auditors will cite…
- Jun 14, 2026
Tools, Technology & Buying Guides
Data Quality Tool Pricing Compared: What You Actually Pay in 2026
The problem with data quality tool pricing: Most enterprise tools use "contact us" pages because the actual price depends on data volume, number of users, integrations, and negotiation leverage. This post gives you realistic ranges based on publicly available information, analyst reports, and…
- Jun 14, 2026
Practical How-To Guides
How to Put a CSV Table into GitHub README or Notion
Fastest method: Use Sohovi's CSV to Markdown converter — paste your CSV or upload the file, and get GitHub-compatible Markdown table syntax in seconds. Copy and paste into your README or Notion page.
- Jun 14, 2026
Practical How-To Guides
Excel's 1,048,576 Row Limit: How to Work with Bigger Files
The limit: Excel can display a maximum of 1,048,576 rows (2^20) per sheet. Files with more rows open, but Excel silently truncates them — you see no warning, just missing data. For any file that exceeds this limit, you need a different approach.
- Jun 14, 2026
Comparisons
Data Ladder and WinPure Alternatives: Modern Dedupe Options (2026)
The issue with Data Ladder and WinPure: Both are capable deduplication tools with solid fuzzy matching algorithms. Both are also Windows-only desktop applications with architectures that feel like 2010. If you're on a Mac, work in a browser-first environment, or want a tool that doesn't require…
- Jun 14, 2026
Practical How-To Guides
How to Generate Realistic Fake Customer Data for Testing
The fastest method: Use Sohovi's test data generator — choose your fields (name, email, phone, address, company, date, etc.), set the row count (up to 100k), and download a CSV. No code, no account required.
- Jun 14, 2026
Practical How-To Guides
How to Reorder, Rename, and Drop CSV Columns in Bulk
Fastest method: Use Sohovi's CSV column picker — upload your file, drag columns to reorder, rename headers, check or uncheck to include/exclude columns, and download the modified CSV. No Excel, no code.
- Jun 14, 2026
Practical How-To Guides
How to Split One Huge CSV into Smaller Files (Without Excel)
The fastest method for Mac/Linux: `split -l 10000 data.csv chunk_ && for f in chunk_*; do (head -1 data.csv && cat $f) > $f.csv; done`
- Jun 14, 2026
Practical How-To Guides
Seed Data for Demos: Generating 100k Realistic Rows in Seconds
For demos without code: Use Sohovi's test data generator — choose your fields, enter 100,000 as the row count, click Generate. Your CSV is ready to download in seconds.
- Jun 14, 2026
Practical How-To Guides
Why 'Remove Duplicates' in Excel Misses Most Real-World Duplicates
The answer in one sentence: Excel's Remove Duplicates uses exact string matching — two cells must be byte-for-byte identical to be considered duplicates. Real-world data is almost never that clean.
- Jun 14, 2026
Comparisons
Soda Alternatives for Teams Without Data Engineers
The quick answer: If Soda is too technical for your team, the most practical alternatives are Sohovi (browser-based, no code), OpenRefine (free, local, some learning curve), or Excel/Power Query (if you're already there). Soda is SQL-first and assumes database access — if your data lives in files…
- Jun 14, 2026
Comparisons
Sohovi vs Excel for Data Quality: What Excel Can't Do
Short answer: Excel handles basic data quality checks well for small, infrequent, simple files. Sohovi is better when you need to check the same file structure repeatedly, when the data is messy (fuzzy duplicates, mixed formats, encoding issues), or when you need a quality report you can share.…
- Jun 14, 2026
Comparisons
Sohovi vs Great Expectations: No-Code vs Code-First Data Quality
The verdict up front: Great Expectations is the industry standard for data engineering teams who want to define quality rules in Python and run them in automated pipelines. Sohovi is for business users — analysts, ops teams, marketers — who need to profile, validate, and clean data without writing…
- Jun 14, 2026
Comparisons
Monte Carlo Alternatives for Small Teams (Data Observability Without the Price Tag)
The honest take: Monte Carlo is one of the best data observability platforms available. It's also priced for large data teams at enterprise companies. If you have a data team of 1–5 people, the ROI math rarely works out. Here are the alternatives that deliver meaningful observability without the…
- Jun 12, 2026
Tools, Technology & Buying Guides
Should You Let AI Clean Your Data? Where It Works, Where It Quietly Destroys Things
Balanced verdict: AI is excellent at suggesting cleaning approaches. It's dangerous when applying changes unsupervised — because it can silently change values it shouldn't, hallucinate plausible-looking fills for missing data, and behave inconsistently across large files. Use AI as an advisor; use…
- Jun 12, 2026
Practical How-To Guides
How to Anonymize a CSV Before Sharing It (So It Can't Be Re-Identified)
The counterintuitive fact: Deleting the name column is not anonymization. A dataset with ZIP code, age, and gender is enough to uniquely identify most individuals — researchers have shown that three quasi-identifiers can re-identify 87% of Americans. True anonymization requires systematically…