Frameworks & methodology: how CMOs build systems that drive business results

Week three of the quarter. Three things land the same afternoon: the CROCROConversion Rate Optimization (CRO) is the systematic practice of increasing the percentage of users who complete a desired action, using data, testing, and user research.View full definition → wants to move a slice of brand budget into a partner incentive, your demand lead has committed to a launch date product never agreed to, and someone asks whether the pipelinepipelineAll active sales opportunities across the stages of the sales process, together with their combined potential value and probability of closing.View full definition → figure in last week's deck matches the one finance uses. Nobody is behaving badly. What is missing is an operating system: a written cadence for setting goals, a fixed rhythm for reviewing them, and an explicit map of who decides what. The commercial remit the foundations lesson describes gives you the authority. This is the machinery that lets you spend it.

What a framework actually is (and what it is not)

A framework is a repeatable decision structure that says what gets prioritised, who can move money, and how you know a choice worked. It is not a 2x2 on a slide. It is not a planning template. Three tests: it forces trade-offs, it creates language your CFO and CRO use without translation, and it produces the same output when two different people apply it.

Partial installation is the usual state. Excellent goals with no review rhythm produce a plan nobody has opened since week six. A weekly review with no decision rights turns every meeting into a renegotiation of the plan, which is the most expensive form of alignment there is. Crude versions of all three layers beat one polished layer.

Sub-concept 1: the goal-setting cadence

Salesforce runs on V2MOM: vision, values, methods, obstacles, measures. Marc Benioff wrote the first one in 1999, and the format still cascades, with employees writing their own so that a campaign manager's methods trace up to the company's measures. Two features matter more than the acronym: it is rewritten each year rather than rolled forward, and obstacles are a named section, so the plan carries its own risks instead of burying them in an appendix.

Underneath the annual spine sits a quarterly layer. OKRs came out of Intel and were carried into Google by John Doerr; the mechanic is a performance contract, not a goal format. "Grow brand awarenessbrand awarenessThe degree to which your target audience recognises or recalls your brand, either prompted or unprompted. It measures how present your brand is in people's minds.View full definition →" is not a key result. This is: objective, establish category leadership in enterprise data security; KR1, 35% unaided recall among CISOs in the Q3 survey; KR2, 500 qualified opportunities from new logo accounts; KR3, three tier-one press mentions framing the company as the benchmark.

Two rules keep that layer honest. Grades stay out of compensation, the separation Google has long argued for, because the moment a key result sets a bonus your team writes targets it has already hit. And anything whose feedback loop runs longer than the cycle does not belong in a quarterly KR: unaided recall, organic search compounding, an enterprise deal cycle of nine to twelve months. Those sit on the annual objective with quarterly leading indicators underneath. Otherwise you will kill a working investment in month five for lack of a signal it was never going to give you.

Sub-concept 2: the review rhythm

Amazon's weekly business review is the reference implementation: a fixed meeting, a standing metrics deck, and a hard distinction between output metrics (revenue, margin) and controllable input metrics, the ones a team can move this week. Selection, page speed and in-stock rate move. The share price does not. Bezos also took PowerPoint out of senior meetings in 2004 in favour of narrative memos read in silence at the start, which pushes the argument into writing and leaves the meeting for the decision.

For a marketing org the layering usually lands as: weekly on inputs (creative in market, response time on inboundinboundA strategy that attracts prospects organically via valuable content (blog, SEO, social) rather than interrupting them.View full definition →, sign-up conversion), monthly on pipeline coverage, CACCACCustomer Acquisition Cost (CAC) is the total sales and marketing spend divided by the number of new customers gained in a period. It measures how efficiently you grow.View full definition → payback and mix, quarterly on segmentssegmentsDividing a market into distinct groups of customers who share similar needs, characteristics or behaviours, so each group can be served with a tailored approach.View full definition → and positioningpositioningThe mental space you want your brand to occupy in your target customer's mind relative to alternatives.View full definition →. The failure mode is cadence inflation. Six leads, three review layers, a day of prep each, and you have spent two working weeks a month on reporting before anyone decides anything. When you add a review, name the one you are deleting.

Sub-concept 3: the demand model the reviews run on

Cadence with the wrong numbers on the page is theatre with a calendar. Two models earn a permanent slot.

