River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Product Stakeholder Alignment Document

Every alignment template gives you a box called open questions. This one arrives full, from what these specific people have already said.

Free download  ·  No account needed

A roadmap review goes wrong in a predictable way. Somebody raises a concern they have raised before, nobody has an answer, and the next forty minutes are spent discovering that in public. The templates on offer all have a box for this and the box is called open questions. It gets filled in during the meeting, which is the meeting you were trying to avoid. Nobody counts how often it has already happened, because notes are filed per meeting and the third raise looks exactly like the first.

So this pack fills it in first, from the record rather than from a workshop. Every objection anybody has raised comes out of the meeting notes, the one-to-ones and the email threads, attributed to a person and dated. Then it is deduplicated on person and substance, and the repeat count comes with it. In the worked example, 54 statements across 14 sources became 29 distinct concerns, and 14 of those had been said more than once by the same person.

The count only means something next to a second column: whether a specific answer was ever given. Concerns with one on the record came back 2 times out of 11. Concerns without came back 12 times out of 18, so about 3.7 times as often. The four left with no answer at all are the agenda, and they hold 23 of the quarter's 61 engineer-weeks. It sits next to the roadmap being reviewed and the planning cycle around it, and it runs before either.

Fifty-four statements, twenty-nine concerns, four real questions

Every objection counted by person, every unanswered one priced against the roadmap, and the review rewritten around what is left.

Concern to Response Map

Illustrative, for a fictional field-service scheduling product called Halverstock. 54 objection statements pulled from 14 sources over five months, deduplicated on person and substance into 29 concerns. The nine most-raised are shown.

Raised byThe concern, in their wordsTimesAnswer on recordWhat the response rests onEng-wks behind it
Raghunathan, CFOA routing rewrite shipped and I still cannot see what a completed job costs5None$1.42 to $1.09, table named0
Ostrowski, VP SalesEvery deal I lost last quarter I lost on mobile, and mobile is never on this roadmap4NoneNothing yet, D-19.0
Deasy, GM EnterpriseEnterprise is 41% of revenue on the same release train as a two-van plumber4NoneNothing yet, D-25.0
Wrightson, ComplianceWe are promising residency work into regulated accounts that is not scheduled3NoneNothing yet, D-36.0
Nkemelu, SupportThe dispatch rewrite doubled ticket volume for six weeks and nobody counted it3NoneNothing yet, D-43.0
Petren, CTOSix things in a quarter where two of my four teams are on the migration3None61 against 64 available0
Bonneau, Cust. SuccessSomething ships every quarter my team hears about from a customer first3NoneRelease notes, 48h, named owner0
Petren, CTOEvery roadmap arrives after the sprint it has to start in2Yes3 weeks before the quarter0
Bonneau, Cust. SuccessOnboarding for the top forty collides with every release window2Yes4 days of slip, accepted0
20 more5 raised twice, 15 raised once259 yesMixed0

Fourteen of the 29 concerns have been raised more than once by the same person, and those fourteen carry 39 of the 54 statements. Two of the eleven concerns with an answer on record came back anyway, 18 percent. Twelve of the eighteen with no answer came back, 67 percent.

Roadmap Exposure

The same 29 concerns joined against the 18 items going into the review. A concern is attached to an item when answering it one way rather than another changes whether the item ships, when, or what it costs.

ItemEng-wksConcerns attachedStatusCommittable today
Mobile job capture, offline first6.0C-02OpenNo
Mobile dispatch acknowledgement3.0C-02OpenNo
Enterprise release channel5.0C-15, C-26OpenNo
Regional data residency, first region4.0C-19, C-03OpenNo
Residency evidence pack for audits2.0C-19OpenNo
Dispatch board performance pass3.0C-11OpenNo
Routing cost reporting4.0C-08, C-28AnsweredYes
Scheduling conflict warnings5.0C-09, C-04AnsweredYes
5 more answered items15.0C-21, C-01, C-04, C-08, C-06, C-23AnsweredYes
5 items nobody has ever mentioned14.0NoneNoneYes
Total61.029 across 13 items6 open, 7 answered, 5 untouched12 of 18

23 of the 61 engineer-weeks, 38 percent of the quarter, sit behind four questions nobody has written an answer to. Three of the four can be settled in ten minutes by somebody already in the room.

Review Agenda

What the 90-minute booking becomes once 25 of the 29 concerns arrive already answered. Last review ran the full 90 with 58 of those minutes on ground already covered.

