There are four status colors in Planview’s portfolio roll-up. Three of them are the ones you expect. The fourth is gray, and it means the project manager did not report (their documentation, plainly).
That gray ball is the most honest object in enterprise portfolio management, and it is worth sitting with. It exists because the color is an opinion somebody has to remember to have. Green does not arrive from anywhere. It arrives from a person, on a Thursday, in a form — and when the person is busy, the system needs a way to say so.
None of which makes Planview a bad product. It is the most complete portfolio database in this comparison and the record says so: a Gartner Leader in strategic portfolio management, and since its acquisition of Sciforma in February 2025 the owner of two of the ten incumbents we measure against. Investment and Capacity Planning gives a portfolio director four working views — Rank, Analyze, Shift and Balance — and lets a chosen scenario be published as the plan of record. Capacity targets are time-phased at any level of the organization. Financials carry an explicit capital-versus-operating split. Hub moves work items across sixty-plus tools, bi-directionally, including GitHub and GitLab. Against that inventory, this product is small.
Here is the argument anyway. Every one of those capabilities takes its input from testimony, and testimony is the thing a delivery organization staffed with agents can no longer supply.
What Planview shipped when the shape changed
Copilot reached general availability in the November 2024 release with status summarization, risk flagging and questions over metrics. Then came Sentiment Analysis: software that judges whether work is on track by reading the emotional tone of comment and status text.
It is a well-built instrument for interviewing humans, delivered at the moment the interview stopped returning signal. You cannot sentiment-analyze an agent. It has no tone. And the humans still in the loop are not a reliable fallback either — in METR’s controlled trial, experienced developers working in their own repositories were 19% slower with AI tools while believing they had been 20% faster. That is not a story about careless engineers. It is a measurement of self-report and reality coming apart under exactly the conditions every enterprise is now creating deliberately.
So the question a portfolio tool has to answer moves. Not what did the project manager say, but what happened, and who accepted it.
What we built instead
A requirement carries its citations. The citations resolve against one pinned commit — a forty-character hash, not “the repository” — and a test result binds the requirement at the rung the evidence actually supports. A skipped test binds nothing. A failing test still binds, at its rung, with the failure as its outcome, because this is a record rather than a dashboard. Two rungs on the ladder are declared and issue nothing at all, and that is the part most likely to be mistaken for an omission: approximating them would be the exact overclaim the ladder exists to prevent.
Then a named human accepts. Not as a policy — the column recording who accepted is a foreign key into the identity table, and an agent has no row there and cannot acquire one. The accepting identity is stamped by the database from the session rather than sent by the client, so a request that supplies a different name is refused instead of trusted.
The objection worth taking seriously
The easy version is that this is a dashboard with better vocabulary. The test that settles it is about inputs. A dashboard’s inputs are things people typed: the color they chose, the percent they estimated, the comment a model scored for tone. Ours are things that happened. You can rebuild any dashboard from our rows. You cannot rebuild our rows from a dashboard, because it never held them.
The harder version is commercial: Planview can add commit-level binding next year and keep everything else it has. Maybe. But most of what we built is refusals, and refusals are expensive to retrofit onto an installed base. A vendor whose roll-up ships a gray ball for “nobody reported” would have to take the gray ball away to ship this, and the gray ball is there because thousands of customers need it on a Thursday afternoon.
Where Planview wins
Resource capacity: theirs is time-phased across the organization, ours is two numbers on a page. Scenario planning: theirs publishes a plan of record, ours does not exist and will not until there are honest inputs to compute over. Connectors: sixty-plus, against our test results and one repository. Timesheets, which we concede on purpose. And two decades of portfolio process encoded in a product a very large number of organizations run today, against one estate of 119 requirements here. If the evaluation weights breadth, this is not a close contest.
Choose Planview Portfolios if…
Choose Planview if the portfolio office’s core job is brokering people and money across a large human delivery estate — if the question that keeps coming back is which of forty proposals can be staffed next quarter, and the answer has to survive a capacity model, a funding envelope and a published plan of record. That is what Investment and Capacity Planning is for, and nothing here replaces it.
Choose it also if breadth is the risk you are managing: a great many organizations run it, the process is encoded, and buying the market leader is a defensible decision in a way that buying a product with one onboarded estate is not.
Come here instead if the delivery organization under the portfolio is getting smaller and more machine-produced, and the question you cannot answer any more is which requirement a build was made against and who accepted the result. The receipt verifies without an account, which is the fastest way to find out whether that is your problem.