Home
Appearance
Text size
100%
Grid

Renato Valdés-Olmos

Essays

Product management
was already doing design.

1 September 2026·Operating·~9 min

The canonical public framework for product management scores user experience design. The canonical one for design scores product thinking. Neither document mentions that the other exists, and the region they are both quietly reaching into has since become the most valuable part of the map.

Ravi Mehta's product competency toolkit, built at Tripadvisor and the closest thing product management has to a canonical public framework, sorts twelve competencies into four groups: product execution, customer insight, product strategy, and influencing people. Sitting inside customer insight, alongside fluency with data and voice of the customer, is user experience design.

The ladder I inherited at Grammarly did the same thing more explicitly. It scored UI and UX expertise as its own competency, footnoted as required for anybody working on user-facing products, rising from partnering with a designer on mocks at the second level to running research projects unsupervised at the fifth.

Meanwhile the design ladder, in the same company, in the same year, listed product thinking as one of its twelve craft skills. I owned both documents, and did not read them side by side until long after either mattered.

Two documents, each crediting the other's discipline, neither acknowledging the other existed.

What the field has to work from

The published ones

Product has one framework the field actually argues about and very little else in public, which is why most product ladders in use are a competency list somebody wrote in a week.

Framework
Year
Shape
What it levels on
2020
Twelve competencies in four groups
Execution, customer insight, strategy, influence
2017–
Company matrices, published piecemeal
Scope, autonomy, impact, in varying proportions
The big-company levels
Numbered bands shared with engineering
Scope, mostly — surface owned, headcount adjacent
The ladder in most companies
Span, autonomy, strategy
Territory, distance from your manager, how far ahead you see

The variation worth noticing is not a framework but a naming argument: whether the job is a product manager who decides or a product owner who administers a backlog. Ladders that never settle it end up scoring the second job and paying for the first.

What product ladders actually measure

Strip the language and most product ladders measure three things: how much surface you own, how little supervision you need, and how far ahead you can see. Span, autonomy, strategy.

Span is the load-bearing one and it is the one that has aged worst. It counts features, then a product, then multiple products, then a line of business. It is a measure of territory, and territory made sense as a proxy when shipping anything took a quarter and owning more of it genuinely meant deciding more.

The autonomy row measures distance from your manager. Contributes ideas to the backlog, then owns the backlog with supervision, then without it. That ladder had six meaningful levels when a product team was fifteen people. On a team of seven senior people it has about three, and the rest is padding.

Neither row asks the question that actually separates product managers, which is what they will refuse.

The exchange rate problem

At Pitch I ran product, design and research as one organization: six product managers, about twenty-five product designers, a research team and an eight-person brand and marketing design team, all reporting up through me. Which meant the exchange-rate problem was not between two departments. It was inside one, and I was the person both ladders escalated to.

There is a mechanical failure here that nobody notices until somebody transfers. At Grammarly, two years earlier, product management, product design, brand design and research all went to seven levels ending at VP, and engineering had seven named titles that were not levels at all, ending at the same word.

So a PM5 and a PD5 were not the same thing, nobody could tell you what the exchange rate was, and compensation bands hung off both. When a designer wanted to move into product, or a product manager into design, the conversation had no arithmetic in it. It was resolved by whoever argued better, which is a bad way to make a pay decision.

And where a ladder stops a level short of its neighbor, as brand did, the drift runs one way. Moving toward the taller ladder reads as a promotion and moving toward the shorter one reads as a step down, regardless of what either person actually decides. That is a fact about the documents, not about the work.

No exchange rate

Four ladders in the same company, in the same year, and no statement anywhere of how a level in one related to a level in another. Compensation bands hung off all of them. Nobody could tell you what a PM5 was worth in design terms, and the question came up every time somebody wanted to move.

