Written by Jason Donnay, VP of Sales, following this year’s ARMCUMS conference
Behind the Pillar: What ARMCUMS Taught Me About Decisions, Not Technology
The organizers put our table in a corner, behind a big pillar. I assumed we would spend the morning watching people walk past.
I was wrong, and the reason is more useful than the booth story. The conversations that mattered most did not start with a demo. They started with someone telling me how their operation actually runs, where it breaks, and what it costs when it does. I mostly listened.
By the end of the day, one pattern was hard to miss. The strongest operators in the room were not shopping for technology. They were trying to stop making the same decision over and over.
This retrospective is built around that idea. I will walk through what I heard using three lenses I bring to every campus conversation: operational efficiency, risk mitigation, and campus experience. Then I will cover what operators told me they actually need, a checklist for evaluating any tracking platform (ours included), and where I think this category goes next.
A Room Full of Logistics, Not Just Mailrooms
The people I met in San Diego run central receiving, campus distribution, last-mile delivery, and warehouse operations. That changes the conversation. Nobody was asking whether students would get a text when a package arrived. They were asking about throughput, exceptions, and accountability.
One leader at a large Midwest research university moves roughly 1,200 packages a day, for faculty and staff only. The same team handles gas cylinders, dry ice, chemicals, campus mail, and courier. The volume is remarkably steady all year, because research does not spike the way student orders do. Steady volume makes planning easier. It also raises the stakes per item, since a delayed lab shipment is not a late dorm delivery.
I also heard how much the org chart decides software. One leader told me her logistics group sits under facilities, which means she has to use a facilities-driven work-order system that was never built for receiving. Another recently started reporting into procurement, and now answers to someone who has never worked a dock. Neither constraint will ever appear in an RFP. Both shape what can be bought.
If you are evaluating tools, ask early who your tool choice has to satisfy besides you.
Lens 1: Operational Efficiency Is a Decision-Count Problem
On most campuses, somewhere between 60 and 75 percent of incoming items arrive with a clear recipient. Any system handles those. The labor, the delay, and the frustration live in the rest.
One operator I spoke with put it plainly: the easy items go out the door, and the hard ones sit around taking up space. I hear a version of that at nearly every campus I visit.
Exceptions have familiar causes:
- A department name that has been abbreviated or butchered (“ACCT” for Accounting)
- A package addressed to the university with no recipient at all
- A street address and nothing else
- A recipient name that is blurry, covered by a sticker, or partly hidden
The fix is not typing faster. It is deciding once. When an operator resolves “ACCT” to Accounting a single time, every future label with that variation is assigned and routed automatically. We call these name variations. The point is to stop spending brain cycles on a question you already answered. A person who decides on the first item and again on the millionth has not been helped by the technology. They have just been given a faster way to repeat themselves.
Capture is also not the finish line. One campus logistics team showed me their internal shipping code: a four-digit building number plus a six-digit room number that almost never prints on the label. The real work is resolving an address pattern to that code so the item can be pre-routed. Reading the label is step one. Resolving it to a destination is where the time is saved.
A final note on method. Older OCR approaches tell the software which region of the label to read. That works for forms you design, like invoices. It breaks on shipping labels, which nobody designed for your software. Reading the whole label and finding the data wherever it sits is a different capability, and it is the one that holds up in a mailroom.
Lens 2: Letter Mail Is Shrinking and Getting More Important
Letter mail volume has fallen by roughly half. The common conclusion is that it matters less. I think the arithmetic says the opposite.
Suppose 10 pieces in every 100,000 are critical: a legal notice, a check, a government document, a regulatory filing. If volume drops to 50,000 and those same 10 still arrive, the critical share of your mail has doubled. You are handling less mail, and each piece is more likely to be one you cannot afford to lose.
This is where image capture stops being an improvement and becomes a new capability. A barcode cannot track a letter, because most letters have no barcode. Capturing an image of the item creates a record of something that previously had no identifier at all. That is not a faster version of the old process. It is a process that did not exist.
Three things follow from that record:
- Proof. When something that was supposed to arrive never reaches its recipient, and you cannot show that you never received it, it becomes your word against theirs. An image and a timestamp end that argument.
- Counting. If you handle hundreds of thousands of letters a year, can you tell leadership what the real number is? Without a count, effort is a mystery and staffing is a guess.
- Classification. Capturing the class of mail, and the Intelligent Mail barcode where one exists, makes the stream reportable instead of anecdotal.
The letters to lockers configuration I described in San Diego shows how simple this can be. A department that receives letters, whether assigned to individuals in the department or the department itself, gets a QR cover sheet. The letters go into a locker, and the scan at the locker triggers the notification to whoever picks up. Every letter is still imaged, so the audit trail stays intact. No barcode on the letter was ever required.
