Goal Setting When OKRs Never Stick

Who they might be — the founder-CEO of a mid-size agri-commodity business: trading, warehousing and a marketplace platform, built from nothing and recently at break-even. Persona — the systems builder. Generates more ideas than his company can absorb, knows it, and has engineered a structure to absorb them. Contributed — the only working cascade.

Facilitator of the 2026-08-22 Goal Setting Workshop and the only person in the room running a fully cascaded goal system. The most talkative participant (32 of 123 segments) and the one presenting material: five pillars, company OKRs, department OKRs, individual sheets, appraisal forms. His company appears here as Alvimora.

What he argues

• Framework is secondary to the cascade — "Das Thema ist Zielsetzung über Hierarchie. Das ist nicht OKR."
• OKR as company language — he uses OKR as a cultural vocabulary, not a control system.
• Goals should be set as hypotheses — early company objectives are hypotheses to be falsified.
• Vertical and Horizontal Alignment plus Bottom-Up Feedback — and admits he personally neglected the bottom-up half.

What he practises

Alvimora Five Pillars · Alvimora OKR Cascade · Alvimora OKR Sheet · Individual OKR Sheet and Signature · OKRs at 70 Percent of the Performance Appraisal · OKR Master · Twenty Percent Capacity for New Topics · Quarter by Quarter not Year Detailed

The self-diagnosis worth keeping

He describes the founder failure mode more precisely than anyone else in the room: dropping half-formed ideas into the Monday call to signal excitement, and the employee hearing an instruction. "Machen wir das jetzt noch oder warten wir?" His whole system is an answer to a problem he created — and he says he needed four attempts at OKRs before it held.

Where the evidence disagrees with him

His strongest practice — OKRs at 70 Percent of the Performance Appraisal — is the one thing every primary OKR source tells you not to do. See Divorce compensation from OKRs and Skip individual OKRs. His Alvimora OKR Sheet also carries a column borrowed from development-aid project management: Means of Verification.

Links: The Single-Metric Founder pushes back on bureaucracy; The Weekly-Rhythm Founder pushes back on nomenclature; The Dynamism Sceptic pushes back on rigidity.

Who they might be — founder of a B2B tech company with Sales, Customer Success and an engineering org; also sits in a peer intervision group outside this one. Persona — the fast thinker who diagnoses himself. Tried OKRs, failed, and blames his own discipline rather than the method. Contributed — the bureaucracy objection.

Second-most active voice (28 segments). Sat in the Future Days OKR workshop last year, tried to integrate OKRs, and concluded "OKR ist es nicht."

What he argues

• It is a discipline problem not a framework problem — the boundary between KPI, Objective and Key Result is knowable; founders skip the sorting because they think too fast.
• The Bureaucracy Objection — a week of work to align company, department and individual OKRs, repeated every quarter, and a sloppy sheet from ten people contaminates the whole thing.
• Tech work cannot be tied to revenue — shipped tickets and features have no causal chain to revenue; Customer Success was the only place he could build one.
• Founders are the operating system — he agrees "100%" and treats it as the reason a system must eventually replace the founder.
• Devil's advocate on the commission model: Incentives do not survive contact with employees.

What he practises

North Star plus Upstream Scorecard — everything aligned to a single revenue metric, with upstream KPIs sought "scorecard-mäßig" in other functions. His own verdict: "mittelmäßig begeistert von unserer eigenen Leistung."

Where the evidence agrees and disagrees

He is right that the KPI/KR boundary has a workable rule (OKR), right that appraisal forms give employees clarity, and right that tech-to-revenue chains are mostly inferred rather than measured (DORA and SPACE). He is arguably wrong that this is only discipline: Cascading does not scale and Skip individual OKRs suggest the cost he is complaining about is structural. On his revenue-only north star, see North Star Metric.

Links: The Host - Cascade Practitioner, The Commission Experimenter, The Dynamism Sceptic.

Who they might be — one of three co-founders of a transaction marketplace at roughly 80 staff, with the engineering leads reporting to him. Persona — the allergic-to-jargon operator. Cannot work with OKR vocabulary and has built a functioning cadence without any of it. Contributed — focus weeks, 50/50 targets.

Contributes 13 segments but supplies most of the concrete machinery in the session. His company appears here as The M&A Marketplace.

What he argues

• The OKR nomenclature is unusable in practice — the weight-loss example: is losing weight the objective or is health? Is weight a KPI or a key result? "Mein Gehirn ist zu einfach gestrickt."
• A target picture only matters above a certain team size — the Munich metaphor: a roadworks detour doesn't change the destination.
• Early stage needs no goal system only hustle — while the founding team can talk daily, strategy is irrelevant.
• Honest about his own governance failure: the Don't List was broken by a co-founder in Q1.

What he practises

Focus Weeks · Monday Assignment Meeting · Friday Written Report · 50-50 Goal Rule · 2028 Target Picture and Five Company Bets · Don't List

Of everyone present he has the tightest weekly loop and the loosest goal vocabulary — the exact inverse of The Host - Cascade Practitioner.

Where the evidence lands

His rhythm has strong analogues he does not seem to know about: 4DX's cadence of accountability, EOS and Rocks' Level 10 on-track/off-track rule, Scaling Up and Quarterly Themes. Two gaps the research surfaces: No credible framework bans discussion outright (he has the reporting half of the loop, not the solving half), and Weekly theme rotation has no precedent. His 50/50 rule: Where the 50 percent rule actually comes from.

Links: The Restless Co-Founder, The Weekly Operating Rhythm, The Dynamism Sceptic (his closest ally in the room).

Who they might be — founder of a transactional platform where customers push volume through the system; Sales, Operations, Support and Tech, currently a small team that is growing. Persona — the structured questioner. Asks about process where everyone else tells stories, and is willing to test his own model out loud. Contributed — incentives instead of goals.

27 segments. Also attended last year's OKR workshop; also dropped it.

What he argues

• His revenue north star works "bis zum Wissensgrad" but has an expiry date as the org grows. Revenue as a single north star has an expiry date
• A commission model may replace a goal system — the most interesting claim of the session: if the incentive already rewards acquisition, retention and development, does pushing goals down add anything but HR and process load?
• Capacity, not motivation, is the binding constraint: an onboarding manager "macht so viel, wie ihm kapazitär möglich ist" — so more goals means a process problem, not a performance problem.
• The three open questions he tabled at the start: How do you handle a priority change mid-cycle, How do you review without a documentation burden, How do you calibrate ambition.

On calibration he frames the dilemma unusually honestly: do you set goals deliberately below the real target to enable overperformance, or so high they're never met?

What he practises

Volume-Based Commission with a 50-50 Split — commission scaling with customer traffic, split 50/50 between the salesperson who won the account and the ops person who runs it, explicitly to remove opposing interests. Live for a few months; he is confident in it.

Where the evidence lands

The strongest counter is structural, not motivational: Multitask Principal-Agent Problem and Reward interdependence must match task interdependence. His volume-as-satisfaction proxy is partly defensible — see Usage as a proxy for satisfaction — but not "near-perfect". His calibration question has a real answer: The Stretch Goal Paradox and Managers set targets to be hit 8 or 9 times out of 10. His fixed 50/50 split is the specific design choice practitioners warn about: Split commissions and team quotas.

