The Experts Nobody Asked
The agency customers loved the dashboard. Then I showed it to the people taking the orders.
The agency customers loved my dashboard, which was encouraging right up until the reps started screaming at me.
I’d spent months reviewing it every week with a broad group of agency users who represented a major share of this wholesale platform’s revenue. They wanted it built. I had advisory sessions with reps too, but I’d been told to wait until the agencies were happy before showing it there. Once I got clearance, the reps needed only minutes to deliver their review.
The platform takes orders: manufacturers make the products, retailers buy them, reps sell them, and agencies employ the reps. One agency user might have hundreds of reps underneath them. Agencies and manufacturers pay; reps and retailers get enabled as features.
I’d asked the people paying what mattered and built a dashboard full of ways to investigate their business. The reps wanted to find problems, fix them, and know what they were getting paid. My dashboard gave them a research assignment.
These people were on the road, balancing laptops, notepads, chargers, and folders of printouts in a Toyota Corolla. Preparing for a trip meant filtering tables, copying addresses, and making a route in Google Maps. In each parking lot, they’d dig through orders and write down products before walking in. Then do it again at the next stop.
The iPad app for taking the order was excellent, even with a bad connection. Can you spot the problem? Getting to the order, and dealing with whatever went wrong afterward, was the rep’s problem. This expensive-ass platform had nailed the bit in the middle.
I started over, made headway around version six, and showed the agencies the work by versions seven and eight. Some couldn’t see why reps needed their own experience. Newer reps would be better with tables, apparently. As though getting faster at copying things makes copying things a useful job.
An agency owner knew how to run an agency; that didn’t tell me what a rep needed before the next visit. I’d let one person’s expertise answer for both, then accepted that the reps should wait until the direction was settled.
We had reps on the research calendar all along. I’d helped make sure the people in the parking lot got their turn after we’d decided what the people in the office wanted.
That’s the short version.
The remaining 7 minutes follow the rep’s working day, the dashboard decisions it exposed, and my part in keeping the wrong people in charge of the answer.
Before I can do useful product design, there’s a question that needs answering: who are the experts? A company can have plenty of them, each able to do a body of work and handle certain kinds of problems. The job is to understand what makes that expertise specific and carry it into the personas and the experience you’re designing.
That’s easier said than done. Telling the expertise of experts A through F apart from the expertise of experts G through J takes observation at the right moments. If you can’t answer that question, every decision afterward inherits whatever answer you stumbled into. Ask me to design a product for musicians who want to do X, and if we decide X means finding a band, we’ve already dropped everyone whose work is solo. They’re gone before the first screen exists.
I’d answered the question. I was convinced the agencies were the experts. Everyone around me was convinced too, and the advisory board was built to match.
So I asked the reps what they cared about, since it wasn’t graphs, detailed tables, or complicated reports.
Problems. Finding them, fixing them, and getting paid, including knowing how much. If they’re getting paid less than they expect, they want to know why. If an order has a problem, they want to know before it spreads to the retailer or the manufacturer. A table of problems would help more than knowing their top retailers or their top products. They didn’t want to do an analysis when they logged in. They wanted to find where the problems were and fix them.
Look at the dashboard through that lens and it’s obvious it wasn’t built for those tasks. It was built for analysis, investigation, and deeper analysis. It was perfect for an agency owner tracking retailer and salesperson performance, or which manufacturers were slow to ship, or which products were moving. It had every dimension you could want, letting an agency manager stop sifting through data inside the platform, filter it on the dashboard instead, and jump straight to the source. For the person it was built for, it was good work.
The reps were objecting to the work it took to turn those rows into action. An agency owner opened the dashboard to investigate the business; a rep opened it because a retailer was about to call about an order, or because they had ten minutes before a visit. Giving them the same analytical starting point made sense only if you ignored why each of them was there.
The more time I spent with reps, the clearer it got that the platform had been built with very little understanding of how a rep actually works. Reps make the sales that let an agency afford its licenses, yet the hundreds of people working under an agency got an experience built for their boss.
Here’s how a rep actually works.
Reps are mobile. Their agency employers work from an office and, outside of market events, mostly stay in it; reps spend a lot of their week on the road. So when they say they don’t want Excel-like tables, take them literally. Reading rows of data in a midsize sedan outside Dallas to decide which retailers to swing by over the next two days isn’t useful. They couldn’t understand why this expensive-ass platform made them write down and print out so much.
Planning a trip from A to B means filtering a table down to retailers that haven’t ordered in sixty days. Then filtering those by city or region. Then opening each retailer and copying or writing down the address. Then, for the more tech-adept, building a Google Maps list of locations and working out where to go and when. Then fueling up and driving to a retailer just outside the Dallas city limits.
