Product Manager

Nummo is hiring a Product Manager. The role is fake, the headcount is fake, and the roadmap you would own does not exist. This is a description you can paste into a form if you need a third sample that is not design, backend, or analytics.

You will decide what gets built next, then write that decision down so engineering can build it and support can explain it. The writing is the job. A sentence in a ticket that names the user, the action, and the result is worth more than a deck that says the team will delight users. Delight is not a requirement. A person finishes the task, or they do not.

A request will arrive from a customer, from sales, or from someone who saw a competitor do a thing. You will ask who has the problem, how often, and what they do today when the product fails them. If the answer is “a few loud accounts” or “it would be cool,” you will say so, and you will not turn it into a quarter’s work to avoid the awkward pause. You will also say yes sometimes. Saying no to everything is not judgment. It is a delay wearing a firm tone.

You will write the outcome before the solution. “A recruiter can see which candidates are waiting on them, and why” is an outcome. “Add a sidebar” is a guess. You will stay with the outcome long enough that design and engineering can propose different shapes. When you already have a picture in your head, you will label it as a picture, not as the requirement. People build what you write. If you write the picture, you will get the picture, including the parts that made no sense once the data showed up.

Scope is the skill people mean when they say “product sense.” A first version does one path for one kind of user and leaves a clear edge where the rest would go. It does not include settings for cases nobody has hit. It does not include a configuration screen so the team can postpone the decision. You will cut items in the open, with the reason next to the cut, so the item can return later without a forensic reading of old comments.

You will sit with engineering while the work is happening, not only at the planning meeting and the launch meeting. Edge cases appear in the middle: a missing field, a permission that blocks the happy path, a state where the job is half done. You will choose a behavior for those cases. “What do you think?” is fine once. The second time, you owe an answer. Shipping with an undefined middle is how support gets a ticket the product should have answered.

Launch is a checklist, not a mood. The empty state has words. The error says what failed. The person who asked for the feature knows it is out, and the person who will get yelled at if it breaks knows what “working” means. You will decide what you are measuring before the release, in one sentence, with the definition an analyst can query. “Engagement” does not count. “Share of new tasks finished without a retry in the first week, excluding internal accounts” counts.

You will talk to customers. You will take notes that separate what they did from what they said they wanted. You will not promise a date in the call. You will bring the contradiction back: two firms described the same pain and prescribed opposite fixes. Your job is to pick the fix that serves the shared pain, and to tell the other firm what you are not building yet.