He also cites The Absent Contrarian as holding the view that commission harms the quality of who applies.

Links: The Quiet Interjector (whose one line punctures the hypothesis), The Single-Metric Founder.

Who they might be — founder of a business he considers structurally unforecastable, with a small product and market team he manages by talking to them. Persona — the anti-framework realist. Never tried OKRs at all, unlike everyone else in the room, and says so without embarrassment. Contributed — the designer question, unanswered.

17 segments, and the clearest anti-framework position in the session.

What he argues

• OKRs are too rigid for a dynamic business — "für mich alles zu starr, weil Sachen sich viel zu dynamisch ändern." Operational reality dictates priorities more than any annual plan.
• Planning credibility: "Unser Business ist nicht berechenbar genug, ich kann nicht wie an der Börse sagen, das ist das Quartal, bin ich da."
• How do you set goals for roles that are not directly measurable — his question about designers is the one nobody in the room answers.
• The founder-temperament problem: how do you keep a person with "eine gewisse Sprunghaftigkeit" inside a system, including yourself — "wenn ich irgendwann keinen Bock mehr habe."
• The deepest epistemic objection in the session: how can I know in January that what I set for November is right, when the causal chain runs "über acht Ecken"?

Where the evidence lands

His rigidity objection has serious backing: Shape Up and the Betting Table is an entire operating model built on it, Cascading does not scale agrees, and Balanced Scorecard's standing criticism is exactly annual staleness. His "over eight corners" objection is the honest state of the art per DORA and SPACE and Input vs Output Metrics. Where he is vulnerable: he manages by high-frequency conversation with The Product Counterpart, which works at his current size and is precisely what A target picture only matters above a certain team size predicts will break.

Links: The Weekly-Rhythm Founder, The Commission Experimenter, The Absent Contrarian.

Who they might be — about to incorporate, a team of four, no product in market yet. Persona — the earnest newcomer. The only person who asks for something rather than presenting something, and apologises for it. Contributed — goals change weekly.

Three segments, the earliest-stage person in the room. His request, at 69:02, is the most quotable moment in the session:

"Gefühlt unsere Ziele ändern sich dreimal pro Woche… gefühlt sich nach jedem Gespräch mit einem Advisor, sich ändern."

It feels like our goals change three times a week — after every conversation with an advisor.

He asks for five to ten minutes on how the others handled this phase, "ganz egoistisch gesehen für mich", and notes that everything discussed so far is "so viel weiter in der Zukunft" than his problem. The Dynamism Sceptic offers to make it concrete ("Was war die letzte Woche, was ist es heute, was war das vor zwei Wochen?"). The host defers it to after the break — so the session's only unanswered request is also its only pre-PMF one.

Earlier he made the one practical process request of the workshop: share an anonymised example OKR sheet with numbers and names changed, so people can see how one is actually written.

What the evidence offers him

How do you set goals when they change three times a week collects it: one growth number instead of a goal system (Startup growth rate as the only metric), Early stage needs no goal system only hustle, and the honest finding that advisor-driven goal thrash is not a well-studied phenomenon — the closest usable guidance is on filtering advice, not on setting goals.

Links: 2026-08-22 Goal Setting Workshop, Future Days.

The top layer of Alvimora's system: five fixed company pillars that do not change from year to year.

1. Brand & Market Excellence — growth in volume, buyers, reach.
2. IT System — a system people understand and experience as a tool that helps their work.
3. Customer Experience
4. Operational Scale — the discipline of channelling the founder's constant new ideas into something that scales. His onboarding question: "Ist das skalierbar, was du heute machst? Wenn du nicht da bist, kann das jemand anderes übernehmen?"
5. Organisation & Culture

Annual company objectives are then defined within the pillars, so the vocabulary stays stable while the targets move.

Why the stability matters

The Commission Experimenter immediately spots that these pillars are near-universal — "das sind Teilbereiche, die fast alle hier betreffen" — and asks the right question: how do you weight them against each other when Operational Excellence is genuinely the only thing that matters for the next twelve months? The host's answer is that everyone new recognises the pillars from previous employers and the interest is in what they mean here — which does not answer the prioritisation question. It stays open.

The published analogue

This is a legitimate variant of strategic themes in the Balanced Scorecard: cross-cutting themes, each with a measurable strategic result, decomposed into objectives. The recommended count there is 3–4, so five fixed is at the upper end, and the Balanced Scorecard Institute's standing warning applies — themes tend to "disappear" in the final scorecard, and skipping careful theme design loses objectives critical to the result.

Links: Alvimora OKR Cascade, Alvimora OKR Sheet.

The mechanism, stated plainly by The Host - Cascade Practitioner and confirmed under questioning at 46:04–46:59:

A company-level Key Result becomes a department-level Objective.

Not every company KR does — "es war nicht immer, dass du alle Key Results von der Company-Ebene auf eine Abteilung kriegst" — but any KR the company commits to must be owned by someone as their objective.

The worked example (2024)

• Pillar: Brand & Market Excellence
• Company Objective: the AgriHub brand "has become synonymous with high capability, reliability, trust"
• Company Key Results: (1) a wholesale/volume target routed through the platform, (2) hard revenue growth of the shops themselves — "wenn sie Saleswachstum haben, dann haben wir auch für die Key Results gewonnen", (3) portfolio expansion to 250 outlets
• Second Objective: Real-Time Data Excellence → became the IT department's objective, with KRs for tool coverage and user adoption
• Third: Customer Experience & Satisfaction → introduce NPS, run surveys
• Plus Sales & Logistics, and Organisation & Culture. Four to five company objectives per year, each with a named owner at senior-management level — "ein Company OKR ist nicht nur der Gründer verantwortlich."

The most instructive friction in the whole session

IT rejected user adoption as their key result: "meine Aufgabe ist es, das Ding zu bauen, aber wie es genutzt wird, ist nicht mein Thema." The host held the line, and the result was a forced dependency between IT and Operations.

He is right, and this is the single best counter to Tech work cannot be tied to revenue in the transcript: adoption is an outcome IT can influence and shipping is not. Compare Input vs Output Metrics — build-completion is an activity metric; adoption is closer to a controllable input. See also HEART Framework, whose Adoption dimension exists for precisely this argument.

Caution

Strict multi-level cascade is the specific pattern the canon has turned against: Cascading does not scale, and Doerr's own objection to goals "handed down the org chart like tablets from Mount Sinai." Alvimora mitigates it by naming owners and by Vertical and Horizontal Alignment plus Bottom-Up Feedback — but the approval load lands on the founder.

Links: Alvimora Five Pillars, Individual OKR Sheet and Signature.

The bottom of the Alvimora OKR Cascade, and the part the host considers the whole point: "wenn wir nur auf der Ebene Unternehmensziele durchgeben, dann landen die nicht bei den einzelnen Mitarbeitern."

Every employee has an OKR sheet answering two questions they kept asking him: what do I contribute, and what am I measured on? Each quarter's sheet is signed by the employee and their manager.