Jobs-to-be-done, from Clayton Christensen's work, treats the customer as someone trying to make progress in a situation rather than a demographic row. Intercom rebuilt its messaging architecture this way, mapping products to jobs such as "help my support team answer faster without adding headcount" instead of leading with features, and moved from a messaging tool to a customer communications platform, reaching a valuation of roughly $1.3 billion in its 2018 round. Note the self-interest: Intercom published the methodology as a book while selling the software that does the job, so the framework was also demand generationdemand generationMarketing activities designed to attract and capture contact information from prospects interested in your offer, creating a pipeline of potential customers.View full definition →.

Growth accounting is the second: split period-over-period revenue into new acquisition, expansion, contraction and churn. The four flows respond to different spend. A team hitting its new logo number while contraction quietly widens will show flat revenue and a rising payback period, and the review will blame the media plan.

Sub-concept 4: decision rights

Write these down before you need them, as a grid: the decision, the owner, who is consulted, the threshold. A workable default: reallocation inside an approved channel budget up to 10%, the channel owner decides within the quarter; above that, you; segment entry or exit, price story and category name, you with the CEO and CFO. Amazon's language sets the line well. Two-way door decisions, reversible and cheap to undo, go to whoever is closest to the work; one-way doors go up. Sorting by seniority instead of reversibility is what creates the bottleneck at your desk.

Amazon pairs rights with a single-threaded owner, one person whose main job is that outcome, which stops shared accountability turning into none. Writing the grid also tends to reveal that pricing, partner marketing or product naming does not actually belong to you. Better to find that out in a planning meeting than in launch week. Targeting is the one decision you cannot delegate: most teams segment, then refuse to target, staying relevant to five segments and owning none.

Real-world cases

Case 1: Salesforce and the measurement system behind category creation. Benioff's playbook had a named enemy (on-premise software, the no-software logo), a repeatable narrative (the end of software), and measures tied to analyst recognition and share of voice. What made it an operating system rather than a campaign is that the narrative sat inside the V2MOM every function wrote, so sales, product and marketing argued from one document. Revenue went from about $5 million in fiscal 2001 to over $1 billion by fiscal 2009.

Case 2: Amazon and working backwards. Before a product is built, the team writes the press release and the FAQ, and that document is the gate. Marketing drafts the message at the point of decision instead of receiving a brief afterwards. The system does not prevent bad bets: the Fire Phone shipped in 2014 and Amazon took a write-down of roughly $170 million on unsold inventory that year. It makes bets legible and quick to stop, which is the realistic claim for any methodology.

CMO action items

• Publish the cadence on one page and put twelve months of dates in calendars: annual objectives, quarterly key results, monthly commercial review, weekly input review.
• Build the decision rights grid for the ten decisions that recur most, including the ones you suspect you do not own.
• Audit your weekly metrics for controllability. Anything your team cannot move within a week is a scoreboard, so move it to the monthly.
• Take the cadence and the grid to the CFO and CRO before planning opens, for input rather than approval. Leave the board-facing metric set to the framing the board and CFO communication lesson sets out.

Common mistakes that kill results

Mistake 1: framework theatre. A CMO who cites jobs-to-be-done but cannot name a targeting decision it changed has bought vocabulary. If prioritisation looks identical before and after adoption, nothing was installed.

Mistake 2: parallel systems that never meet. Jobs-to-be-done for product marketing, OKRs for performance, segmentation for brand, each run separately, produce contradictory signals. Every segment should map to at least one job and every job to a named segment; otherwise the tension surfaces in the planning meeting instead of the market.

Mistake 3: decision rights the CEO overrides most weeks. Rights people watch being ignored are worse than none, because the team learns that the real system is access. If overrides cluster in one area, the grid is wrong there, so rewrite it to match who decides in practice.

Mistake 4: a cadence that never changes with scale. At $10 million and 50 people, one weekly meeting is the whole operating system. At $500 million the same meeting is a status report and the decisions have migrated to side channels. The reverse fails too: an Amazon-style metrics deck imposed on a twelve-person team generates reporting nobody can act on. Re-cut the cadence at every rough doubling of the org.