Before you automate a report, find out if it is right.
Most companies have one report that drives what they order, how they schedule, or what they charge. It is rebuilt by hand every week, usually by one person, and no one has ever tested whether it is right. We test it — before anyone builds automation on top of it.
WHO THIS IS FOR.
If a report drives a real decision in your business, what to order, how many people to schedule, what to charge, which customers are actually profitable, and that report is pieced together in spreadsheets from more than one system, this is for you.
It is a good fit if at least one of these is true:
The report takes someone most of a day, every week.
Two people can produce different numbers for the same month.
Nobody is quite sure where one of the figures comes from any more.
The person who built it has left, or is about to.
You are being asked to automate it, and you are not confident the thing being automated is correct.
If you are not sure whether your report qualifies, it almost certainly does. Bring it to the call.
The weekly production report.
Built every Monday by a supervisor from the shop floor system, a maintenance log and two spreadsheets. It decides the week’s scheduling, and it has never been checked against the source.
The service level report
On-time performance pulled from an order system and a carrier portal that count "on time" differently. Leadership reviews it monthly. Nobody can say with confidence whether performance is improving.
Margin by customer
Revenue from one system, freight and handling from another, rebates from a spreadsheet somebody maintains by hand. It is used to decide which accounts to keep and what to charge them.
Three reports we see most often
What we do.
01 Trace the numbers
We follow each figure on the report back to the system it came from, and write down the path it takes to get there. In most cases this has never been documented.
02 Rebuild it independently
We reproduce the report from source data without using your existing spreadsheet. This is the part that finds things — an independent build and the original rarely agree on the first pass.
03 Quantify the gap
Where the two disagree, we say by how much, why, and whether it is a one-off or it happens every cycle.
04 Tell you what it means
We translate each gap into the decision it affected: units over-ordered, shifts over-staffed, margin reported that was not there.
What a finding looks like.
A typical finding
A distributor’s monthly margin report pulled revenue from the order system and freight cost from the carrier invoices. The two were matched on order date. But freight was billed on the date the shipment left, not the date the order was placed, so every order that crossed a month end carried its freight into the wrong month. On any single month it looked like noise. Over a quarter it moved reported margin on the largest customer group by enough to change which accounts the sales team was told to protect.
Nothing in that is exotic, and nobody did anything wrong. It is the ordinary way reports drift. It is also the kind of thing that is close to invisible from inside the spreadsheet and obvious the moment somebody rebuilds it from the source.
What you get.
A written report, in plain language, around eight to ten pages. No dashboard you need a licence to open.
A reconciliation that shows exactly where your current report and the independent rebuild diverge.
A ranked list of causes : data entry, gaps between systems, logic that was right once and is not any more, timing mismatches between sources.
A one-page recommendation: fix, rebuild, or automate. In that order, and often the answer is the first one.
A 60-minute presentation with you and whoever owns the report.
What we need from you.
The report itself, in whatever form it currently exists. A spreadsheet is fine. A PDF is fine.
A read-only extract of the underlying data for the period we agree. We do not need access to your systems, and we do not ask for credentials.
One hour with whoever builds the report, so we understand what they do and why.
About four hours of your team’s time in total, across two to three weeks. We do the rest.
Your data.
We sign your NDA, or ours, before anything is shared. Your data is handled in Canada and used for one purpose: this assessment. Nothing you send is used to train a model, and nothing is shared with any third party.
At the end of the engagement, your data is returned or destroyed - your choice, in writing.
What this is not.
It is not an audit. We express no opinion on your financial statements and we do not report to anyone but you.
It is not a software purchase. There is nothing to install and nothing to subscribe to.
It is not a pitch for a platform. If the answer is that your report is sound and you need nothing else from us, that is the answer you will get.
Timeline
Cost
Two to three weeks from the day we have the data. We need roughly four hours of your team’s time across that period — one kickoff conversation, one data handover, one walkthrough.
A fixed fee, quoted after a 30-minute call. You will know the scope, the price and what you will be holding at the end before any work starts.
Common Questions
-
Then you have something you did not have before: a documented, independently checked report you can defend when someone challenges it. That is a real outcome, and it is the one we hope for. It does not change the fee, because the work is the same either way.
-
No. The assessment is a complete piece of work with a complete deliverable. If the recommendation is to fix something small in-house, we will tell you that and you can do it yourselves.
-
No. Reports drift for ordinary reasons — a system changed, a definition moved, a workaround became permanent. The report names causes, not people, and the person who built it is usually the one most relieved to have it checked.
-
Yes, but we would rather not at the start. One report, done properly, tells you more about the state of your data than three done quickly.
Why Ainfore?
Ainfore is led by Luisa Ferreira. She spent twelve years building the traffic forecasts at Toronto Pearson and led data science delivery at Circana, where models ran in production for some of the largest consumer brands in North America. The thing that goes wrong is almost never the model. It is the data underneath it.
“Fix the data first. Then automate”
- Luisa Ferreira, Founder & CEO
Start with the report that matters most.
Bring the one report your decisions actually rest on. Thirty minutes is enough for us to tell you whether this is worth doing — and to say so if it is not.