The worked example

An Operations Manager, 2025:

• Objective: "Enhanced AgriHub Engagement and Digital Adoption" — visibly the 2024 company objective, reworded
• Key Results: 500 new customers (up from the prior year's 250-outlet company target), NPS 60
• Initiatives: the employee's own — "welche Initiativen muss ich in meinem Department durchführen, um diese Key Results zu erreichen"

Initiatives are stable enough to double as job description material: "das könnte genauso die Job Description sein… ich brauche jemanden, der Market Analysis und Target Identification kann." The host explicitly calls this the bottom-up half.

The Pre-Launch Founder asked for an anonymised copy of exactly this sheet — the one concrete artefact request of the session.

The two problems

1. The canon says don't do this at all. Rick Klau, author of the famous Google OKR video: "Skip individual OKRs altogether. Especially for younger, smaller companies. They're redundant" — because they turn into task lists and get conflated with performance reviews. See Skip individual OKRs.
2. The signature is a commitment device pointing the wrong way. It binds the employee to a target set above them, in a system where OKRs at 70 Percent of the Performance Appraisal makes the target pay-relevant. That is the textbook setup for sandbagging — and the field data says people negotiate targets they can hit 8 or 9 times out of 10. See Managers set targets to be hit 8 or 9 times out of 10.

Links: Alvimora OKR Sheet, The Bureaucracy Objection.

The most consequential practice in the session, and the most contested. The Host - Cascade Practitioner, 54:06:

"So haben wir dann das ganze Thema Key Results für eine Person in die Performance Appraisal gebracht. Und das schlägt so zu 70 Prozent der Bewertung von der Person. Und dann so Kompetenzen, so Team-Mitarbeit… Und wenn ich weiß, hey, daran werde ich gemessen — ich kann in den Performance Improvement Plan kommen, oder ich kriege mehr Gehalt — dann ist jeder wirklich darauf fokussiert, das Ding richtig zu machen."

Mechanics: 70% key results, 30% competencies and peer feedback. Reviews run twice a year (quarterly was attempted and abandoned as too expensive). Consequences are explicit in both directions — PIP or raise.

The Single-Metric Founder's reaction was admiration: "Das ist ja ultra sexy eigentlich. Weil ich als Mitarbeiter kriege sofort eine absolut klare Performance Indication."

Every primary source says do not do this

• Doerr, Measure What Matters: "Divorce compensation (both raises and bonuses) from OKRs. These should be two distinct conversations, with their own cadences and calendars."
• Laszlo Bock (ex-Google SVP People Ops): "Objectives and Key Results should be totally divorced from compensation." Google tried it once; "people started gaming the system to get their bonuses" — "detrimental to both the product and the broader culture."
• Google re:Work: "OKRs are not synonymous with employee evaluations… not a comprehensive means to evaluate an individual."
• Perdoo: "Not achieving OKRs is just as important as achieving them."

Full detail and the counter-position (Gartner: OKRs should augment rather than replace performance goals) in Divorce compensation from OKRs.

The predicted failure mode, and how to check for it

Coupling pay to targets produces sandbagging, not focus: employees negotiate targets they know they can hit. The tell is not complaints — it is a suspiciously high hit rate. If Alvimora's individual key results are being achieved most of the time, the system is calibrating downward and nobody will say so.

Diagnostic worth running: compare last year's individual attainment distribution against the company-level one. If individuals beat the company, the cascade is being gamed at the bottom.

Two mitigations that keep most of the benefit: exclude aspirational goals from the appraisal weighting entirely (Committed vs Aspirational OKRs), or keep the clarity artefact P2 admired while removing the pay formula — the appraisal form was what he liked, not the 70%.

Links: Individual OKR Sheet and Signature, Goals Gone Wild, The Bureaucracy Objection.

The Weekly-Rhythm Founder, 05:56. The origin story is deliberately unflattering: The Restless Co-Founder brainstormed a grandly-named internal "OS" with an LLM on a long train journey — "total überkanditelt, ganz viele Begriffe drin, die überhaupt keinen Sinn ergeben." But one idea inside it worked.

"Wir haben uns einmal zusammengesetzt und wirklich zwei Tage lang die fucking Karten gelegt und gesagt: was ist eigentlich unser Zielbild 2028? … Wenn du das einmal zwei Tage lang machst, wo du dir wirklich die Köpfe einschlägst und dann halt nicht sagst, ja das wird sich eh noch ändern, sondern wirklich sagst: ich fixiere jetzt einen Satz, einen Absatz — da möchte ich hin."

From it: five company bets, then this year's slice, then the quarter. His claim is not that this eliminated churn but that it made quarterly goals stable enough: "wir exekuten schon sehr auf diese Vision."

Why the two days matter more than the document

The mechanism is refusing the escape hatch. The thing that makes target pictures worthless is the sentence "it'll change anyway" — spoken in advance, it licenses vagueness. Two days of argument to fix one paragraph is a commitment device, not a planning exercise.

Analogues

• BHAG (Collins & Porras) and Amazon's PR/FAQ — write the destination before the route. Amazon Working Backwards
• Spotify DIBB and Bets — the word "bets" is Spotify's, and the DIBB chain would give him what his version lacks: the belief each bet rests on, so a dropped bet updates a stated belief rather than looking like a whim.
• EOS and Rocks's V/TO holds exactly this stack — 10-year target, 3-year picture, 1-year plan, quarterly rocks — on two pages.

The unresolved tension

The bets are supposed to stabilise the quarter. The Don't List was supposed to protect the year. His co-founder broke the don't-list within one quarter. A target picture cannot bind a founder who does not consent to being bound; see Founder discipline needs an external enforcer.

Nothing about the OS, the target picture or the bets appears in public sources — all participant-stated. See The M&A Marketplace.

Links: Don't List, A target picture only matters above a certain team size.

The Weekly-Rhythm Founder, 05:56 — an explicit annual list of things the company will not do:

"Da stand Anfang des Jahres drauf: wir wollen keinen Eigenkapitalfonds machen, weil das eigentlich eine gute Idee für uns ist. Wir machen das nicht, weil das würde Ressourcen ablenken."

Note the reasoning: it was banned because it was a good idea. That is the whole point of a don't-list — it defends against attractive distractions, not bad ones.

What happened next

"Dann haben wir in Q1 zurückgeschaut, hat [der Mitgründer] gesagt, er will diesen EK-Fonds machen. Dann haben wir das auf der Don't-Liste durchgestrichen, haben den Eigenkapitalfonds raufgepackt. Da hat [er] ein Quartal daran gearbeitet, und nach dem zweiten Quartal: nee, ist doch nicht so gut. Und dann haben wir es wieder runtergeschmissen."

Struck through within one quarter, pursued for two, abandoned. His own verdict on the practice: "das hat geholfen… Also es passiert trotzdem immer noch."

Lineage

• Jim Collins' stop-doing list: "A 'once-in-a-lifetime opportunity' is irrelevant if it is the wrong opportunity." His companion discipline, the 20 Mile March, is consistent bounded progress — including deliberately not overreaching when conditions are good.
• Porter: "The essence of strategy is choosing what not to do." A position without trade-offs is imitable.
• 37signals adopted the stop-doing list independently.

