Skip to main content
Blog

How to use Excel Agents for Batch Payment testing

  • July 23, 2026
  • 0 replies
  • 1 view
Tips & Tricks

Hi everyone! Sangeetha here, Senior Product Marketing Manager at DataSnipper, and welcome to this week's Tips & Tricks.

Quick intro for anyone who's new here. Before I moved into product marketing, I spent years as an auditor. So when I write these, I'm writing about the tasks I actually did, not features I read about on a slide. Batch payment testing is one of those tasks, and it's the one I want to go deep on this week.

If you've done it, you know the shape of it. You have a batch payment (one lump sum that covers many invoices), and you need to prove two things. First, that every invoice inside the batch is legitimate and adds up to the batch total. Second, that the batch actually left the bank. That means pulling together the batch payment schedule, the individual invoices behind it, and the bank statement, then reconciling across all three. During busy season, with dozens of batches, this is exactly the kind of repetitive, high-volume work that eats your week.

This is where Excel Agents come in.

What Excel Agents actually do

Excel Agents bring the work into Excel and do it for you based on a plain-language instruction. Instead of stepping through each action yourself, you describe the outcome you want and the Agent executes the full procedure end to end. It reads your source documents, reconciles the data, applies the audit and finance logic, and returns output that's explainable and traceable back to the evidence. Everything it produces is meant to be reviewed, not taken on faith.

Before you start, make sure you've got the setup right. You'll need DataSnipper v26.1 or later, an Accelerate or Elevate package, and an internet connection. If any of those are missing, sort that out first, because the Agent won't appear otherwise.

Running a batch payment test, step by step

Here's the full workflow.

  1. Open your Batch Payment Testing workpaper in Excel and import the relevant supporting documents. For this procedure that's usually the batch payment schedules, the individual invoices that make up each batch, and the bank statement showing the payments cleared.
  2. Click the Excel Agent icon in the DataSnipper ribbon to initiate the Agent.
  3. Enter your instruction in the prompt. This is where you tell the Agent what to do, in your own words. A strong instruction for this test might read: "For each sample, match every individual invoice to the batch payment schedule, confirm the sum of the invoices equals the batch total, then agree the batch total to the bank statement. Flag any invoice that doesn't match and any batch where the totals disagree."
  4. Once the Agent has your instruction it gets to work, and when it's ready to document its output it pauses and asks for your go-ahead before updating the Excel. At that point you can approve, or add more instructions before it writes anything. This is a deliberate checkpoint, not a formality. Nothing lands in your workpaper until you say yes, so you're always in control of what goes in.
  5. Let it run and read the summary. Once the procedure finishes, the Agent gives you a detailed summary of the actions it completed.
  6. Review the output in the workpaper. Your workpaper now includes all the generated outputs, along with the relevant snips and Excel formulas wherever they're needed. Because the snips link straight to the source documents, you can click any figure and see the exact invoice, statement line, or bank entry behind it.

Where the auditor stays in control

I want to spend a minute here, because it's the difference between a shortcut and a defensible workpaper.

The Agent gets you to a complete, traceable draft quickly. It does not sign off the test. That's still your call, and it should be. Use your professional judgment to review what the Agent produced. Walk the flagged exceptions and decide whether they're real issues or timing differences. Spot-check a sample of the snips to confirm they point to the right evidence. Make any edits the engagement requires. Humans stay at the judgment points, and the Agent's job is to get you to those points faster with the evidence trail already attached.

This is also why traceability matters so much. When a reviewer or a regulator asks how you reached a conclusion, you can show the snip, the formula, and the source document behind every number. That's a far better position than "the tool did it."

Getting sharper prompts (and better output)

The single biggest thing that changes your results is how you write the prompt. A vague instruction gives you vague work. A specific one, with the reconciliation logic, the fields to match on, and the tolerance you'll accept, gets you output that's close to review-ready on the first pass. If you want to get good at this quickly, DataSnipper has a dedicated article on structuring prompts effectively, and it's worth ten minutes of your time.

Building this into your team's practice

If you're leading an engagement, standardize the prompt. Write one good batch payment testing instruction, agree it with your team, and reuse it so everyone's output looks the same and reviews faster. Pair that with a consistent workpaper structure and a quick review checklist covering exceptions and snip spot-checks. That combination gives you speed without giving up consistency, which is usually the thing that worries reviewers most about any kind of automation.

Want the full walkthrough, including the video demo? Here's the knowledge base article: https://knowledge.datasnipper.com/en/articles/741967-how-to-efficiently-perform-batch-payment-testing-with-excel-agents

If you've run a batch payment test with Excel Agents, I'd genuinely like to hear how it went. Drop it in the comments.

I started this series because I've seen how one small tip can completely change the way someone uses DataSnipper. Every week, I'll share something practical you can put to work right away. And if there's a feature or workflow you'd like me to cover next, let me know in the comments. Visit our knowledge base if you'd like to learn more.