Welcome back to our series on dealing with imperfect data. Last week, we discussed how to turn bare-bones datasets into directional recommendations that help stakeholders generally understand where opportunities and issues lie. But those tips and tricks will only take you so far.
Relying on thin datasets is a great way to begin an analysis and generate preliminary insights. Now, it’s time to expand past those initial findings by securing more high-quality data. There are several ways we can expand datasets, but today we’ll be focusing on three primary methods: Targeted Customer Surveys, Internal Operational Data Fusion, and External Third-Party Enrichment.
By using these methods of data collection and integration, you’ll be able to continue delivering value to your clients while a more comprehensive data pipeline is being built.
Let’s dive in!
Three Agile Expansion Channels
1. Ask Users Directly (Targeted Quick Surveys) When your spreadsheet lacks crucial context, ask users directly. For example, if you don’t know how satisfied recently attrited customers were or which competitor they switched to, the easiest way to gain that understanding is to send them a survey.
Keep it Gated: Put 1 or 2 screening questions right at the start (e.g., “Did you cancel in the last 30 days?”) so you don’t waste time analyzing responses from the wrong people.
Use Simple 1-to-5 Scales: Ask users to rate their experience on a 1-to-5 scale. Number scales turn squishy personal opinions into clean, measurable averages.
Ask Them Why: Scales are a great way to quantify feelings, but they mean nothing if you can’t identify the root cause behind the number.
Watch Out for Trick Questions: Avoid poor-quality queries like double-barreled questions that ask two things at once (e.g., “Was our app cheap and easy to use?”) or pushy wording that forces people toward a specific answer.
2. Connect the Dots Inside Your Company: Your company is likely collecting great data right now that never makes it into the main data warehouse. Customer service reps log complaints, sales reps take notes on lost deals, and marketing tracks email clicks. Invest time in finding this data! Instead of waiting for engineers to connect those systems automatically, ask those teams for a quick CSV export. Matching customer email addresses or User IDs across spreadsheets lets you see if the 40 people in your small sample also submitted support complaints before canceling.
3. Look Outside Your Company (Public & Third-Party Data) :You don’t always have to collect new data yourself. Plenty of it is already published for free. You can grab census stats, regional economic reports, or industry benchmark studies. Match this third-party data to your small sample using basic functions like XLOOKUP in Excel or Google Sheets (matching by ZIP code or industry category). Here’s a video that shows you how easy and valuable lookup functions can be. Comparing your small numbers against industry standards instantly shows whether your problem is unique to your company or happening everywhere.
Theory to Practice: The Analyst’s Playbook
Now that you know what tools you’ll be using to expand your dataset, it’s time to put a strategy into play. You will have the most success if you break this process down into three distinct steps:
Step 1: Isolate the Missing Variable. Before you jump into gathering extra data or writing customer surveys, pause and inspect your core hypothesis. Ask yourself: “What exact piece of information is missing from my dataset that stops me from giving leadership a reliable answer today?” Pinpointing the single, critical gap keeps your expansion laser-focused. You avoid wasting valuable time gathering massive amounts of extra noise, ensuring you collect only the precise variable needed to unlock an immediate, risk-managed decision.
Step 2: Select the Lightest-Friction Expansion Channel. Always look for the path of least resistance before bothering customers or clients with fresh requests. If the missing detail already sits in an internal spreadsheet owned by customer service, sales, or marketing, tap those channels first. If the data can be pulled from free public databases, go get it. Direct surveys take time to design, test, and collect; using existing internal logs or public benchmarks lets you plug your data gaps in hours or days rather than weeks, keeping your analytical momentum moving forward.
Step 3: Stitch Your Interim Dataset Together. Once you gather your quick expansion data, connect it directly back to your small starting baseline so you can analyze everything as one unified picture. You don’t need a fancy automated database pipeline to do this. Again, simple, reliable matching functions in your spreadsheet will do the trick. Standardize common identifiers across your files and merge the new attributes directly into your baseline spreadsheet. Perfect match keys include anonymized Customer_ID, email address, order number, or ZIP_Code.
This creates a lightweight “interim dataset” that gives you immediate cross-system visibility into user behavior without waiting on IT.
Conclusion
You have now created a baseline using small data effectively, and you’ve created the most comprehensive dataset possible using just a small amount of resources. By doing this, you provide immediate strategic value to your stakeholders long before they could create a workable data pipeline.
The next step? Helping coach those clients or executives on how best to build that permanent, automated pipeline.
In the meantime, Keep Analyzing!