The missing half

A don't-list is a commitment device with no enforcer. Commitment devices work when there is an external party and a cost for defection — the Ulysses pact, stickK's deposit contracts. This one had neither: the person who wrote it crossed it out.

The mechanism EOS and Rocks supplies is worth noting because it is cheap and non-confrontational: in a Level 10 meeting the founder's rocks appear on the same scorecard as everyone else's and receive the same two-word verdict. Enforcement by symmetry of exposure, not willpower. See Founder discipline needs an external enforcer.

Links: The Restless Co-Founder, 2028 Target Picture and Five Company Bets.

The Weekly-Rhythm Founder, 10:10:

"Ich habe im Endeffekt einen Snapshot am Ende abfotografiert, und am Montag, die Woche startet mit dem Meeting, dass ich draufgucke und sage: okay, was kann ich diese Woche erreichen, was auf eins dieser Sachen einzahlt. Und dann wird es halt wirklich im Seniorkreis in dieser Woche diskutiert. Weil sonst, wenn ich weiterplane, funktioniert es nicht."

Weekly ownership is assigned in the senior circle against a photographed snapshot of the quarterly goals. Planning horizon: one week, deliberately. Anything further out "funktioniert nicht."

Paired with the Friday Written Report and, one step earlier, his own Friday session with the tech leads so he arrives at Monday prepared.

The published analogue

This is 4DX's cadence of accountability with the third element removed. The WIG session is 20–30 minutes, same day and time, three parts: account for last week's commitments, review the scoreboard, plan and clear the path — each person commits to the one or two things they will do this week. See 4DX.

P3 has account (on Friday, in writing) and plan (on Monday, in the meeting). What he does not have is clear the path — the part where blockers get removed rather than reported. That is the same gap as No credible framework bans discussion outright, seen from the other end of the week.

Second gap: the photograph. 4DX Discipline 3 and Rockefeller habit #10 both require a durable, visible scoreboard the team maintains, not the leader.

Links: Focus Weeks, 50-50 Goal Rule, The Weekly Operating Rhythm.

The Weekly-Rhythm Founder, 11:37 — the session's most concrete answer to How do you review without a documentation burden:

"Wir haben einen Channel. Und da ist gar nicht Diskussion, sondern da muss jeder reporten, ob er es erreicht hat, auch ganz kurz und schriftlich. Hast du es erreicht? Ja, dann easy. Nein, warum nicht? Für alle nachvollziehbar… Aber halt nicht nochmal ein Diskussionsmeeting, sondern ganz klar: Verantwortlichkeit wurde am Montag vergeben, am Freitag erwarte ich End of Day einen Report."

Three properties: written, short, no meeting. If missed, it goes on Monday's agenda for the following week.

Strong precedent for the writing

Grove, High Output Management: "Reports are more a medium of self-discipline than a way to communicate information. Writing the report is important; reading it often is not." The value is in forcing the author to be precise and confront their own trouble spots. Amazon's narrative memos and silent-reading meetings, Basecamp's automatic check-ins, GitLab's async updates and Automattic's P2s all rest on the same claim. See Written reports are a medium of self-discipline.

The gap, stated precisely

Nobody credible advocates status without discussion. They advocate status outside the discussion meeting.

• EOS Level 10: on-track/off-track only during scorecard and rock review — then 60 of the 90 minutes go to IDS on the issues that surfaced.
• 4DX: whirlwind topics are barred from the WIG session, but "clear the path" is a fixed agenda item.
• Grove's written report feeds a one-on-one.
• Bezos's formulation: crisp documents, messy meetings.
• Toyota's andon makes a problem visible in order to trigger immediate help; obeya is the room where cross-functional people go to solve what the visuals surfaced. Visible signal without a response mechanism is the documented failure mode of visual management.

P3 has removed the discussion without designating a venue for it. The predictable result is the very thing The Commission Experimenter fears — a compliance ritual. The fix is small: the "warum nicht" answers become Monday's agenda, which he already half does, so name it and time-box it.

Links: Monday Assignment Meeting, No credible framework bans discussion outright, EOS and Rocks.

The most direct contradiction of any practice presented in the 2026-08-22 Goal Setting Workshop. Verdict on OKRs at 70 Percent of the Performance Appraisal: every primary source says don't.

• Doerr, Measure What Matters: "Divorce compensation (both raises and bonuses) from OKRs. These should be two distinct conversations, with their own cadences and calendars."
• Laszlo Bock, ex-Google SVP People Operations: "Objectives and Key Results should be totally divorced from compensation." Google tied them once — "people started gaming the system to get their bonuses", which was "detrimental to both the product and the broader culture." He adds that people mentally discount a future bonus by ~80% anyway.
• Google re:Work: "OKRs are not synonymous with employee evaluations… not a comprehensive means to evaluate an individual."
• Ryan Panchadsaram (What Matters): the mechanism is sandbagging — "you don't want to penalize people for aiming high" — plus "a culture of sharp elbows". He cites Volkswagen's emissions scandal and Wells Fargo.
• Perdoo: "Not achieving OKRs is just as important as achieving them."
• Christina Wodtke: keep OKRs out of individual performance reviews; use continuous conversations.

Two honest qualifications

1. Doerr is not absolutist. He allows that OKRs can "provide important feedback for the review cycle" and act as "an antidote to ratings bias." The canonical rule is decouple from the pay formula, not ignore at review time. That distinction is the practical escape route for The Host - Cascade Practitioner: keep the appraisal artefact, drop the 70% weighting.
2. The counter-position exists. Gartner: OKRs "should augment, not replace, employee performance goal setting" — an explicit institutional rejection of full separation.
3. Grove's own stance is weakly sourced. The often-quoted line about MBO not being "a legal document on which to base a performance review" could not be verified against High Output Management. Treat as unconfirmed.

Sources

Bock: divorce compensation from OKRs · Should you connect OKRs and compensation? (spoiler: no) · Google re:Work · Castro quoting Measure What Matters · Gartner

Links: Goals Gone Wild, Managers set targets to be hit 8 or 9 times out of 10, Committed vs Aspirational OKRs.

The strongest evidence base in this entire vault: 400+ studies over ~35 years, summarised in American Psychologist (2002) with a fifty-year retrospective in 2019.

Core findings

• Specific difficult goals beat "do your best": effect sizes d = .42–.80.
• Goal difficulty and performance are related linearly, not curvilinearly — "the highest or most difficult goals produced the highest levels of effort and performance." This is a direct empirical rebuttal to any inverted-U or "50% optimum" story. See 50-50 Goal Rule.
• Performance plateaus "when the limits of ability were reached or when commitment to a highly difficult goal lapsed."
• Moderators: goal commitment, feedback, ability and task knowledge, situational constraints, self-efficacy, task complexity. Complexity attenuates the effect substantially — d = .48 for the most complex tasks versus .67 for the least.
• Four mechanisms: goals direct attention, mobilise effort, sustain persistence, and prompt strategy and task-relevant knowledge.

The decisive practical caveat (2019): impossible goals can raise performance in the lab because there is no penalty for failure; in the field "impossible-to-attain goals can lead to demoralization and punishment." Where the authors endorse stretch goals it is under two design conditions: no penalty for missing, and a parallel minimum attainable target that must be hit.