One caution from the other side of the ledger: you do not need to track everything. Bulletins and flyers may not be worth the effort unless a mandate says otherwise. Decide on purpose what belongs in the system.
Lens 3: The Notification Email Is the Most Underused Tool on Campus
Students are easy. They ordered the item, they are excited about it, and even a poorly worded email gets them to the mail center. Faculty and staff are different. Many do not know what is coming or when, and when they do not know, they call.
That is why I tell every campus the same thing: the email you send faculty and staff is the most underused way to control the narrative about your operation. A few practical moves:
- Send it from your domain, with your branding. It should come from you, not from a vendor.
- Change one verb. “You have a package” invites a phone call. “Your package is being prepared for delivery and will arrive by 2 p.m.” sets an expectation and answers the question before it is asked.
- Put an image of the item in the email. A researcher can tell at a glance whether the item is critical lab material or a box of pens. That lets them turn their attention on or off.
- Use every notification to teach. Ten times a year or a thousand, a short reminder of your delivery practices does more over time than any policy memo.
You will never eliminate “where is my package” calls entirely. But the most demanding customers on campus are often the most important ones, and they already know it. Giving them information before they ask is how you earn the right to run your operation without interruption.
People barely talk on the phone anymore, yet most people at work live in their inbox. The mail center that shows up there, clearly and on time, shapes how the whole campus sees it.
What Operators Told Me They Actually Need
The best part of the day was the questions I could not answer cleanly. Three themes came up often enough that I think they deserve their own list.
Two layers of status
Customers see a status like “exception.” Operators need to know which exception: an attempted delivery held in a freezer, held at a dock, held because the information was incomplete, or returned to sender. One leader called these “statuses of statuses,” a customer-facing layer and an operational layer beneath it. It is a fair requirement, and it is the kind a buyer should raise before a decision, not after.
The switching cost is vocabulary, not data
Several operators have built their processes around features that were never designed for the way they use them. They would have to rebuild that operating language on a new platform. Moving records is the easy part. Mapping the vocabulary your team has built over years is the real work, and any vendor who skips it is underestimating your change.
Integration versus control.
Five years ago, almost every buyer asked how a system would integrate with theirs. Today more of them want manual control over who is in the directory and where items route, because someone upstream owns that gate. The right answer is to support both: integrate with a campus system, or load recipients from a CSV.
A Practical Checklist for Evaluating Any Tracking Platform
This applies to us, to the legacy barcode systems we compete with, and to anyone else you are considering. I would rather you choose well than choose quickly.
- Start earlier than feels necessary. If your contract ends in the next 12 to 18 months, begin the conversation this winter. Public-sector procurement alone can take many months, and the decision itself should not be rushed against that clock.
- Clean your recipient directory. Names, emails, buildings, routes. Any platform is only as good as the list it starts with.
- Bring your worst labels. Pull a month of the items that sat on your shelf because no one could tell where they belonged. Test every vendor on the same real items, not their sample set. No honest vendor should promise 100 percent. Ask what happens to the rest.
- Write down your operating vocabulary. Internal statuses, routes, handoffs, the exceptions your team handles by habit. Make each vendor show you how it works.
- Ask references a better question. “How is it working for you?” almost always gets “great.” Ask instead: which solutions did you evaluate, and why did you choose this one?
- Build the leadership case in three lenses. Efficiency (hours and decisions removed), risk (what you can now prove), and experience (calls avoided, trust gained). Liking a tool is one thing. Articulating the case to your team and your leaders is where evaluations stall.
That last point deserves emphasis. Most stalled decisions I see are not about the product. They come down to uncertainty about return, fear of change management, an internal selling process that has not started, or IT and procurement concerns. Naming those early is the cheapest way to keep a good decision moving.
Where the Category Is Heading, and an Honest Note on Why It Is Hard
Every vendor in the room now says AI. Saying it is easy. Making it work on a label that is torn, taped, blurred, or partly hidden under a sticker is not.
We started this work in 2019, and it has taken years of development, much of it unglamorous. People traced labels by hand, region by region, down to the pixel, so the models could learn what a name, a PO number, or a mail stop code looks like wherever it appears. Our development teams are dedicated to this product, which keeps them from being pulled toward unrelated work. None of that is visible in a demo. All of it shows up in how an item behaves on the thousandth label.
I will not tell you every item routes and captures perfectly. It does not, and no one’s does. What I can say is that the system was built to handle the exceptions, with a place for people to work them, rather than pretending they do not exist.
Two directions we are working toward in 2027, offered as direction and not commitment:
- Conveyor-based package capture. A prototype is already underway.
- High-speed letter capture, potentially using the mail machine already in your operation. The details depend on the machine.
Our internal goal for capture speed is something close to telepathy: you present the item and the system already knows. We will not fully get there, but the target keeps us driven.
