Carbon Accounting Software Trial: How to Test on Your Own Data

A carbon accounting software trial should use your own data, include a hard case and end with a comparison against what you already have, whether that is a spreadsheet or an adviser report. A demonstration on clean sample data shows how a product looks. A trial on your data shows whether it works for you. Agree the scope, the data, the test cases and the outcome you want before the trial starts, and ask the vendor to confirm them in writing. These steps apply to any product, including ours.

Choose the sample

Take one full period for a part of the business, for example two offices and a vehicle fleet. Include a mix of data: bills, fuel records and purchase lists. Choose a sample for which you already have a result from a spreadsheet or an adviser, so you can compare.

Add a hard case

Include something awkward: a missing bill, a site that changed address, an estimated meter reading, a supplier with only spend data. See how the product shows it, flags it and records what you did. A product that handles only clean data is a risk.

What to test in a carbon accounting software trial

Compare with your own result

Compare totals and, more importantly, lines. Where results differ, find out why. The cause may be a different factor set, a different boundary or an error in your spreadsheet. A good trial can find mistakes on either side.

Ask the vendor about the limits

Ask what the product does not do. Ask which standards and methods it supports and which it does not, and ask for each answer in writing. Where a vendor states a capability, ask to see it on your data.

Agree what happens after the trial

Ask what happens to your data, how long you can export it, and what a contract involves. Ask who supports you through your first period. Ask for pricing in writing, including what is and is not included.

Who should be in the trial

Include the person who will enter data, the person who will review and, if there is one, the person who will present the result to a customer or lender. Each will judge the product differently. Ask each to write down what they liked and what slowed them down, and collect the notes before the final meeting.

Write the success test first

Agree before the trial what a good result looks like: totals within a set tolerance of your known figure, estimated lines marked, one line traced to its source, and an export your reviewer can read. Decisions made against a test written in advance are easier to defend than decisions made on impressions.

An example of a trial plan

Illustrative example, not client proof: a finance lead agrees a trial on two offices and a fleet for one year, with a known spreadsheet result. The plan includes one missing bill and one spend-based line. The success test is that totals reconcile within an agreed tolerance, estimates are marked and a reviewer can trace one line to its source. The team compares results and notes two differences, one of which was a spreadsheet error.

Common mistakes

Judging on a demonstration is the first. Using only clean data is the second. Skipping the exit question is the third.

FAQ

How long should a trial be?

Long enough to load a full period and review it, which is often a few weeks for an SME. Agree this with the vendor.

What should I do if results differ?

Find the cause line by line. Differences often come from factor sets or boundaries.

Should my auditor see the trial?

If assurance matters, ask them what evidence they would expect and test for it.

Can I use the trial result in a report?

Only if you are satisfied with the data and the method. Treat it as a draft until reviewed.

What if the product fails the hard case?

Ask the vendor how it plans to handle it, and weigh that in your decision.

Method reference: the inventory principles behind these checks are set out in the GHG Protocol Corporate Standard.