Every messy portfolio I have ever seen had a tidy little secret. The mess did not start in delivery. It started at intake, months earlier, in the small moments when work slipped in through side doors, hallway asks, forwarded emails, and that one executive who never quite fills out the form. By the time the chaos shows up on a status report, the real cause is long gone.
Nobody plans it this way. It happens because saying yes in the moment feels helpful, and because there is rarely one obvious place for a request to land. So requests land everywhere. And a portfolio that accepts work from everywhere is not really a portfolio at all.
Ten doors means no portfolio
When work can enter from anywhere, you do not have a portfolio, you have a pile. Requests arrive as favors. They arrive as meetings that end with, "Can your team just..." They arrive three levels down, promised by a manager who had no idea the team was already full.
None of it is visible. None of it is compared against anything. And all of it quietly competes for the same people who are supposed to be delivering the work you actually planned.
The tragedy is not that the surprise requests are bad. Some are excellent. The tragedy is that they never had to prove it, because they never stood next to the alternatives. That is what an open intake really costs you. Not a few extra projects, but the ability to choose. When everything is already in flight, prioritization becomes a fiction. You are not deciding what matters most. You are discovering, late, what other people already decided for you.
Build one front door
The least glamorous and most powerful thing a PMO can do is insist that every request ends up in the same intake process, described the same way, no matter where it starts.
Notice that this is about the destination, not a single screen. In a real organization, requests show up through a customer portal, a service desk ticket, an internal form, a sales conversation, a security escalation. That variety is fine. What cannot vary is where they land: one system of record, with the same minimum information, so you have one place to look instead of ten.
And keep the ask short. Not a fifty-page document that nobody reads and everybody resents. A handful of honest fields is enough:
- What is being requested?
- Who is requesting it, and who benefits?
- What happens if we do nothing?
- Roughly how large is the effort?
- How urgent is it, and why?
Five questions a reasonable person can answer in ten minutes.
The point is not paperwork. The point is that demand becomes visible and comparable. Once every request looks the same on the page, you can lay them side by side and actually see the portfolio, instead of reconstructing it from memory and hallway conversations. A front door does not slow good work down. It stops invisible work from jumping the line.
The example that tests the rule
Here is where every intake process meets its real exam.
It is Thursday afternoon. A sales leader messages you: the biggest prospect of the quarter needs a custom integration, and the deal walks without it. This is urgent, it is real, and it is exactly the kind of request that usually bypasses every process ever written.
The instinct is to make an exception. Resist it, gently. The front door does not mean slow. It means visible. That request can move through the same short case in the same ten minutes, and if it is genuinely the most valuable thing in the building this week, it will win on its merits and jump to the top of the list, out in the open, where everyone can see what it pushed aside.
That is the difference that matters. Urgent work is not the enemy of intake. Invisible work is. When the emergency goes through the door, you can say yes to it and still know what it cost. When it goes around the door, you lose that knowledge, and you pay for it three months later when the planned work is mysteriously late and nobody can say why.
Triage it, then let it flow
A front door is only half the system. The other half is what happens once a request walks through it.
Getting through the door is not the same as getting a yes. Submission earns consideration, not commitment. Intake makes the request visible. Triage decides whether it is valid, a duplicate, or genuinely mandatory. Prioritization decides whether and when it should actually proceed. Keeping those steps separate is what stops "I filled out the form" from turning into "so you promised me my project."
So someone has to triage. Not a committee that meets once a quarter, but a regular, lightweight review where a few people with authority look at the incoming cases together. Advance the ones that are valid, mandatory, or genuinely worth considering, and send them into the same prioritization as everything else, so a shiny idea from a loud sponsor has to earn its rank against the work already in flight.
Be careful with the word park. A parked request needs a specific reason and a date to be looked at again. Without both, park just means a graveyard nobody visits, and people quietly learn that the honest answer would have been no. So if it is a no, say no, clearly and kindly. A clean decline respects people more than a backlog that never moves.
The front door also has to answer back. It earns trust only when every request has an owner, a visible status, and a rough date by which the submitter will hear something. People route around intake when the official path feels like a black hole. Show them who holds their request and when they will get an answer, and most of them will happily use the door.
Do this consistently, and many of the worst intake surprises become visible before they disrupt delivery. Far fewer projects appear in month three that nobody remembers approving, and there is much less archaeology to find out who said yes. The work in your plan becomes mostly the work you chose, which sounds obvious until you notice how rarely it is actually true.
This is also where a tool starts to earn its place. A demand intake pipeline keeps the whole trail, from submission to decision, in one home. Perspective gives each request a consistent business case and routes it through review. Qualified demand then flows directly into portfolio prioritization and capacity planning, alongside everything already competing for the same people.
But the software is not the point, and it never was. You can run the whole thing with a shared form and a standing thirty-minute meeting. The tool saves effort. The discipline is what saves the portfolio.
If you want to start on Monday
You do not need a governance overhaul to fix this. You need one door and the will to use it:
- Pick one system of record every request must land in, whatever channel it arrived through.
- Give it five fields: what is being requested, who is asking and who benefits, the cost of doing nothing, the rough size, and the rough urgency.
- Hold a short weekly triage to validate, decline, or advance each request, and give every request an owner and a status.
- Feed the advanced work into the same prioritization as your existing commitments.
The first week will be awkward, because people will still reach for the side doors out of habit. Hold the line politely. Every request you route through the front door is one you can finally see, compare, and stand behind.
One door, every time. Because work you cannot see is work you cannot choose.