You can conduct a meaningful data quality assessment in one day by focusing on a single dataset, working through five quality dimensions systematically, and producing a prioritized list of findings by end of day — ready to act on the next morning.
Most businesses put off data quality assessments because they picture them as multi-week projects with dedicated resources. They don't have to be. A focused one-day assessment on your most important dataset produces findings that take months of ad-hoc cleanup to replicate.
This guide gives you a structured schedule you can follow today.
The One-Day Structure
Morning (2–3 hours): Profiling and measurement
- Choose your dataset
- Define your standards
- Run the quality checks
- Capture raw findings
Sohovi applies your data quality rules automatically across the whole dataset and highlights every violation — so nothing slips through.
Afternoon (2–3 hours): Analysis and prioritization
- Categorize findings by type and severity
- Score by impact and effort
- Draft your recommendations
- Document your output
Total time: 4–6 hours of focused work.
Morning: Profiling and Measurement
Hour 1: Choose and Prepare Your Dataset
Pick the right dataset. Choose the single dataset with the most business impact — your customer contact database, your CRM pipeline data, your product catalog, or your financial records. Don't try to cover multiple datasets in one day.
Sohovi finds gaps, duplicates, and format errors in your CRM data — so your team is working from records they can trust.
Export and document the baseline. Export the dataset as a CSV. Note the row count, column count, and the source system and date of export. This baseline matters if you run future assessments on the same dataset.
Review the structure. Before looking at values, confirm:
- All expected columns are present
- Column names are correct
- There are no completely empty rows or columns
- The row count matches what you'd expect from the source system
Hour 2: Run Completeness and Uniqueness Checks
Completeness: For each critical field, calculate the percentage of non-null values. In a spreadsheet: =1 - COUNTBLANK(column)/ROWS(data). In a data quality tool, this is automatic.
Sohovi profiles every column in your dataset for completeness and flags the exact rows where values are missing — free to try.
Critical fields to check first:
- Primary contact identifier (email, phone, or customer ID)
- Name or company name
- Any field used for segmentation or filtering
Uniqueness: For your unique identifier field (email, customer ID, order number), count how many values appear more than once.
Use =COUNTIF($B:$B, B2)>1 across all rows to flag duplicates in a spreadsheet.
Note: total duplicate count, not just the percentage. "847 duplicate customer records" is more actionable than "4.2% duplicate rate."
Hour 3: Run Validity and Consistency Checks
Validity: Check whether values match the expected format.
- Emails: look for missing "@", multiple "@", no domain
- Phones: look for letters, fewer than 7 digits, more than 15 digits
- Dates: look for values that aren't recognized as dates, or dates outside plausible ranges
Consistency: Check for mixed representations of the same value.
- State/region: "California" vs. "CA" vs. "california"
- Status: "Active" vs. "active" vs. "ACTIVE"
- Country: "US" vs. "USA" vs. "United States"
List all distinct values in your key categorical fields using a pivot table or =UNIQUE(). Look for values that mean the same thing but are spelled differently.
Uploading to Sohovi gives you completeness rates, duplicate counts, format validation results, and categorical value distributions for every column automatically — completing Hours 2 and 3 in under 5 minutes.
Afternoon: Analysis and Prioritization
Hour 4: Categorize and Score Your Findings
For each finding from the morning, record:
- Field affected: which column has the problem?
- Problem type: completeness, uniqueness, validity, or consistency?
- Severity: what percentage of records are affected?
- Business impact: what breaks or degrades if this isn't fixed?
Then score each finding on a 1–3 scale for both impact and ease of fix. Use the two-axis framework from the prioritization guide:
- High impact + low effort = fix immediately
- High impact + high effort = schedule as a project
- Low impact + low effort = fix opportunistically
- Low impact + high effort = defer or accept
Hour 5: Draft Your Recommendations
For each finding, write:
- What the problem is
- How widespread it is (specific number/percentage, not just "significant")
- What fixing it requires (estimated effort)
- What the consequence of not fixing it is
- Your recommended action and timeline
Aim for 5–10 findings total. More than 10 suggests you've gone too broad — narrow to the findings that matter most.
Hour 6: Document and Share
Create a one-page assessment summary:
Data Quality Assessment: [Dataset Name] — [Date]
Dataset: [Name, row count, source] Assessment scope: [Which dimensions were checked]
Top findings:
- [Finding, severity, recommendation, priority]
- [Finding, severity, recommendation, priority] ...
Recommended next steps:
- Immediate (this week): [high-impact, low-effort items]
- Near-term (this month): [high-impact, high-effort items]
- Ongoing: [monitoring and prevention measures]
This document is your handoff to whoever will own the remediation work.
Frequently Asked Questions
Q: What is a data quality assessment? A data quality assessment is a structured evaluation of a dataset across multiple quality dimensions — completeness, uniqueness, validity, consistency, and timeliness — that produces a prioritized list of findings and recommendations. A one-day assessment focuses on the most important dataset and the most impactful findings.
Q: How much data can you realistically assess in one day? For a dataset up to 100,000 rows, a thorough one-day assessment is very achievable with the right tools. For larger datasets, use a representative sample (10% of rows) for the manual review portions while running automated checks on the full dataset.
Q: What's the difference between a one-day assessment and a full data quality audit? A one-day assessment is a targeted, rapid evaluation focused on your most important dataset and most critical findings. A full audit is more comprehensive — covering more datasets, more dimensions, and more systemic analysis. The one-day assessment is the right starting point for most businesses.
Q: What tools do I need to conduct a one-day data quality assessment? A spreadsheet tool (Google Sheets or Excel) and a data quality profiling tool are sufficient. The profiling tool handles the automated statistical analysis (completeness rates, duplicate counts, format validation), leaving you free to focus on analysis and prioritization.
Q: What if I find too many problems to address in the afternoon? Prioritize ruthlessly. A one-day assessment should produce 5–10 actionable findings, not 50. If you've identified more than 10 significant findings, apply the two-axis framework to narrow to the issues with the highest impact and lowest effort to fix.
Q: Who should be involved in a one-day data quality assessment? Ideally: the person who uses the data most (to provide business context on what matters), and the person responsible for maintaining the data (to understand what's fixable). For small businesses, this might be the same person.
Q: What do I do with the assessment findings the day after? Share the prioritized findings with the dataset owner and get agreement on who is responsible for each remediation item. Schedule the high-impact, high-effort items as formal projects. Start the high-impact, low-effort fixes immediately.
Q: Can I use a one-day assessment to justify a data quality investment? Yes — and this is one of the best uses of a one-day assessment. The specific findings (X% of email records invalid, Y duplicate customer records, Z% null rate on a critical field) translate directly into a cost-of-bad-data estimate that supports a business case for tools or process improvement.
Q: How do I know if my one-day assessment was thorough enough? If you checked all five quality dimensions for your key fields and documented specific, measurable findings with recommendations — it's thorough enough. A one-day assessment isn't meant to be exhaustive; it's meant to be actionable.
Q: How often should I conduct a data quality assessment? Quarterly for your most important datasets. After any significant system change. Before any major decision or campaign that depends on the data. The findings from each assessment inform the prevention work that reduces the findings at the next one.
The biggest obstacle to a data quality assessment isn't the work — it's starting. Block a day on your calendar, choose your most important dataset, and run through this structure. By end of day, you'll know more about the quality of your data than you ever have before.
Sohovi makes Hours 2 and 3 of the morning session instant — upload your CSV and get a complete quality profile in under a minute. Free to try, no credit card, no IT team, no code required.