Run the 90-day execution cadence
Turn the academy artifacts into a weekly system for shipping, learning, and compounding SEO and AEO results.
Turn the nine artifacts into a weekly operating rhythm, with one objective, one sequenced board, honest capacity, and review gates, so the system ships instead of shelfing.
You will have a sequenced 90-day board with owners, capacity, review gates, and measurable milestones.
The playbook fails when audits, briefs, and dashboards live as disconnected documents. The operating cadence creates one visible queue and protects the small set of work most likely to improve buyer discovery.
IF YOU SKIP IT / Nine good artifacts and no cadence is a folder of documents. The strategy deck nobody opened after week three is the most common way this playbook fails.
- Stage gate
- A checkpoint where work stops until an artifact exists and has been reviewed. Gates are what keep a plan from becoming a wish.
- Capacity
- The real hours each role can give per week, counted before promising anything. The plan's honesty metric.
- Input metric
- What a team shipped (pages updated, pitches sent), reported beside outcomes so movement can be explained.
- Parked list
- Good ideas deliberately deferred. Parking is how you say yes later instead of no forever.
Each phase feeds the next. Don't skip ahead, the artifact at the end is only trustworthy if every phase ran.
- 01Choose outcome
Set one commercial objective, one priority segment, and explicit constraints for the quarter.
- 02Baseline
Complete the demand, technical, page, prompt, source, and measurement artifacts needed for decisions.
- 03Build backlog
Convert findings into outcome-sized work with impact, confidence, effort, dependency, and pass condition.
- 04Load capacity
Reserve named engineering, design, expert, legal, content, and distribution time.
- 05Sequence
Schedule foundational fixes and evidence before dependent pages and promotion.
- 06Operate weekly
Review shipped work, blocked dependencies, quality failures, and next seven-day commitments.
- 07Measure monthly
Read leading and lagging indicators with context and document decisions.
- 08Close quarter
Audit every artifact, preserve learnings, reprioritize the backlog, and set the next outcome.
- ✓You now have
90-day execution board
Set one 90-day objective
The objective should describe a buyer-visible change, not an activity count.
'Publish twelve pieces of content' is not an objective; it is a way to stay busy while nothing changes for the buyer. Teams hit activity targets all the time in programs that produced no observable difference. The objective has to name a change someone outside your company could notice: a shortlist you now appear on, a cohort you now win, demos arriving pre-sold.
Then write the guardrails, the list of what this cycle will not attempt. Guardrails do their work in week six, when someone important discovers an exciting keyword and the parked list from Module 01 gives their idea a dignified place to wait.
- 01
Choose the segment and demand area from Module 01.
- 02
Name the visibility and pipeline outcomes you expect to influence.
- 03
Add guardrails: what this cycle will not attempt.
- Write the objective in one sentence using the template: become a credible option for [segment] on [demand area], measured by [cohort metric], [visibility metric], and [pipeline metric], by [date].
- List three guardrails: things this cycle will not attempt, by name, with the parked list linked.
- Get one leadership signature on the doc. Not a meeting, a signature. It is what you point at in week six.
The objective names a buyer-visible outcome with numbers attached, and the "not this cycle" list has real things on it that someone wanted.
- PRO TIP
Pick an objective a cohort can actually prove in 90 days. Visibility movement yes; closed revenue usually needs two quarters, so frame it as leading evidence.
- WATCH OUT
If the objective contains 'publish N pieces', it is an activity count wearing a suit. Rewrite it as a change a buyer could notice.
OPTIONAL / AI CO-PILOTRun this lesson with AI, the prompt, sources and expected output+
Choose the business priority, interpret exceptions, protect confidential data, challenge weak evidence, and approve the final decision.
Clean, classify, compare, calculate, and draft rows from supplied evidence. AI may surface patterns; it does not own strategy.
