AI Engine/Strategy/Objectives + Key Results (OKRs)

Strategy 04

Objectives + Key Results (OKRs)

A target without control over the inputs is not accountability.

The company goal sets the destination. The marketing objective names the outcome marketing can materially influence. Key results define the observable changes that would show progress. Budget, decision rights, time, and talent must support the result before anyone is held responsible for it.

Its boundary: Company leadership sets the vision and company outcome. The CEO and marketing leader agree on the marketing objective, resources, authority, and acceptable risk. Measurement verifies what changed without rewriting the target after the fact.

01Intelligence02Strategy03Creation04Distribution05Pipeline06Lifecycle07Operations08Measurement

The Control Radius

Own the result only as far as the decision rights extend.

Marketing can influence revenue without controlling the entire path to revenue. Sales may control follow-up and closing. Product may control signup conversion and activation. Customer Success may control adoption and renewal. A revenue target can align the company, but it cannot become a clean marketing accountability when essential decisions sit elsewhere.

Separate direct control, shared influence, and downstream ownership before choosing the objective. If another leader can override a material marketing decision, accountability for the affected result is shared.

The Metric Tree

Connect the company outcome to evidence, then to the work.

An objective states the change worth pursuing. A key result measures whether that change is occurring. A supporting KPI explains where movement is coming from. A bet is work chosen because the team believes it can move the result.

Publishing twenty articles is an activity. Increasing qualified organic demand from a declared baseline to an approved target could be a key result. The activity belongs at the bottom of the tree, not where the outcome should be.

The Accountability Contract

Set the outcome, authority, and resources in the same conversation.

The CEO sets the company vision and outcome. The CEO and marketing leader then agree on the marketing objective, budget, hires, decision rights, time horizon, and acceptable risk. Once those conditions are clear, the marketing leader chooses the audience, message, channels, team, priorities, and execution model inside that agreement.

Sales, Product, Customer Success, and Finance should contribute facts and constraints. They should not turn marketing strategy into a group vote. Accountability requires one owner with authority over the choices being judged.

The Shape of Startup Growth

Growth often arrives in steps, not a smooth monthly staircase.

A perfectly linear target assumes that learning, market response, and large wins arrive on schedule. Early-stage companies rarely work that way. A team can improve the baseline every week while an experiment produces little visible movement, then one message, channel, product change, or campaign shifts the trajectory.

The plan should include reliable work, adjacent tests, and bounded high-upside bets. The downside must be explicit. So must the fact that the timing of a major win cannot be guaranteed before the market responds.

Metric Gaming

The easiest number to move is often the least useful.

A single metric can reward behavior that harms the company. More leads can lower sales productivity. More traffic can attract the wrong audience. More output can create rework. Attribution can give credit to the last visible touch while ignoring the system that created demand.

Pair every volume target with a quality or downstream guardrail. Define the metric before the work starts and preserve the definition, source, exclusions, and change history.

Raw trafficEasy to increase with irrelevant attention
Qualified attentionPriority audience, intended page behavior, and a later quality signal
Lead or MQL volumeCan transfer poor-fit work to Sales
Accepted demandDeclared qualification plus Sales acceptance and later conversion quality
Attributed pipelineCan change with the attribution rule
Reconciled pipelineOne definition, source, acceptance rule, and conversion view
Asset volumeRewards production regardless of usefulness
Useful outputTarget-audience response, reuse, business effect, and rework rate
AcquisitionCan hide a broken product or promise
Acquisition + activationThe right customer arrives and reaches verified value

The Operating Cadence

Review decisions, not status theater.

Set the overall marketing objective quarterly or every six months, depending on the company and the work. Keep the scorecard current from source systems. Do not add a recurring meeting merely to read numbers that could have been updated asynchronously.

Bring people together when an assumption breaks, a result crosses a decision threshold, or a resource commitment must change. At the end of the cycle, decide whether to scale, adjust, stop, or replace the bet, then preserve the learning.

The OKR Brief

Make every target inspectable before the work begins.

The brief should be short enough to read and precise enough to audit. If the definition, baseline, target, source, owner, authority, or revision condition is missing, the score will become a negotiation later.

Where AI helps

AI can assemble scorecards from approved systems, flag definition drift, compare scenarios, summarize what changed, and surface the evidence behind an exception. It should not invent a baseline, target, causal explanation, or certainty level. The accountable human chooses the objective and accepts the tradeoff.

Current Tools

Start with the systems already holding the truth.

The best OKR software cannot repair an undefined metric or missing decision right. Use one operating record, connect the actual customer and funnel sources, and add planning software only when budget, headcount, and scenario complexity justify it.

01
NotionDefault operating record
Keeps objectives, key results, owners, timelines, source links, decisions, and linked work in one visible place without adding a specialized platform too early.
Best fitSeed to Series B teams that already use Notion and need a lightweight source of truth.
02
HubSpotFunnel + CRM evidence
Tracks sales, service, marketing, and custom goals against the records already holding lead, pipeline, and customer activity.
Best fitCompanies whose key results depend on agreed CRM definitions and funnel behavior.
03
AmplitudeProduct behavior evidence
Provides governed product metrics, cohorts, and event definitions for objectives involving activation, engagement, retention, or verified customer value.
Best fitProduct-led or hybrid companies that must connect acquisition with what customers do after signup.
04
PigmentIntegrated planning
Connects business planning, budgeting, forecasting, and scenario analysis when spreadsheet models and cross-functional planning have become difficult to govern.
Best fitLarger or more complex teams with real planning complexity. Usually unnecessary for an early team with a clean operating record.

My default: use Notion as the decision record, HubSpot for funnel evidence, and Amplitude for product behavior. Add Pigment only after planning complexity becomes a genuine constraint.

Examples Worth Studying

Two public examples that keep the metric subordinate to the strategy.

These examples show operating choices documented by the companies themselves. They do not prove that one practice caused company performance.

Change the metric when the strategy changes

Amplitude

Amplitude publicly documented replacing its original product North Star after revisiting its strategy, studying product and customer behavior, and concluding that the old metric no longer represented the value it wanted to create.

Lesson: a metric is a model of the strategy. It should not become sacred when the strategy or evidence changes.

Read Amplitude's account
Keep the reporting layer lightweight

GitLab

GitLab's public handbook separates objectives, key results, and initiatives, assigns a directly responsible individual, links progress to existing work records, and asks for short updates instead of duplicating detailed status reports.

Lesson: the OKR layer should expose progress and health without becoming a second project-management system.

View the public handbook

Tool sources: official documentation from Notion, HubSpot, Amplitude, and Pigment.

Operating examples: Amplitude's public account of changing its North Star metric and GitLab's public OKR handbook.

Tools, links, and rankings reviewed Q3 2026. Recheck current capabilities, permissions, and pricing before procurement.

Strategy 04

Make the objective executable.

Agree on the company direction, marketing objective, authority, resources, key results, guardrails, and revision conditions. Fund a portfolio of work, keep the scorecard source-backed, and meet only when the evidence requires a real decision.