That is Committed vs Aspirational OKRs, arrived at independently by psychologists rather than by Google.

Why this note anchors the vault

Because it is the evidentiary basis for Framework is secondary to the cascade. The effects come from goal properties — specificity, difficulty, commitment, feedback, task knowledge — none of which any framework owns. Every framework in this vault is a delivery vehicle for the same five variables.

And the fourth mechanism is the answer to A commission model may replace a goal system: money supplies effort; only a goal supplies direction and strategy.

Sources

Locke & Latham 2002 (PDF) · Locke & Latham 2019, 50 years (PDF) · Five principles overview

Links: The Stretch Goal Paradox, Goals Gone Wild.

Ordóñez, Schweitzer, Galinsky & Bazerman (2009), Academy of Management Perspectives 23(1). The standard citation for goal-setting's side effects, and the counterweight to Goal-Setting Theory Locke and Latham.

Five documented harms: a narrow focus that neglects non-goal areas; a rise in unethical behaviour; distorted risk preferences; corrosion of organisational culture; reduced intrinsic motivation.

The prescription that gives the paper its force: goal setting should be treated "as a prescription-strength medication that requires careful dosing, consideration of harmful side effects, and close supervision."

The three cases worth knowing

• Ford Pinto — Iacocca's 1969 goal: under 2,000 pounds, under $2,000, 1971 model year. Safety checks skipped, fuel tank ten inches behind the rear axle, ~53 deaths.
• Sears — a $147/hour sales quota for auto-repair advisers produced unnecessary repairs and overcharging. The chairman conceded the "goal setting process for service advisers created an environment where mistakes did occur."
• Wells Fargo — an eight-products-per-household cross-sell target, ~2 million unauthorised accounts, 5,300 employees fired, $185m in fines (2016).

Relevance

All three attached a number to pay or to job security. That is the specific configuration in OKRs at 70 Percent of the Performance Appraisal, and the reason the canonical rule exists (Divorce compensation from OKRs).

Also relevant to Volume-Based Commission with a 50-50 Split: "narrow focus that neglects non-goal areas" is the same mechanism as the multitask distortion, arrived at from behavioural ethics rather than economics.

Sources

Goals Gone Wild (SSRN) · Knowledge@Wharton discussion · Wells Fargo fine

Links: Goodharts Law and Campbells Law, On the Folly of Rewarding A While Hoping for B.

Sitkin, See, Miller, Lawless & Carton (2011), Academy of Management Review 36(3), with an HBR follow-up in 2017.

The thesis: "stretch goals are, paradoxically, most seductive for organizations that can least afford the risks associated with them."

The two enabling conditions: recent strong performance and available slack resources. Firms possessing both rarely use stretch goals; firms in decline reach for them and "nearly always fail." The canonical failure case is Mayer-era Yahoo, declaring double-digit growth targets while revenue fell from $7.2B to $4.9B.

Chet Miller's framing: stretch goals rarely end well "unless an organization has had very strong recent success and has avoided advice to be very, very lean."

Applied to this room

This is the actual answer to How do you calibrate ambition — the question is not "how ambitious" but "am I allowed to be ambitious right now?"

• Diagnostic: recent wins, and real slack. The Host - Cascade Practitioner has just reached break-even (a recent win) and has explicitly budgeted slack (Twenty Percent Capacity for New Topics) — he qualifies.
• A founder setting stretch targets because the quarter was bad — the exact scenario The Dynamism Sceptic describes at 13:20, where a weak Q1 makes the founder nervous and reaching — is the failure pattern named.

Sources

Sitkin et al. 2011 (SSRN) · HBR 2017: The Stretch Goal Paradox · Miller interview

Links: Stretch goals raise variance not median performance, 50-50 Goal Rule.

The measurement pathologies, in their actual wording rather than the paraphrases.

Goodhart (1975), original: "Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes." Note this was a claim about monetary aggregates, not incentives. The famous compression — "when a measure becomes a target, it ceases to be a good measure" — is Marilyn Strathern's (1997), not Goodhart's.

Campbell's law (Donald Campbell, 1979, versions from 1969 — so predating Goodhart): "The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures."

Ridgway (1956), Administrative Science Quarterly 1(2): 240–247, "Dysfunctional Consequences of Performance Measurements" — the founding text, and the most useful of the three because it is specific about which kind of measurement fails how:

• Single measures motivate effort that is "wasted or detrimental to the organization" — see North Star plus Upstream Scorecard and Volume-Based Commission with a 50-50 Split.
• Multiple measures leave staff unsure which to prioritise — see the unanswered prioritisation question in Alvimora Five Pillars.
• Composite measures suppress trade-offs but still distort — see OKRs at 70 Percent of the Performance Appraisal's 70/30 weighting.

Ridgway anticipated all three failure modes present in this room, seventy years ago.

Sources

Goodhart's law · Campbell's law · Ridgway 1956 summary

Links: Goals Gone Wild, Net Promoter Score, DORA and SPACE.

The most useful single finding for The Weekly-Rhythm Founder's system.

Nobody credible advocates status without discussion. They advocate status outside the discussion meeting.

• EOS Level 10: during scorecard and rock review, members give only "on track" or "off track" — then 60 of the 90 minutes go to IDS (Identify, Discuss, Solve) on the issues that surfaced. See EOS and Rocks.
• 4DX: whirlwind topics are barred from the WIG session agenda, but "clear the path" is a fixed part of it. See 4DX.
• Grove: the written report feeds a one-on-one.
• Bezos: crisp documents, messy meetings.
• Bain's IDD taxonomy (Inform / Discuss / Decide): state the purpose up front, push informational content into pre-reads, and don't mix content types in one meeting. One client deliberately separated short-term operations from long-term planning — separate meetings, not no meeting.
• Rhythm Systems distinguishes status meetings (backward-looking reporting) from adjustment meetings (forward-looking course correction) and argues only the latter justifies the room.
• Toyota: andon makes an exception visible in order to trigger immediate help; obeya is the room where cross-functional people solve what the visuals surfaced. Visible signal without a response mechanism is the documented failure mode of visual management.

The diagnosis

Friday Written Report has removed the discussion without designating a venue for it. The predictable output is a compliance ritual — which is exactly the documentation burden The Commission Experimenter is right to fear.

The fix is small and he half does it already: the "warum nicht" answers become Monday's agenda. Name it, time-box it, and the loop closes.

Sources

EOS Level 10 meeting · Bain: decision-focused meetings · Rhythm Systems: adjustment vs status meetings · Lean.org on obeya

Links: How do you review without a documentation burden, Written reports are a medium of self-discipline.

Lineage. Drucker's MBO (1954) → Andy Grove's iMBOs at Intel in the 1970s, documented in High Output Management — his two questions were "Where do I want to go?" (objective) and "How will I pace myself to see if I am getting there?" (key results) → John Doerr, who learned them at Intel in 1975 and brought them to Google in 1999 → Measure What Matters (2018).

Contested: the Balanced Scorecard Institute argues Grove invented nothing but a faster review cadence — "if I cook pizza in the microwave and call it Pizzaz, have I invented a new food?" And the ubiquitous "Intel grew from $1.9B to $26B because of OKRs" is Doerr's own marketing claim with no causal evidence.

