The Very Big Spreadsheet

In a busy town, everybody needed the same list. It showed the business risks and what people were doing about each one.

The list lived with Big Sheet, a spreadsheet whose house was so crowded that finding anything took a flashlight.

Updates came by email and flew around like runaway envelopes. Some never came home. Leaders waited for slides made by hand. And whenever the auditors asked for proof, everyone went digging.

Everyone agreed that Big Sheet needed a new home. The team even wrote it on the first slide of their proposal: “From spreadsheets to an automated, auditable, role-based workflow.”

But a new home needs a blueprint first. Writing it would be my job.

I joined the team in late summer 2025 as the new helper. Before drawing a single wall, I wanted to learn how the family lived day to day.

The business owner looked after the list, and she showed me around. She knew exactly what the family needed, and she became my partner for the whole build.

Then I went to the library. I watched how the work happened, and I took a lot of pictures. My very first sketch still counted calories.

I made pretend copies of Big Sheet, full of made-up data, so the whole team could practice without touching real records.

Together we kept four groups of neighbors in mind, each with a need and a hurt.

Before ECS built the new house, we needed to agree what went inside. The team made a dashboard demo. I used Makey, an AI design helper, to build clickable prototypes: models we could review, not working apps. The business owner could try a screen, explain what was missing, and make a decision I could put into the requirements.

And then the shiniest idea of all rolled up.

That was the catch. Before an assistant could help with this register, people needed a reliable way to record the risks and updates. Our assistant screens showed ideas using sample data. They did not prove a connected service or reliable AI answers.

So we asked what the family needed on the first day. A front door comes before a chandelier. The team prioritized entering and updating the risk records. ECS would build that in Power Apps. The assistant stayed a proposal; it never shipped.

Ours had one big rule. The new home had to use the company’s Microsoft tools. SharePoint kept things, Teams was for talking, and Power Apps ran the app itself.

The builders were ECS, a skilled team working from another time zone. That meant the work kept moving even while our side of the world was asleep. Handy!

So the house had three partners. The business owner knew what the family needed. ECS knew how to build it. And I drew the plans and kept everyone on the same page. Before anyone broke ground, we also lined up the safety checks, like privacy reviews.

I started with the business owner’s spreadsheets and mapped how the work moved between people. Then I wrote the blueprint: who could do what, which fields each form needed, and what belonged in the release. ECS needed requirements it could build. The business owner needed a way to check that they matched her process.

A long blueprint was hard to discuss across time zones and different tools. I made new sample spreadsheets, flow diagrams and clickable Figma prototypes. We used them in calls and reviews, then I traced the feedback back into the spec. Its history records 18 versions; those are requirements revisions, not 18 releases.

I built little models: clickable prototypes and flow diagrams. I also made one reference pack. It put the old spreadsheets, the team’s dashboard, my screens and ECS’s first wireframes side by side. The business owner reviewed the behavior, ECS checked what it could build, and I updated the requirements. The models gave all three of us something concrete to point at.

Now we could point at real buttons instead of guessing. The business owner picked up her red pen.

Click a quarter, she wrote, and show its progress without making me edit. ECS said yes, and she approved the sketches with one fix.

Small words mattered too. Before one big review in January, my notes started with a thank-you to ECS for their progress on the wireframes. Then I suggested clearer names: “Audit Trail” became “Version History”, and the delete button got a trash can instead of a green check mark. Nobody should have to guess what a button does.

In one January session, we even agreed a new way to handle approvals, right there in the room. Compliance would approve, and the leadership team would watch. Reports would also export as PDFs, so nobody could change them after they left the app.

The models helped us agree what to build. They did not settle what would fit in the release. That needed traceable requests, effort estimates and the business owner’s decisions. Inspection day put that agreement to the test.

Just when the walls were going up, a sneaky little vine crept onto the building site.

The business asked for changes. I traced every one back to the blueprint or to the person who asked, so none of them shook the foundation. We agreed to save a few bigger things for the next phase, like the dashboard and a rule for flagging late work. That kept the first house solid. (Keep an eye on that late-work rule. It comes back.)

Then came inspection day. The business owner logged 43 observations in four days. ECS fixed 32 inside the window, but marked all four major items for Phase 2 without her agreement. A prototype could show the problem; it could not decide which work left the release.

Some were tiny and a little funny. One tester had to click an X five times to remove a single risk area.

I checked the build against the requirements and helped write the testing rules. ECS could propose a deferral with a reason, an effort estimate and its business value. The business owner had to approve it. I would step in when an item stayed open for two business days or disputed across two stand-ups. The owner kept the final decision.

In March 2026, Big Sheet moved in. ECS built the app in Power Apps and launched it. The blueprint was mine, the building was theirs, and the home belonged to all of us.

The first release was the built house. The pictures that follow are requirements and prototypes for changes we discussed after launch, not screenshots proving those changes shipped. Remember the red pen? The business owner needed each plan’s progress in her committee report, and the released app couldn’t show it yet.

I sketched a way to pick the report fields. The business owner said that was what she needed. I recorded the requirement for the next phase; I do not have a record of its later release.

After the US release, we planned a new wing for Global in the same app. The business owner described it as about 95% the same, with a few more risk areas and names. I specified separate records and access, plus a way for contributors to help without owning a risk. These were next-phase requirements, not a confirmed Global launch.

The blueprint was getting long again. So I made a picture guide for the whole crew. It was a plain-language deck with short words and one idea per slide. That way everyone could agree on the plan before building it.

My guide said a task was late if nobody touched it for a while. Two days later, the business owner spotted a problem. The work might be done, she said. Someone just hadn’t written it down yet.

She was right, and this is where I learned the most. Together we rewrote the rule around missing updates instead. Remember that late-work rule we saved for later? It came back with a kinder name.

Then came the busiest week of all. The proposed plan added policy exceptions. That workflow still needed Compliance's approval; it was separate from the already released risk register. “Deny” became “Decline”. On one day we worked through five new versions of the blueprint together.

At the end of April I finished my last version. In May I walked a teammate through the blueprint, started handing it over, and set off on my next adventure: an AI Workbench prototype. That’s a different book.