A data quality checklist is a structured list of checks you run on a dataset before using it — designed to catch completeness gaps, duplicate records, format errors, and consistency issues before they damage your reports, campaigns, or decisions.
Most data quality problems don't happen because people are careless. They happen because there's no process. No one checks for duplicates before importing a list. No one validates email formats before a campaign send. No one confirms that a new data file matches the expected schema before loading it into a system.
A data quality checklist gives your team a repeatable, low-effort process that catches most problems before they cause damage. This guide shows you exactly how to build one.
In this guide
- The five categories every data quality checklist should cover
- A ready-to-use template for each category
- How to customize the checklist for your most important datasets
- How to make the checklist a team habit, not a one-time task
- Common mistakes to avoid when building your first checklist
Why a Checklist Works Better Than Ad-Hoc Review
Ad-hoc data review is fragile. It depends on whoever is handling the data that day to remember what to check, know what "good" looks like, and have the time to do it thoroughly. That's three points of failure.
Sohovi profiles your datasets for quality issues in minutes — see what's broken before it breaks your pipeline — try Sohovi free.
A checklist removes the memory and judgment requirement. Once you've decided what to check, the checklist runs the same way every time — regardless of who's handling the data, how busy the day is, or whether the problem is obvious on first glance.
Airlines use pre-flight checklists not because pilots forget how to fly, but because a structured process is more reliable than memory under pressure. Data quality works the same way.
Category 1: Completeness Checks
Completeness checks verify that required fields actually have values.
Sohovi profiles every column in your dataset for completeness and flags the exact rows where values are missing — free to try.
Checklist items:
- [ ] Identify the "required" fields for this dataset's intended use
- [ ] Count the null/empty rate for each required field
- [ ] Flag any required field with more than 5% null rate
- [ ] Check whether null values are truly missing or are placeholders ("N/A", "none", "0")
- [ ] Verify row count matches the expected total from the source system
Example threshold: A customer email field used for campaign sends should be 98%+ complete. A phone number field used for SMS campaigns should be 95%+ complete. Set your thresholds based on what the data is actually being used for.
Category 2: Uniqueness / Deduplication Checks
Duplicate records are one of the most common and most expensive data quality problems. This check should run on every dataset that represents distinct entities — customers, leads, products, orders.
Sohovi automatically finds every duplicate in your dataset — including near-matches — and shows you exactly which rows are affected.
Checklist items:
- [ ] Identify the field(s) that should be unique for this entity (email for contacts, order ID for orders, SKU for products)
- [ ] Count exact duplicates on the unique identifier field
- [ ] Check for near-duplicates: same name with slightly different spelling, email with/without company domain
- [ ] Verify that a contact with multiple records doesn't have conflicting values (e.g., two different phone numbers)
- [ ] Flag any unique identifier field with a duplicate rate above 1%
Category 3: Validity / Format Checks
Validity checks confirm that values match the expected format or fall within an acceptable range.
Checklist items:
- [ ] Email addresses: do they match a valid format (contains @, has a domain)?
- [ ] Phone numbers: do they contain the expected number of digits for the region?
- [ ] Dates: are they in a consistent format? Are any dates outside a plausible range?
- [ ] Numeric fields: are values within an expected range (e.g., price > 0, age between 18–120)?
- [ ] Categorical fields (status, type, industry): do all values appear in the approved list?
- [ ] IDs and codes: do they match the expected pattern (e.g., invoice numbers start with "INV-")?
Category 4: Consistency Checks
Consistency checks catch cases where the same information is stored differently across fields or datasets.
Checklist items:
- [ ] Are state/country fields consistent (e.g., "California" vs. "CA" vs. "california")?
- [ ] Do related fields agree? (e.g., if status = "Active", is there a non-null start date?)
- [ ] If this file will be joined to another dataset, do the shared fields use the same format and casing?
- [ ] Are date fields in the same format across all rows and across all source files?
- [ ] Check for mixed encoding or special characters in text fields that could break downstream processing
Category 5: Schema / Structure Checks
Schema checks confirm that the file itself is structured as expected before any row-level checking begins.
Checklist items:
- [ ] Does the file have the expected number of columns?
- [ ] Are column names exactly as expected (no extra spaces, no typos, no reordering)?
- [ ] Does the file have a header row, and is it in the expected position?
- [ ] Are there any fully empty rows or columns?
- [ ] Does the encoding match what the receiving system expects (UTF-8 vs. Latin-1)?
- [ ] Is the delimiter consistent (all commas, all tabs — no mixed delimiters)?
Building Your Master Checklist Template
A practical master checklist combines the above categories into a single document. Here's a condensed version you can adapt:
Pre-Use Data Quality Checklist
| Check | Field/Column | Pass/Fail | Notes | |---|---|---|---| | Required fields complete (≥98%) | Email | | | | Required fields complete (≥95%) | Phone | | | | Unique identifier — duplicate rate (<1%) | Customer ID | | | | Email format validity | Email | | | | Date format consistency | Created Date | | | | Categorical values in approved list | Status | | | | Column count matches expected | All | | | | Row count matches source system | All | | |
Add or remove rows based on what matters for your specific dataset.