Definitions, and the boundary rule this room needed

• Objective — "what is to be achieved, no more and no less"; significant, concrete, action-oriented.
• Key Result — "specific, time-bound, and aggressive yet realistic"; "you either meet a Key Result's requirements or you don't." Usually 3–5 per objective.
• Doerr's own discriminator: "OKRs are more than just KPIs. They're measures for change, whereas KPIs are measures of health."
• Perdoo's test: KPIs are the gauges on your car's dashboard; OKRs are the roadmap. Initiatives are the projects that move key results.

The operational rule that survives every source: a key result names a delta (baseline → target, by a date). A KPI names a level monitored indefinitely. An initiative names work. A KPI becomes a key result only when you attach a target change and a cycle.

Health metrics — the "business as usual" numbers you monitor but do not try to change this quarter — are Felipe Castro's and Christina Wodtke's concept, not Google's, despite constant attribution. Worth knowing, because it is exactly the category that confuses people in practice.

Sources

What is an OKR (whatmatters) · OKR origin story · Wikipedia · Perdoo guide · Castro: OKR vs KPI · Radical Focus summary · BSI: Grove didn't invent OKRs

Links: The OKR nomenclature is unusable in practice, Committed vs Aspirational OKRs, MBO.

The distinction that dissolves most of this session's arguments about rigidity and ambition, and which nobody in the room mentioned.

From Google's OKR Playbook, verbatim:

• Committed OKRs — "OKRs that we agree will be achieved, and we will be willing to adjust schedules and resources to ensure that they are delivered." Expected score: 1.0. Less than 1.0 "requires explanation for the miss" and shows errors in planning or execution.
• Aspirational OKRs — "express how we'd like the world to look, even though we have no clear idea how to get there and/or the resources necessary to deliver." Expected average: 0.7, with high variance. Falling short "is okay, even expected."

Why it matters for this room

• It answers OKRs are too rigid for a dynamic business: re-planning resources mid-cycle is a feature of the committed class, not a violation of the method.
• It answers How do you calibrate ambition: you don't pick one number, you run two classes. And it matches, independently, Locke & Latham's own prescription for stretch goals — no penalty for missing, plus a parallel minimum attainable target that must be hit.
• It is the safe version of OKRs at 70 Percent of the Performance Appraisal: weight the committed class, exclude the aspirational class entirely.

And the misquote to stop repeating: "70% is good" applies only to aspirational OKRs. Any team applying it to delivery commitments is misreading the source. See Where the 70 percent norm actually comes from.

Sources

Google's OKR Playbook (PDF) · committed vs aspirational · Google re:Work

Links: OKR, 50-50 Goal Rule.

Drucker, The Practice of Management (1954). Manager and subordinate jointly agree specific measurable objectives for a period; performance is appraised against them; objectives at each level serve the level above. Cascade: the original — corporate → division → department → individual, negotiated one-to-one and tied to appraisal and pay. Mid-cycle change: weak; objectives are annual and locked to the review cycle.

The Host - Cascade Practitioner names it in the session as one of the models available ("diese klassische, sehr strukturierten Prozesse"). Worth noticing: what he actually runs — cascaded objectives with 70% appraisal weighting — is closer to MBO than to OKRs. The OKR canon's distinctive move was to decouple the two.

Deming's rejection

Point 11b of the 14 Points: "Eliminate management by objective. Eliminate management by numbers, numerical goals. Substitute leadership." His argument: numerical targets set without a method are met by distorting the system; most variation is systemic rather than individual; and ranking people against targets destroys cooperation.

That is the strongest single argument against everything the host presented, and it is 40 years old.

Sources

Deming's 14 Points · Deadly diseases of Western management · Deming vs Drucker on metrics · Does MBO stifle excellence?

Links: OKR, OKRs at 70 Percent of the Performance Appraisal.

Kaplan & Norton, HBR 1992, from a year-long study of twelve companies. Measure strategy across four perspectives — Financial, Customer, Internal Process, Learning & Growth — and, from Strategy Maps (2004), draw the causal chain upward: capabilities enable processes, processes produce customer outcomes, customer outcomes produce financial results.

Relevance here: the BSC term of art for Alvimora Five Pillars is strategic themes. The Balanced Scorecard Institute recommends 3–4, each cutting across all four perspectives, each with a measurable strategic result, each decomposed into objectives. Five fixed themes annually is a legitimate variant at the upper end of the recommended count.

Cascading: "third-generation" BSC cascades tier-1 maps to business-unit and team scorecards, each level owning what it can influence.

Known weaknesses, and every one of them was audible in this session: it degenerates into a KPI dashboard with no causal logic; it becomes an annual ritual with no monthly review; consultant-built means nobody owns it; "balanced mediocrity" — equal weight across perspectives dodges prioritisation, which is exactly the unanswered question The Commission Experimenter put to the five pillars; and the measurement trap, where metrics become gaming targets.

Sources

BSI: strategic themes (PDF) · BSC basics · Honest criticism of the BSC · 2025 encyclopedia review

Links: Alvimora Five Pillars, MBO.

Gino Wickman, Traction (2007). A V/TO (Vision/Traction Organizer) holds Core Values, Core Focus, 10-Year Target, 3-Year Picture, 1-Year Plan and quarterly Rocks on two pages; a Scorecard of 5–15 weekly numbers; an Accountability Chart (seats before people). Rocks are 90-day priorities, 3–7 per leadership-team member, one owner each, no shared accountability.

The Level 10 Meeting — the piece this session needed

90 minutes, fixed: Segue 5 · Scorecard 5 · Rock Review 5 · Headlines 5 · To-Dos 5 · IDS 60 · Conclude 5.

During scorecard and rock review, members give only "on track" or "off track" — explanations are resisted. Off-track items drop to the Issues List and are handled in IDS (Identify, Discuss, Solve — permanently), 3–5 issues per meeting.

So EOS bans discussion in the status portion and then gives it 60 dedicated minutes. The Weekly-Rhythm Founder has independently invented the ban (Friday Written Report) without the 60 minutes. See No credible framework bans discussion outright.

Second borrowable mechanism: the founder's rocks sit on the same scorecard as everyone else's and get the same two-word verdict. Enforcement by symmetry of exposure — relevant to the broken Don't List.

Weaknesses: rigid and dogmatic vocabulary, a paid Implementer ecosystem, tuned to 10–250-person owner-operated SMBs rather than product companies, and Rocks that are often output-shaped rather than outcome-shaped.

Sources

The Level 10 Meeting · L10 structure · EOS glossary · What are Rocks · Critical implementation review

Links: The Weekly Operating Rhythm, 4DX, Scaling Up and Quarterly Themes.

McChesney, Covey & Huling (2012), FranklinCovey. Four disciplines:

1. Focus on one or two Wildly Important Goals — stated "from X to Y by when" — accepting that the day job ("the whirlwind") will otherwise consume everything. Teams are asked "what is the fewest battles necessary to win the war?" rather than handed a decomposed number.
2. Act on lead measures — predictive and influenceable — rather than lag measures.
3. Keep a compelling scoreboard that the team, not management, maintains.
4. Cadence of accountability — a weekly WIG session, 20–30 minutes, same day and time: account for last week's commitments, review the scoreboard, plan and clear the path. Whirlwind topics are barred from the agenda.