Product
Design
Research
Engineering
1
PM1
PD1
UXR1
Junior
2
PM2
PD2
UXR2
Mid
3
PM3
PD3
UXR3
Senior
4
PM4
PD4
UXR4
Tech Lead
5
PM5
PD5
UXR5
Eng Manager
6
PM6
PD6
UXR6
Director
7
PM7 · VP
PD7 · VP
UXR7 · VP
VP

Same numbers, no exchange rate: PM5 and PD5 were both written as fifth levels, and nothing said whether they were the same job at the same price. Engineering’s titles could not be mapped onto either. That is a fact about the documents, not about the work.

What the merge did to the role

The interface between product and engineering used to be an estimate. A product manager described what they wanted, an engineer priced it, and the negotiation between those two numbers was most of the job. A large part of the product ladder is really a measure of how well somebody runs that negotiation.

A product manager who can read the schema and run the query does not need an engineer to price the decision. An engineer who can see the funnel does not need a product manager to justify one. The estimate conversation has largely collapsed into the work, and with it a good chunk of what the middle levels of the ladder were measuring.

The same thing happened on the design side, faster. Writing the spec, framing the problem, deciding what not to build, prioritizing against a goal: all of that now sits between the two lanes rather than inside one. It is the most valuable region on the map and it belongs to nobody, which is why no ladder scores it.

Product and Design

Design keepsInteraction design. Service design. What the product refuses to ask of a person.
Now sharedProblem framing. Prioritization. Data fluency. Writing the spec. Deciding what not to build.
Product keepsExperiment design. The judgment to stop something. Which future is worth the quarter.

This one was already true on paper. What changed is that a designer can now produce and test the artifact that used to require a product manager to schedule it.

Product and Engineering

Product keepsMarket judgment. Sequencing against a business outcome. Stakeholder cost.
Now sharedSystem reasoning. Technical trade-offs. Querying your own data. Reading a schema. Costing a change.
Engineering keepsArchitecture. Keeping a decision reversible. What the system can still do in three years.

The estimate conversation, which was most of the interface between these two lanes, has largely collapsed into the work.

Read down the middle of both bands and almost nothing in there is a craft skill. The shared region is made of judgment, which is exactly what span and autonomy were never measuring.

What I would score instead

Decision rights, not span. Not how much surface, but which decisions this person makes alone. A product manager who owns three products and cannot kill any of them owns nothing. The test is whether they can stop something without asking.

Opinions held. The scarcest thing in product work is somebody who will say what should exist and be judged for saying it. Not a prioritization framework, not a well-run process. A position, held when holding it cost time or scope or goodwill, that the company can name.

Judgment density. Whether the room decides faster because they are in it. The product manager who runs an excellent process and generates options indefinitely is adding capacity, not decisions, and those are different levels.

Range across the boundary. Which calls are they trusted to make in engineering's lane or design's? Querying your own data is now the floor, not a credit. Being the person engineering routes a sequencing question to is a level.

And what survives them. A product manager at the higher levels is not remembered for what they shipped. They are remembered for what the product refuses to do, and whether that refusal held after they left.

The six axes, in product’s language

Three of the six levels. Note that none of them mention how many products a person owns.

Axis
P3 · Which problem
P4 · What we refuse
P5 · What outlives us
Decision rights
Decides what is worth solving next inside their area, and brings the framing before anybody asks for it.
Can kill a funded, working project without asking. Can hold a launch date against commercial pressure.
Sets what the product is for, in terms the company argues inside rather than about.
Judgment density
The roadmap conversation is shorter and better because they are in it.
Called into other teams’ prioritization because they end the argument rather than adding options.
Teams make the call the same way when they are not there, because the reasoning is legible.
Opinions held
Has a position on what should exist next and can say why it beats the alternatives.
Has held a position that cost scope, revenue or goodwill, in public, with their name on it.
The product’s refusals are theirs, and they survived the person leaving.
Range
Writes their own queries. Reads the designs well enough to argue with them.
Trusted with calls that belong to design or engineering — a sequencing decision, an interface trade-off — because they have been right about them.
Brought into other lanes’ framing as a matter of course.
What survives them
A definition of done, or a metric the team actually uses.
A written constraint on what the product will not do, which later decisions are checked against.
A strategy still shaping decisions three years and two reorganizations later.
Ownership
Involved: picks up the adjacent problem without being assigned it.
Invested: carries a company problem that is nobody’s and gets no credit.
Invested, and the reference point for what the word means on the team.