Linear / Asana / ClickUp, plus performance inputs from GSC + audit + citation tools.
| Commercial outcome | Segment | Baseline | Target | Constraint | Executive owner | Decision date |
|---|---|---|---|---|---|---|
| Increase qualified demos | Mid-market agencies | 74 / qtr | 100 / qtr | 1 engineer | VP Marketing | 2026-10-01 |
CLAUDE / CHATGPT PROMPTAnalyze the evidence without outsourcing the decision+
Remove personal or confidential data. Attach the named exports, explain every column, and tell the model when the dataset was collected.
You are assisting a human B2B SaaS SEO and AEO operator with Module 10: Run the 90-day execution cadence. Current lesson: Set one 90-day objective Objective: The objective should describe a buyer-visible change, not an activity count. Required artifact: 90-day execution board BUSINESS CONTEXT I WILL PROVIDE - Product, category, target market, pricing model, sales motion, and geography - The priority customer segment and the commercial outcome for this 90-day cycle - Operating backlog exported from Linear / Asana / ClickUp - Performance inputs exported from GSC + audit + citation tools - Definitions for any internal fields, stages, scores, and abbreviations TASK 1. Choose the segment and demand area from Module 01. 2. Name the visibility and pipeline outcomes you expect to influence. 3. Add guardrails: what this cycle will not attempt. REQUIRED OUTPUT Return a table using these exact columns: Commercial outcome | Segment | Baseline | Target | Constraint | Executive owner | Decision date. For every recommendation, cite the source row, URL, call note, or data point that supports it. Add a confidence column in your analysis: High, Medium, or Low. List missing evidence separately instead of guessing. Finish with a section named HUMAN DECISIONS REQUIRED. RULES - Do not invent search volume, revenue, customer statements, product capabilities, or competitor facts. - Do not treat correlation as causation. - Preserve contradictory evidence and explain why it conflicts. - Do not make the final priority or publishing decision. Prepare the evidence for a human owner. - Use this example only as a format reference, not as evidence: “Become a credible shortlist option for agency reporting buyers,” measured across 30 prompts, 12 queries, and qualified demos.
The objective names a buyer-visible outcome. A named human owner must verify this before the lesson is complete.
“Become a credible shortlist option for agency reporting buyers,” measured across 30 prompts, 12 queries, and qualified demos.
Sequence dependencies
Technical access and evidence often need to precede content volume.
Order here is physics, not preference. Content on an unindexable template is invisible. Measurement bolted on after launch has no baseline. And outreach started in week ten will not publish before the quarter closes, because editors reply on their calendar, not yours: a pitch sent in week 3 lands around week 9.
So the first two weeks are always the same, unblocking and baseline, however boring that feels while competitors are shipping posts. Then money pages and passages, because they are the destination everything else points at. Start the relationship work early and let it cook.
● WEEKLY SHIP REVIEW ◆ MONTHLY PROMPT RE-RUN
- GATE / END W2 A dated baseline everyone has seen and agreed on.
- GATE / END W6 Every commercial page fetchable, current, and quotable.
- GATE / END W12 Cohort win rate moving against a named competitor.
- 01
Put blocking technical fixes and measurement setup in Weeks 1-2.
- 02
Sequence money pages and answer passages before supporting content and source promotion.
- 03
Start relationship-driven source work early because response and publication times vary.
- Set up the board in whatever your team already uses (Notion, Trello, Linear) with columns: Done, Evidence/QA, This week, Blocked, Next.
- Every card gets a dependency field filled in before a date. If the dependency is another team, name the person, not the department.
- Fix weeks 1-2 as unblocking and baseline only, and start source outreach cards by week 3 regardless of when they will pay off.
Weeks 1-2 contain only unblocking and baseline work, and everything on the board lists what it depends on before it lists its ship date.
- PRO TIP
Editors and partners reply on their own schedule. Outreach started week 3 lands in week 9; outreach started week 9 lands next quarter.
- WATCH OUT
The most common sequencing failure is content before crawlability. It feels productive and it is invisible.
OPTIONAL / AI CO-PILOTRun this lesson with AI, the prompt, sources and expected output+
Choose the business priority, interpret exceptions, protect confidential data, challenge weak evidence, and approve the final decision.