Why it is the closest analogue to what this room already does

"The whirlwind" is precisely The Weekly-Rhythm Founder's founding problem — quarterly initiatives eaten by daily business — and Discipline 4 is his Monday Assignment Meeting plus Friday Written Report, with two additions he lacks: a durable team-maintained scoreboard (he has a photograph) and "clear the path" (he has no venue for removing blockers).

It is also the answer to How do you review without a documentation burden that doesn't require OKR vocabulary: a hard time cap and a fixed three-part template.

Weaknesses: extreme narrowness fits badly when several goals are genuinely important; prescriptive ritual; lead measures are hard to identify and, once tracked, gameable; little to say about cross-functional or portfolio-level goals.

Sources

Perdoo's 4DX guide · Mooncamp glossary · 4DX vs OKRs

Links: The Weekly Operating Rhythm, EOS and Rocks.

Post-war Japan; formalised at Bridgestone Tire in the 1960s while studying Deming Prize winners, with Yoji Akao generally credited as its developer (book 1987, English 1991). Adopted by HP, P&G and Xerox in the late 70s and 80s.

Mechanic. 2–3 breakthrough objectives (3–5 years) → annual objectives → improvement priorities → targets to improve, arranged on the X-matrix with correlation marks showing which contributes to what, plus owners and resources. The whole cycle runs PDCA with monthly and annual review.

Catchball — the answer to the question the host didn't answer

Catchball (nemawashi) is an explicit two-way negotiation protocol. A level proposes a target; the level below responds with what it can actually commit to given capacity and what it needs; the ball goes back and forth until both agree — often through nested matrices, one per level, where a lower level's objectives are the upper level's priorities. The stated purpose is bottom-up ownership and reality-testing of top-down targets.

That is exactly what The Commission Experimenter asked for at 21:02 and never got: not "do you gather feedback" but what is the protocol. See Vertical and Horizontal Alignment plus Bottom-Up Feedback.

Weaknesses: slow — catchball across many levels takes weeks or months; the X-matrix is famously unreadable to the uninitiated; it presumes stable operations and a mature continuous-improvement culture. Note also that the practitioner literature is almost entirely promotional and lists no disadvantages, which is itself a warning sign.

Sources

Strategic Management Insight · Wikipedia · Catchball · The 7 steps of Hoshin Kanri · Tennant & Roberts, Long Range Planning

Links: MBO, Framework is secondary to the cascade.

Input metrics are controllable, customer-facing levers — selection, price, delivery speed, in-stock rate. Output metrics are results — revenue, profit, churn — which "generally can't be directly manipulated sustainably over time." Leaders allocate resources to move inputs; outputs respond downstream.

This is the single most useful reframe available to this room. It dissolves Tech work cannot be tied to revenue — you were never supposed to tie tech work to revenue, you were supposed to find the lever tech controls — and it repairs North Star plus Upstream Scorecard, which is the right instinct anchored to the wrong metric.

The cautionary tale nobody in the room knows

Amazon's first selection input metric was "number of new detail pages created." Teams gamed it by adding "tens, hundreds, even thousands of items" with no demand, filling fulfilment centres with dead inventory. The metric was iterated over roughly a year: detail page views → in-stock percentage → Fast Track In Stock (available for two-day ship) → demand-weighted in-stock.

The safety net was the weekly business review repeatedly asking whether moving the input actually moved the desired output: "controllable input metrics are only important if they drive desirable outcomes."

Read that against Volume-Based Commission with a 50-50 Split, which is a volume input metric with money attached and three months of history.

Note also that Amazon tracked hundreds of interlocking metrics and reviewed 200+ in an hour — the model is not "one metric", it is "one owner per metric plus a forum that challenges the chain."

Sources

Working Backwards: input metrics · Commoncog: Amazon's category-expansion incentives · How Amazon measures

Links: Amazon Working Backwards, North Star Metric, Usage as a proxy for satisfaction.

Asked by The Commission Experimenter at 03:18, first of his three:

"Wie gehe ich damit um, wenn sich Prioritäten ändern in einer Zielplanung? Was in unserer historischen Erfahrung auch im OKR-Modell zu oft passiert, das für so viel Friction sorgt, dass dieses Modell nicht mehr ernst genommen wird, weil du mindestens einmal im Quartal eine solche Sache bekämpfst."

The killer detail: the friction itself destroys the model's credibility. It is not that plans change; it is that changing them costs a fight, at least once a quarter, and after enough fights nobody takes the plan seriously.

Answered in the room: not directly. Quarter by Quarter not Year Detailed reduces the frequency. Goals should be set as hypotheses reframes it. Neither gives a protocol for the change that arrives in week five.

Candidate answers from outside the room

1. Make cancellation the default, not the exception. Shape Up's circuit breaker: a project that doesn't ship in its cycle is cancelled and must win a fresh bet. The change is the mechanism, so it costs nothing politically. Shape Up and the Betting Table
2. Route it to an issues list, not to a re-plan. EOS holds Rocks immovable for 90 days and sends everything else to IDS. EOS and Rocks
3. Class the goal correctly up front. Committed OKRs are explicitly the class where "we will adjust schedules and resources to ensure delivery" — re-planning is permitted by definition. Aspirational ones are expected to move. Committed vs Aspirational OKRs
4. Write the belief the goal rests on, so a change updates a stated belief rather than embarrassing a person. Spotify DIBB and Bets
5. Budget slack so small changes don't need renegotiation at all. Twenty Percent Capacity for New Topics

The common structure: the friction comes from treating change as failure. Every working answer makes change a named, pre-authorised move.

Links: OKRs are too rigid for a dynamic business, How do you review without a documentation burden.

The Commission Experimenter, 03:18: "Wie review ich das effizient ohne den Leuten irgendwie massive Dokumentationspflichten aufzuerlegen oder mich totzumüten?" — without imposing massive documentation duties on people or exhausting myself.

The best-answered question of the session, and the answer arrived from two directions that each have half of it.

The answer, assembled

From the room: Friday Written Report — written, short, in a channel, no meeting. Plus Monday Assignment Meeting to assign, and Means of Verification so a claim points at evidence rather than at itself.

From the record, the three things that make it lightweight rather than merely brief:

1. A hard cap plus a fixed template. Every working example has one: EOS's two words per line (whole scorecard in ~5 minutes), 4DX's 20–30-minute capped session, Basecamp's single automatic question. An open-ended report drifts. See Meeting and documentation load.
2. A designated venue for the discussion you banned. EOS gives status two words and then 60 minutes to solve; 4DX has "clear the path". Reporting with no route to resolution becomes a compliance ritual. See No credible framework bans discussion outright.
3. A durable scoreboard the team maintains — not the leader, and not a photograph of a whiteboard. 4DX Discipline 3 and Rockefeller habit #10 both insist on this.

