Why Pre-Project Audits Are Non-Negotiable
You've agreed on a scope and a price. Then you open the client's data — and discover it's far messier than described. Now you're in an awkward position: either absorb the extra work or have an uncomfortable conversation about scope creep.
A pre-project data audit prevents this situation entirely. It's standard due diligence, and clients who are serious about their projects will respect it.
What You're Auditing For
The pre-project audit isn't as exhaustive as a formal data quality audit. You're answering three questions:
- Is the data complete? Does it cover the full scope of the project?
- Is the data usable? Are there issues severe enough that you can't work with it as-is?
- Is the data what the client thinks it is? Often, clients overestimate the quality of their data.
You're not fixing anything at this stage. You're assessing.
See exactly what's wrong with your data — try Sohovi free — try Sohovi free.
The 15-Minute Pre-Project Audit
Step 1 (3 min): Open the file. Note the number of rows, columns, and tabs. What's the date range? What does the client say this file contains?
Step 2 (5 min): Scan each column. For each one: is the type consistent? Are there blanks? Does the column name match the content?
Step 3 (4 min): Check for duplicates in the primary identifier. Count distinct values in key columns. Do the numbers feel right?
Sohovi automatically finds every duplicate in your dataset — including near-matches — and shows you exactly which rows are affected.
Step 4 (3 min): Write a 3-sentence summary of your findings. What's good, what's concerning, what needs clarification?
Turning Findings Into Scope Protection
After the audit, email the client:
"I've done a quick review of the data you sent. Good news: the core data is solid. A few things worth noting before we start:
- The Email column is blank for about 15% of records
- There appear to be ~200 duplicate customer IDs
- Date formats are inconsistent across 3 columns
My quote assumed clean, ready-to-use data. Cleaning these issues would add approximately [X hours] to the project. Can we discuss how you'd like to handle this?"
This email does three things: it establishes your professionalism, it documents the pre-existing data issues, and it creates space to adjust scope or price before work begins.
When to Walk Away
Some data is too broken to work with within the agreed scope and budget. Signs you might need to renegotiate or decline:
- More than 30% of rows have critical missing values
- The data structure doesn't match what was described
- Multiple conflicting versions of the same dataset
- No documentation of what the data represents
Declining gracefully is fine: "After reviewing your data, I don't think I can deliver what you're expecting at the quoted price. Here's what I found, and here's what it would take to address it. I'd be happy to re-quote if you'd like to proceed."