Read the P4 column and every cell is a refusal with a cost. That is the level most product ladders cannot see, because span and autonomy both rise while a person refuses nothing.

The uncomfortable part

If the shared region between product and design is problem framing, prioritization, data fluency, spec writing and deciding what not to build, then a fair question is what is left that is distinctly product.

The AI Daily Brief gave the surviving role a name this July: when prototypes are cheap and ideas are abundant, the decision rights move to the editor, the person who says which of the forty deserves to be built. That is a fair description of the top of this table.

I think the answer is narrower than most product managers would like and more durable than they fear. It is the judgment about which of the possible futures is worth this quarter, and the willingness to stop something that is working in order to do it. Everything else on the list is now shared, and pretending otherwise is how you end up with a product organization that is mostly coordinating people who no longer need coordinating.

The whole system

You can read the argument for free. Running it is the hard part.

An essay can tell you the six rows stopped working. It cannot sit in the room in November when two managers disagree about the same person and neither can say why. That is what the workbook is for: one shared core across Engineering, Product, Research and Design (EP(R)D), six levels defined by what a person decides, and the mechanics to grade against them without the cycle turning into a negotiation.

Get the workbook · $39 →

Decides
Design
Research
Product
Engineering
1How
D1
R1
P1
E1
2What
D2
R2
P2
E2
3Which problem
D3
R3
P3
E3
4What we refuse
D4
R4
P4
E4
5What outlives us
D5
R5
P5
E5
6What we will not do
D6
R6
P6
E6

Six levels defined once, by what a person decides alone. The shaded band is the overlap, where most of the work now belongs to no single lane. The marks to the left of each cell are doors: a move sideways at the same level, which is a transfer and not a demotion.

Written to be opened during a cycle rather than read once: on the page, as a PDF set to print, and with the templates as a spreadsheet.

Define it

01Why the old ladders stopped workingSample chapter
02The spine: six levels, six axes
03Four lanes, and what each one keeps
04Overlap, doors and the management fork

Grade against it

05Behaviors: the full matrix
06Craft: what to score, at what depth
07Evidence: the decision trail
08Grading, with worked examples
09Nine ways grading goes wrong

Run the cycle

10Running a cycle
11The calibration room
12Promotion cases
13Templates
Three people, graded in fullA designer who should not be promoted yet. An engineer who looks excellent and is not a level 5. A researcher moving into product. Marked axis by axis, with the argument written out the way you would have to make it in the room.
Five templates you can use in NovemberEvidence log, self-assessment, calibration scorecard, promotion case, transfer record. Short enough that people actually fill them in.
The rule that stops grade inflationDecision rights carries a veto. Below on that axis and the level is not held, whatever the other five say. Most frameworks leave this implicit, which is how everybody ends up a four.

v1. The cycle, the calibration room and the evidence standard are the ones I have run at companies of 25, 250 and 2500 people; the six axes are new. One payment, every revision by email.

Compensation$39For CFOs and Finance teams setting pay for Engineering, Product, Research and Design, with EPD leaders supplying the role and market evidence behind every decision.
Development$39For Engineering, Product, Research and Design leaders and People teams to run together: turn review evidence into funded growth work, protected time and explicit manager commitments.
The new standard$117$99A shared standard for levels, pay and growth. All three workbooks, their printable documents, 24 editable templates and reference sheets, plus three worked examples.

One of four, each taking a function through the same framework; the system they share is set out in full in the workbook. The others are on design, engineering and research. Written from having inherited, run and rebuilt leveling and review cycles at three companies of very different sizes, and against the public frameworks that preceded them: the levels framework in Org Design for Design Orgs by Peter Merholz and Kristin Skinner, Rent the Runway’s engineering ladder, and Ravi Mehta’s product competency toolkit. Figures from the AI in Design Report 2026 and Figma’s State of the Designer 2026.

© 2026 Renato Valdés-Olmos