And the structural saving nobody mentioned: most documentation burden in goal systems is transcription between systems that don't talk to each other — goals in a sheet, numbers in Metabase, progress in a channel. Reducing the number of places a number lives is worth more than shortening any report.

Links: The Bureaucracy Objection, OKR Master.

The Commission Experimenter, 03:18, the third of his three, and the one with the best-developed literature behind it:

"Ich möchte mein Team enable to overperform, das heißt ich setze die Ziele vielleicht bewusst ein bisschen unter dem eigentlichen Ziel… Oder ich setze sie bewusst so hoch an… Wie tariere ich dieses Pensum am besten aus?"

Answered in the room: 50-50 Goal Rule — set targets you'll hit half the time.

What the evidence actually supports

Not a number. A portfolio plus a precondition.

The portfolio (Committed vs Aspirational OKRs):

• Committed goals calibrated near-certain. Field practice sits at 80–90% attainment, and the accounting literature says that is rational for planning, not cowardice — Managers set targets to be hit 8 or 9 times out of 10.
• Aspirational goals with no penalty for missing, where 60–70% is a diagnostic average rather than a target — Where the 70 percent norm actually comes from.

This is also, independently, Locke & Latham's own prescription: stretch goals work under two design conditions — no penalty for missing, and a parallel minimum attainable target that must be hit (Goal-Setting Theory Locke and Latham).

The precondition (The Stretch Goal Paradox): you are only entitled to stretch goals if you have recent strong performance and available slack. Firms in decline reach for stretch goals and nearly always fail. So the first question is not "how ambitious" but "am I allowed to be ambitious right now?"

And the mechanism to be honest about (Stretch goals raise variance not median performance): a stretch goal buys variance, not median improvement — a fatter right tail and a fatter left tail. Ask whether you can survive the left tail.

Direct answer to his either/or: deliberately under-setting to enable overperformance is what the field already does, and it is defensible for committed goals. It is not defensible as your only class, because it forecloses the upside. Run both, and never let pay ride on the aspirational half.

Links: Where the 50 percent rule actually comes from, OKRs at 70 Percent of the Performance Appraisal.

The most interesting question of the session. The Commission Experimenter has built Volume-Based Commission with a 50-50 Split, believes it produces acquisition, retention and account development on its own, and asks whether pushing goals down adds anything but load.

Short answer

No — but the hybrid he is one step away from is probably better than either.

Why not

• Attention, not effort, is what incentives allocate. Incentive pay "serves to direct the allocation of the agents' attention among their various duties", and the optimal intensity on one measured task falls with the difficulty of measuring the others. Onboarding quality, documentation, incident prevention, migrations, product feedback and strategic accounts are all unmeasured here. Multitask Principal-Agent Problem
• Money buys quantity, not quality — corrected correlation .34 with quantity, none with quality. His premise that the individual "does the job as well as they can" regardless is the part the data doesn't support. Financial incentives buy quantity not quality
• Interdependence breaks it. Ops depends on engineering, IT and compliance — none of whom are in the split — so the rational response to a blocked dependency is to attribute rather than escalate. That is the "IT hat nicht funktioniert" excuse, predicted. Reward interdependence must match task interdependence
• Four things a commission structurally cannot do: coordinate across interdependent teams, fund non-revenue bets, sequence work (a commission is a scalar gradient: it can say "more", never "this before that"), or distinguish signal from exogenous customer volume.
• The Quiet Interjector said it in six words: "Du hast ja schon Ziele formuliert." He has goals. He formulated them as an incentive, which means he cannot change them without changing people's pay — the most expensive possible goal-setting mechanism.

What to do instead

Keep the commission as one component; add three or four explicit non-revenue goals covering quality, strategic accounts and cross-team dependencies; set the split by contribution rather than a fixed 50/50. That is what the evidence supports (Hybrid rewards beat pure individual or pure shared) and what practitioners actually run (Snowflake's territory-weighted hybrid; team-based CS targets).

What would change the verdict

Volume-based commission surviving a full year without account-quality drift, small-account starvation, or credit disputes. Three good months is not evidence — Amazon's structurally identical volume metric took about a year to reveal it was being gamed. Input vs Output Metrics

Links: A commission model may replace a goal system, Incentives do not survive contact with employees, The Absent Contrarian.

The Dynamism Sceptic, 57:42, and the one question the room genuinely failed to answer:

"Kannst du das mal für jemanden zeigen, der so etwas [nicht] Messbares macht?" — "Designer oder sowas?" — "Jemand, der eine marktnahe Tätigkeit macht, die aber nicht so vertrieblich messbar ist."

The host said he'd find an example. The conversation moved to commission models instead and never came back.

The answer the room didn't have

Stop measuring output; measure the behaviour change the work is for.

• HEART Framework — Google's UX metrics with the Goals → Signals → Metrics process. A designer's work is measurable through Task success (completion rate, error rate, time-on-task) and Adoption — outcomes they genuinely influence. You are not meant to instrument all five dimensions.
• Input vs Output Metrics — pick the controllable lever the role owns rather than the distant result. This is what P5's "über acht Ecken" objection is really about: he is tracing the chain from the effect backwards, which is always eight steps. Forwards from the lever it is one.
• DORA and SPACE for engineering — with both frameworks' explicit warnings that these metrics must never be used to reward or penalise individuals.
• Skip individual OKRs — the canon's own answer: don't set individual goals in these functions at all. Put the goal on the team.

And the best answer was already in the room, unrecognised. Alvimora OKR Cascade gave IT a user adoption key result over IT's objection ("meine Aufgabe ist es, das Ding zu bauen"). That is exactly the move: not shipped-features, but did-anyone-use-it. The host had solved P5's problem an hour earlier and neither of them noticed.

Links: Tech work cannot be tied to revenue, The Product Counterpart.

The Pre-Launch Founder, 69:02 — a team of four, pre-launch, goals reshuffling "gefühlt nach jedem Gespräch mit einem Advisor". He asked for five to ten minutes. The session broke for a pause and the transcript ends.

So this is the vault's one genuinely unanswered question, and the only one about the stage most founders are actually at.

What was offered, indirectly

Early stage needs no goal system only hustle — The Weekly-Rhythm Founder's view that while the founding team can talk daily, structure is irrelevant. True but incomplete: it tells him to stop worrying, not what to do on Monday.

The Dynamism Sceptic offered the right diagnostic instead: what was the goal last week, what is it today, what was it two weeks ago? Written down, that list is either evidence of learning or evidence of thrash, and you cannot tell which until you can see it.

The best-sourced answer

One number, held for a quarter, that doubles as an advice filter.

• Graham: "If there's one number every founder should always know, it's the company's growth rate." 5–7% weekly is good, 10% exceptional.
• Andreessen: before PMF "the only thing that matters is getting to product/market fit."

The point is not the target — it is that a single number lets him ask of every advisor suggestion: does this move the weekly rate? Advice that doesn't gets written down and ignored. See Startup growth rate as the only metric.

Honest caveat: advisor-driven goal thrash is not a studied phenomenon. The closest usable material is on filtering and sequencing advice. And note the asymmetry he correctly sensed — everything else in the session was designed for problems he is two years away from.

Links: The Pre-Launch Founder, Founders are the operating system.