Clean, classify, compare, calculate, and draft rows from supplied evidence. AI may surface patterns; it does not own strategy.
GSC + audit + citation tools, plus system of record from Docs / Sheets.
| Work item | Impact | Confidence | Effort | Dependency | Pass condition | Priority |
|---|---|---|---|---|---|---|
| Fix integration canonicals | High | High | Medium | Platform sprint | 47 self-canonical URLs | P1 |
CLAUDE / CHATGPT PROMPTAnalyze the evidence without outsourcing the decision+
Remove personal or confidential data. Attach the named exports, explain every column, and tell the model when the dataset was collected.
You are assisting a human B2B SaaS SEO and AEO operator with Module 10: Run the 90-day execution cadence. Current lesson: Sequence dependencies Objective: Technical access and evidence often need to precede content volume. Required artifact: 90-day execution board BUSINESS CONTEXT I WILL PROVIDE - Product, category, target market, pricing model, sales motion, and geography - The priority customer segment and the commercial outcome for this 90-day cycle - Performance inputs exported from GSC + audit + citation tools - System of record exported from Docs / Sheets - Definitions for any internal fields, stages, scores, and abbreviations TASK 1. Put blocking technical fixes and measurement setup in Weeks 1-2. 2. Sequence money pages and answer passages before supporting content and source promotion. 3. Start relationship-driven source work early because response and publication times vary. REQUIRED OUTPUT Return a table using these exact columns: Work item | Impact | Confidence | Effort | Dependency | Pass condition | Priority. For every recommendation, cite the source row, URL, call note, or data point that supports it. Add a confidence column in your analysis: High, Medium, or Low. List missing evidence separately instead of guessing. Finish with a section named HUMAN DECISIONS REQUIRED. RULES - Do not invent search volume, revenue, customer statements, product capabilities, or competitor facts. - Do not treat correlation as causation. - Preserve contradictory evidence and explain why it conflicts. - Do not make the final priority or publishing decision. Prepare the evidence for a human owner. - Use this example only as a format reference, not as evidence: Fix indexation -> ship comparison -> publish supporting test -> pitch benchmark -> retest prompts.
Dependencies determine the sequence. A named human owner must verify this before the lesson is complete.
Fix indexation -> ship comparison -> publish supporting test -> pitch benchmark -> retest prompts.
Plan against real capacity
A smaller finished plan beats an overloaded backlog.
Every plan gets built for the team you wish you had. Then reality checks in: the product expert has four hours a month, not twelve, and the plan quietly becomes fiction that everyone still reports against. Honest capacity planning feels pessimistic and is the only optimism that survives a quarter.
Count real hours by role, price each work item in those hours, and hold 20% back. The reserve is not slack. It is the budget for whatever Module 02 finds, and it will find something. A plan with no reserve converts every discovery into a schedule crisis.
- 01
List weekly capacity for SEO, engineering, product experts, writing, design, customer marketing, and analytics.
- 02
Estimate work in reviewable units and expose external dependencies.
- 03
Reserve 20% of capacity for discoveries, fixes, and revisions.
- Make a capacity table: role, person, honest hours per week. Have each person confirm their own number.
- Estimate every board card in hours from that table, and stop adding cards when 80% of capacity is allocated.
- Pre-book the scarcest resource (usually product experts) as recurring calendar holds for the whole quarter.
Every work item has an hours estimate a named person agreed to, and at least 20% of total capacity is unassigned on day one.
- PRO TIP
The 20% reserve is not slack; it is the budget for what Module 02 will find. Plans without reserve convert every discovery into a crisis.
- WATCH OUT
Writer hours are rarely the constraint. Expert review hours are. Plan around the bottleneck, not the byline.
OPTIONAL / AI CO-PILOTRun this lesson with AI, the prompt, sources and expected output+
Choose the business priority, interpret exceptions, protect confidential data, challenge weak evidence, and approve the final decision.
Clean, classify, compare, calculate, and draft rows from supplied evidence. AI may surface patterns; it does not own strategy.