Sohovi automates most of these checks — upload a CSV and get instant results for completeness, uniqueness, validity, and format consistency across every column. It's a good way to run a first pass before working through a manual checklist for the edge cases a tool might miss.
Making the Checklist a Team Habit
A checklist only works if people actually use it. Here's how to make it stick:
Attach it to a trigger, not a schedule. The checklist should run every time a specific event happens — every time a file is imported, every time a new data source is connected, every time a campaign list is prepared. Not "once a month" — on a trigger.
Keep it short. A checklist that takes 45 minutes to run will be skipped. A checklist that takes 10 minutes will be used. Start with 5–8 checks on your most critical fields and expand from there.
Make it a gate, not a suggestion. The most effective data quality checklists are required before the next step — you can't import the file until the checklist is signed off. Even informal gates ("I need to see the completeness check before we send this campaign") work better than optional reviews.
Store it where the data lives. If your team works in Google Sheets, keep the checklist in Google Sheets. If they work in Notion, put it in Notion. A checklist no one can find is a checklist no one uses.
Common Mistakes to Avoid
Checking everything at once. A checklist that covers 40 different checks across all possible datasets is unusable. Build dataset-specific checklists — one for your customer list, one for your product catalog, one for your financial exports. Each should be focused and short.
Setting thresholds without thinking. "Zero errors allowed" sounds rigorous but isn't practical. A customer email field at 97% completeness is fine for most uses. Setting a 100% threshold means the checklist is always failing, so it stops being treated seriously. Set thresholds based on actual use requirements.
Skipping the schema check. Most teams jump straight to row-level checks and miss the structural problems that invalidate everything else. A file with a missing column, extra header rows, or a different encoding than expected will produce wrong results on every downstream check.
Never updating it. Datasets evolve. Checklists need to evolve with them. When you add a new field, update the checklist. When a threshold proves too strict or too lenient in practice, adjust it.
Frequently Asked Questions
Q: What is a data quality checklist? A data quality checklist is a structured list of checks you run on a dataset before using it — verifying completeness, uniqueness, validity, consistency, and structure. It standardizes what "good data" means for your team and makes quality review repeatable rather than dependent on individual judgment.
Q: What should a data quality checklist include? At minimum: completeness checks (are required fields populated?), uniqueness checks (are there duplicate records?), validity checks (are values in the right format?), consistency checks (is the same data stored the same way?), and schema checks (does the file have the expected structure?). Add dataset-specific checks based on what each dataset is actually used for.
Q: How often should I run a data quality checklist? Run it every time data is received, imported, or used in a high-stakes process. For recurring datasets (weekly exports, monthly reports, campaign lists), set a recurring review process. For one-off imports, run it before the import. The trigger is the event, not the calendar.
Q: How long should a data quality checklist take? A practical first-pass checklist should take 10–15 minutes for a typical business dataset. Automated tools can do most of the heavy lifting in under a minute. A full manual audit of a complex dataset might take 30–60 minutes, but that's an audit, not a routine checklist.
Q: Can I automate a data quality checklist? Yes, for the most common checks. Completeness rates, duplicate counts, email format validity, and date format consistency can all be automated. Tools like Sohovi handle these automatically when you upload a CSV. Manual judgment is still needed for business-rule-specific checks — whether a status value is appropriate given context, whether a date range makes business sense, and so on.
Q: What's the difference between a data quality checklist and a data quality audit? A checklist is a routine pre-use review — a short, repeatable process you run every time data moves through your systems. An audit is a deeper, periodic assessment of a dataset's overall health — typically done quarterly or before a major system change. The checklist is the daily habit; the audit is the periodic deep clean.
Q: Who should be responsible for running the data quality checklist? Whoever is working with the data. Data quality checklists work best when they're a standard part of the workflow for the person importing, using, or sharing the data — not something delegated to a separate "data quality" function. The person closest to the data is the first line of defense.
Q: What should I do when a checklist item fails? Document it, investigate the cause, and decide whether to proceed. Minor failures (a few invalid email formats in a large list) can often be cleaned quickly. Systemic failures (30% null rate on a critical field) should trigger an upstream investigation before the data is used. Never ignore a failing checklist item without a documented decision about how to handle it.
Q: How do I get my team to actually use a data quality checklist? Attach it to a workflow trigger, keep it short (5–8 items maximum for routine checks), and make it a gate — data doesn't move forward until the checklist is signed off. Store it where work already happens. The goal is zero friction between the checklist and the workflow it's protecting.
Q: Can a small business benefit from a data quality checklist? Absolutely — and small businesses often benefit more than large ones. Without a dedicated data team, a simple checklist is the most affordable, practical way to catch problems before they cause damage. A 10-item checklist that takes 10 minutes to run can prevent hours of downstream cleanup.
A data quality checklist won't prevent every problem. But it will catch the ones that happen repeatedly — the same format errors, the same duplicate patterns, the same missing fields — before they have a chance to damage your operations again.
If you're ready to run your first data quality check, Sohovi is free to try. Upload your most important CSV and get an instant quality breakdown — completeness, duplicates, format issues — across every column. No credit card, no setup, no code.