MinSlotWho it is forWhat is being askedEng-wks riding on it
5Confirmation, not discussionEveryoneConfirm the 25 answered in the briefs, or flag one to reopen38.0
15D-1 Mobile in the quarter, or a dateOstrowski, PetrenMobile goes in and two named items come out, or a date for two accounts9.0
15D-2 A separate enterprise trainDeasy, Petren, RaghunathanA yes or a no. Two quarters of ambiguity is what is being escalated5.0
10D-3 Residency this quarterWrightson, RaghunathanSchedule the 6 engineer-weeks, or accept in writing the accounts it costs6.0
10D-4 Support headcount against loadNkemelu, RaghunathanWhether headcount moves with estimated support load, as a standing rule3.0
5The one number worth reading outEveryoneCost per completed job, $1.42 to $1.09, because it settles the oldest concern in the room0.0
606 slots against a 90-minute booking7 stakeholders4 decisions and 1 confirmation61.0

The register takes 6 hours of one person to build and about 40 minutes to update before each later review. A year of four reviews costs 8 hours of preparation against the 27 person-hours currently spent re-covering old ground.

What is in the pack

01

A repeat count, which no alignment template has

Statements deduplicated on person and substance, so the third raise stops looking like the first one.

02

A strict test for whether it was ever answered

A number, a date or a written trade-off, given to the person who raised it. A reassurance grades as no answer.

03

One response per concern, with an owner

Written to the person who raised it, in their vocabulary, naming a source they can check without asking you.

04

A grade for what each response rests on

A checkable number, a dated commitment, an accepted trade-off, or nothing yet. The last one is a legitimate answer.

05

The roadmap effort behind every open question

One join, and it puts the four remaining questions in cost order instead of in whatever order they get raised.

06

Decision requests rather than dressed-up updates

Question, options priced, recommendation, who decides, and what it costs if nobody does.

How it works

  1. 1

    Send the record

    Meeting notes, one-to-ones, email threads, channel exports. Six months is plenty and raw beats summarised.

  2. 2

    Count the repeats

    Every objection attributed and dated, then deduplicated on person and substance so the raise count is real.

  3. 3

    Check for an answer

    A number, a date or a written trade-off, or it grades as unanswered no matter how senior the person who said it.

  4. 4

    Write the responses

    One per concern, graded by what it rests on, and the ones resting on nothing become the agenda.

Frequently asked questions

What do I actually need to send?

The record, not a stakeholder list. Meeting notes, one-to-one notes, email threads, channel exports, the last review deck and whatever came back after it. Raw beats summarised, because a summary has already dropped the name and the date this runs on. Then the roadmap with rough effort per item.

Does this make my colleagues look difficult?

It does the opposite, and the arithmetic is why. People repeat themselves when nothing was ever given to them that would let them stop. The register is a list of unpaid debts rather than a list of awkward people, and in the worked example the loudest one turned out to be the cheapest to settle.

What counts as a real answer?

A number they can check themselves, a dated commitment with a named owner, or a trade-off somebody accepted in writing. Nothing else. "We are aware of that" and "let us take it offline" grade as unanswered, and grading them generously is what produces the second raise.

What happens to concerns with no answer?

They become decision requests with options priced, a recommendation and a named decider. That is the most useful output here. Writing an answer-shaped paragraph instead buys one quiet meeting and costs the next three, which is how most of these concerns reached their fourth raise.

Is a register like this really necessary?

Federal rulemaking has run one since 1946. An agency must consider and respond to every significant comment before a rule is final, and a rule that skips one can be challenged on that basis. The mechanism is old and well tested; product reviews are just unusual in not having it.

What if somebody still disagrees at the end?

Then record that they disagree and will support it. Amazon's shareholder letter is direct about the difference: a decision cycle that needs conviction rather than commitment is slower for no gain. This pack is built to produce commitment, and scoring the trade-offs is a separate question.

How is this different from a stakeholder map?

An influence and interest grid describes people. This counts sentences. Nobody has ever changed a roadmap because of a quadrant, and the register works from the outcome and the evidence each person has actually cited rather than from where they sit.

Find out how many times they have already said it

Send the meeting notes, the one-to-ones and the threads. What comes back first is the count: how many distinct concerns, how many have been raised more than once, and how many of those ever got a specific answer.

Prepare my roadmap review