Docs / Sheets, plus weekly commitments from Slack / meeting cadence.
| Week | Workstream | Owner | Capacity hrs | Dependency | Blocker | Escalation |
|---|---|---|---|---|---|---|
| 3 | Technical | Engineering | 16 | QA environment | Release freeze | VP Product |
CLAUDE / CHATGPT PROMPTAnalyze the evidence without outsourcing the decision+
Remove personal or confidential data. Attach the named exports, explain every column, and tell the model when the dataset was collected.
You are assisting a human B2B SaaS SEO and AEO operator with Module 10: Run the 90-day execution cadence. Current lesson: Plan against real capacity Objective: A smaller finished plan beats an overloaded backlog. Required artifact: 90-day execution board BUSINESS CONTEXT I WILL PROVIDE - Product, category, target market, pricing model, sales motion, and geography - The priority customer segment and the commercial outcome for this 90-day cycle - System of record exported from Docs / Sheets - Weekly commitments exported from Slack / meeting cadence - Definitions for any internal fields, stages, scores, and abbreviations TASK 1. List weekly capacity for SEO, engineering, product experts, writing, design, customer marketing, and analytics. 2. Estimate work in reviewable units and expose external dependencies. 3. Reserve 20% of capacity for discoveries, fixes, and revisions. REQUIRED OUTPUT Return a table using these exact columns: Week | Workstream | Owner | Capacity hrs | Dependency | Blocker | Escalation. For every recommendation, cite the source row, URL, call note, or data point that supports it. Add a confidence column in your analysis: High, Medium, or Low. List missing evidence separately instead of guessing. Finish with a section named HUMAN DECISIONS REQUIRED. RULES - Do not invent search volume, revenue, customer statements, product capabilities, or competitor facts. - Do not treat correlation as causation. - Preserve contradictory evidence and explain why it conflicts. - Do not make the final priority or publishing decision. Prepare the evidence for a human owner. - Use this example only as a format reference, not as evidence: One expert article requiring product review may consume more constrained capacity than three page refreshes.
Work fits real team capacity. A named human owner must verify this before the lesson is complete.
One expert article requiring product review may consume more constrained capacity than three page refreshes.
Run the operating rhythm
Use a short cadence that keeps quality and learning visible.
The rhythm is the system. Weekly, thirty minutes, one board: what shipped, what is blocked, what is next, what evidence is needed. Monthly, re-run the prompts and re-read the cohorts. Day 90, write down what worked, what missed, and the next compounding bet.
Every review looks at evidence instead of effort. The question is never 'were we busy'; it is 'did the artifact ship, did the cohort move, did the source publish'. Same four charts every time, so the room can spot a real change in ten seconds. And on a quiet week, a five-minute 'nothing blocked, queue confirmed' still counts. Skipping is how cadences die.
- 01
Weekly: review shipped work, blockers, next queue, and evidence needed.
- 02
Monthly: rerun prompts, inspect cohorts, review influential sources, and update priorities.
- 03
Day 90: document results, misses, lessons, and the next compounding bet.
- Create the recurring invites now: weekly 30-minute review (fixed agenda: shipped, blocked, next, evidence needed), monthly re-run day, and the day-90 close-out.
- Run the weekly from the board, not from memory, and end it by confirming next week's queue.
- At day 90: write expected vs actual per artifact, three lessons, and the next cycle's compounding bet, then loop back to Module 01 with everything you learned.
The weekly and monthly reviews are recurring calendar events with owners, and the day-90 close-out has a date before the quarter starts.
- PRO TIP
Same four charts in every review. Familiarity is what lets a room spot a real change in ten seconds.
- WATCH OUT
Skipping the review on a quiet week is how the cadence dies. A five-minute 'nothing blocked, queue confirmed' still counts.
OPTIONAL / AI CO-PILOTRun this lesson with AI, the prompt, sources and expected output+
Choose the business priority, interpret exceptions, protect confidential data, challenge weak evidence, and approve the final decision.
Clean, classify, compare, calculate, and draft rows from supplied evidence. AI may surface patterns; it does not own strategy.