From the parking lot, they sift through tables of retailers to find the one whose parking lot they’re sitting in. They pull up that retailer, sort its orders by recency, open each of the last three, and write down the top products from each, while keeping a running list in their head of new products this retailer has never ordered but probably would if someone told them about it. Then they walk in. Then the process repeats at the next parking lot.
Ordering itself is done through a custom iPad app built to survive bad data connections. Reps loved it when it came out and still love it. It takes every kind of order in every kind of environment, shows retailers what products look like, and lets them browse manufacturer-managed catalogs. It’s about as good as an order-taking tool gets. Can you spot the problem?
The platform built a tool that’s perfect for the moment an order gets taken, and getting a rep to that moment is a tangle of handwritten notes, Google Maps, and a working knowledge of a platform optimized for someone else. Nobody had worked through how the people involved in an order find each other, deal with each other, or solve problems for each other. The platform did orders. Getting to an order, or handling what went wrong afterward, was largely left to the rep. And none of the reporting is worth much if a rep doesn’t know where to go, how to get there, what to do when they arrive, or how to handle a call about an order they took an hour ago while driving to the next one.
My dashboard couldn’t close those gaps by getting easier to read on an iPad. It started from an agency manager’s questions, and rearranging the answers would have left that choice in place.
So I started over for the reps. The conversations had changed the questions the product needed to answer: what needs my attention right now, which retailer is worth visiting, what should I know before I walk in, and what’s changed about an order or the money I expect to earn? I iterated concept after concept, trying to design something compatible with a rep’s day instead of a rep’s manager’s day.
What came out of it started from the day, not the data. At login, a rep saw a launchpad: the day’s work as a short list, ordered by what would cost them if it waited. Orders a supplier had put on hold came first, with the money they were holding up. Accounts that had stopped ordering came next, counted as commission at risk rather than drawn as a chart. Then the visits to plan, the orders waiting on approval, and a campaign ready to send. Reports, including a forecast of commission, were still there, last on the list.
Problems got one place. Every held order sat in one table, and opening a row showed why it was held and what was on it, beside the ways to act on it. The point was to hear about a held order before the retailer called about it.
Trips started on a map. Instead of filtering a table to retailers who hadn’t ordered in sixty days and copying addresses into Google Maps, a rep drew a shape around part of a city, ticked the retailers inside it, and had a visit list.
The parking lot got a page of its own. A retailer’s page opened with the address and how far away it was, the last visit, and the contact, with buttons to call or email. Below that came the recent orders, the account notes (when the owner prefers visits, how often they reorder), open follow-ups, and a short note on what to bring up. That replaced the three orders a rep used to open one at a time and copy into a notepad.
In the advisory sessions, each demo got about five minutes and four questions. How do you start your day now? How do you plan a route? How often do orders get stuck? Would this change how you work, or is it just nice to have?
Somewhere around the sixth version, I started making headway. By the seventh and eighth, I took the risk of showing the findings and the results to the agency users.
They were, to put it plainly, unsure why I’d waste time on a separate experience for their reps. They didn’t understand why tables were such a problem. A few suggested the newer reps were far more tech-literate and shouldn’t struggle with tables at all, which came close to saying the legacy reps would be phased out anyway.
Technical literacy could help someone operate the existing platform faster, but it couldn’t explain why getting ready for one visit meant reconstructing the same information from several places. Being better at the workaround doesn’t make the work disappear.
I couldn’t find an agency user in those sessions who understood what a rep’s week looked like. Many had been reps themselves, but long enough ago that they’d forgotten the balancing act of a laptop, a spiral notepad, a charger, and folders of printouts crammed into the front seats of a Toyota Corolla.
Their expertise was real, and so were its limits. An agency owner knew things about staffing, team performance, and running an agency that the reps couldn’t tell me. A rep knew things about preparing for a visit, working between appointments, and handling a retailer’s problem on the way to the next one. Treating the agency as the expert for both jobs let knowledge of one stand in for observation of the other.
That gap didn’t come from anyone being careless. It came from a decision made so long ago that nobody remembered it as a decision. At some point in the platform’s past, agencies were named the experts, and it stuck long enough that they came to be treated as experts in their own workflows and their reps’ workflows. Everything since had inherited it.
I helped keep it going. I can point to the instruction to hold off on the rep review, but I’d also accepted that the dashboard’s direction should be settled before the reps saw it. By the time they were invited to react, the work already had an answer to whose needs mattered most. That’s why a user group somewhere on the research calendar isn’t enough. Timing decides what their expertise is allowed to change.
The rep had always been somewhere outside Dallas, getting ready for the next visit. We’d been designing as though the important work happened back at the office.