Why suppliers don’t use your portal
Updated August 2026
Your brackets come from one machine shop, and that shop serves about forty other customers. Your portal asks the shop to treat you differently from the other thirty-nine: another login, another format, another screen where its sales types the PO dates, quantities, and prices. But every one of the shop’s forty customers has the same ask, and it simply cannot maintain forty portals. It maintains the one channel that reaches all forty at once: email.
That is the entire failure point. Portals don’t fail on software quality; they fail on incentive. The supplier asked to do the extra work (sign up, log in, re-key) captures little of the value. A supplier can comply once, as a favor to a customer it can’t afford to lose. But that adoption doesn’t scale.
Twenty years of supplier portals, supplier networks, and collaboration platforms are the same bet under different names: that a thousand small businesses will change how they work for one customer. The uncomfortable reality is: the bet keeps losing.
Why the demo keeps winning
From your seat, a portal sells consolidation: one system, all four hundred suppliers inside it. From the supplier’s perspective, it is fragmentation: one more system, stacked on every other customer’s. The demo is always given from your side. But what decides the outcome lives on the other side.
Vendors know this — it is why every proposal includes a “supplier enablement program”: months of onboarding, mail campaigns, and escalation procedures to persuade your suppliers to use the thing. A tool that actually fits the supplier’s workflow would not need one.
And enablement is not a one-time cost. The trained contact changes jobs, the login lapses, and the urgent exception goes wherever answers come fastest, which is email.
The adoption arithmetic, from the supplier’s perspective.
Where portals do work
Portals do work the best in one scenario: when you are most of the supplier’s business. Walmart’s Retail Link and the automotive OEMs’ EDI mandates work because those suppliers cannot afford to say no. So the mandate is credible, and the supplier assigns people to comply.
When you think about adopting a portal solution, we suggest running the same test on your own base: what share of your key suppliers’ revenue are you? For a mid-market manufacturer, the honest answer is usually a single-digit percent. At that share, the mandate isn’t going to work: the handful of suppliers who need you most will log in, for a while. The long tail, where your single-source shops usually sit, never will.
Partial adoption is worse than none
Suppose the enablement campaign lands anyway and the head of your supplier base complies. What you have bought is a second source of truth.
The compliant suppliers’ acknowledgments now live in the portal; everyone else’s still arrive by email; and your buyers now reconcile two systems against the ERP instead of one. The work you meant to retire is still there, plus a login.
You have also become tech support for your suppliers’ sales (password resets, retraining, escalations), a never-ending tax on exactly the relationships you most need goodwill from.
Partial adoption splits the truth.
What to do instead
The fix is not a better portal. It is to stop asking suppliers to come to your data and to read the channel they already use. Everything a portal promises is already arriving in your buyers’ inboxes today, in free text: the acknowledgment, the date push, the partial-ship notice, the price change. With more than half of POs changing after issuance, those threads are where the real delivery picture gets negotiated daily.
The gap was that extracting free text, comparing it line by line against every open PO, and writing corrections back to the ERP was manual work no team could keep up with at volume.
That extraction is what large language models got good at, and it is what we build: an agent that reads the PO messages, reconciles what suppliers say against every open line, drafts the chases and the ERP updates, and leaves your buyer one decision: approve. The supplier signs up for nothing, installs nothing, changes nothing.
AI agents reading email inboxes and structuring the data.
Three questions for any vendor
Whoever you are evaluating, three questions separate the fixes from the portals:
1. Does any of the value depend on supplier adoption? If yes, ask what happens to the POs of the suppliers who never log in, because some never will. Then ask a reference customer for adoption twelve months after rollout — counted by supplier, not by share of spend. The spend-weighted number hides the tail, and the tail is where your line-down risk lives.
2. Does it write back to the ERP? Reading email and drawing a dashboard is half the job. Your planners run MRP against the ERP; if corrected dates don’t land there, you have bought a second place to check, not a source of truth.
3. Who approves? Unattended writes to your ERP should worry you; fully manual updates are the status quo you are trying to leave. The answer you want is in between: the system reads, compares, and drafts; a buyer approves every write.
Good answers to all three describe a system that asks nothing of your suppliers, keeps the ERP as the single source of truth, and keeps your buyers in control.
Assumptions & sources — PO change rate: SourceDay 2025 published data, >50% of POs change after issuance. Supplier counts (forty customers per shop, four hundred suppliers per site, single-digit revenue share) are illustrative of mid-market manufacturing. Enablement decay and compliance stratification: direct observation across EMS and industrial sites. Retail Link and automotive EDI cited as public examples of dependence-backed mandates.