Slack / meeting cadence, plus operating backlog from Linear / Asana / ClickUp.
| Artifact | Expected | Actual | Evidence | Learning | Carry forward | Owner |
|---|---|---|---|---|---|---|
| Money pages | 5 shipped | 4 shipped | Index + CRM | Proof took longer | 1 comparison | Growth |
CLAUDE / CHATGPT PROMPTAnalyze the evidence without outsourcing the decision+
Remove personal or confidential data. Attach the named exports, explain every column, and tell the model when the dataset was collected.
You are assisting a human B2B SaaS SEO and AEO operator with Module 10: Run the 90-day execution cadence. Current lesson: Run the operating rhythm Objective: Use a short cadence that keeps quality and learning visible. Required artifact: 90-day execution board BUSINESS CONTEXT I WILL PROVIDE - Product, category, target market, pricing model, sales motion, and geography - The priority customer segment and the commercial outcome for this 90-day cycle - Weekly commitments exported from Slack / meeting cadence - Operating backlog exported from Linear / Asana / ClickUp - Definitions for any internal fields, stages, scores, and abbreviations TASK 1. Weekly: review shipped work, blockers, next queue, and evidence needed. 2. Monthly: rerun prompts, inspect cohorts, review influential sources, and update priorities. 3. Day 90: document results, misses, lessons, and the next compounding bet. REQUIRED OUTPUT Return a table using these exact columns: Artifact | Expected | Actual | Evidence | Learning | Carry forward | Owner. For every recommendation, cite the source row, URL, call note, or data point that supports it. Add a confidence column in your analysis: High, Medium, or Low. List missing evidence separately instead of guessing. Finish with a section named HUMAN DECISIONS REQUIRED. RULES - Do not invent search volume, revenue, customer statements, product capabilities, or competitor facts. - Do not treat correlation as causation. - Preserve contradictory evidence and explain why it conflicts. - Do not make the final priority or publishing decision. Prepare the evidence for a human owner. - Use this example only as a format reference, not as evidence: A 30-minute weekly review uses one board: Done, Evidence/QA, This week, Blocked, Next.
Weekly and monthly reviews are scheduled. A named human owner must verify this before the lesson is complete.
A 30-minute weekly review uses one board: Done, Evidence/QA, This week, Blocked, Next.
Assign every surface an owner
AI reads surfaces owned by six different teams. The program only scales when each surface has a named owner, a brief, and a goal that was already theirs.
Google read one thing, so SEO was one team's job. Answer engines read your site, your G2 reviews, your partner listings, your docs, your founder's Reddit replies, and the promises your sales team makes on calls. Six surfaces, six teams, and not one of them thinks it works in search.
So the SEO lead becomes a conductor: mapping each surface to its owner, writing the brief, tracking the result, and owning almost none of it directly. The move that makes this work is never asking a team to adopt your goal. Find the goal they already have, show how the brief serves it, and let them keep the credit. CS gets review volume for their NPS story. PMM gets positioning that finally stays consistent. You get consensus, which is the only thing the model was ever measuring.
Conductor, not owner: writes the briefs, tracks the scorecard, owns none of the six surfaces directly.
- 01
Map every surface AI reads to the team that controls it: product and pricing pages to PMM, docs to Product, reviews and customer quotes to Customer Success, partner and directory listings to Partnerships, press and community to PR or founders, call transcripts to Sales.
- 02
Set two goal levels: channel goals (cohort win rate, organic plus AI-referred traffic, self-reported attribution) for the program, and campaign goals (visibility on one attribute, citation of one page, sentiment on one objection) for initiatives. Never tie a single campaign to revenue.
- 03
Give each owning team a brief that serves a goal they already have, and report input metrics (pages updated, profiles corrected, pitches sent) beside the outputs so movement can be explained.
- Copy the six-surface table from the diagram above into a doc and replace the team names with actual people at your company. Gaps in the owner column are the finding.
- Write each owner a half-page brief: the surface, what AI reads from it, the one campaign goal, and how it serves a goal they already have. Deliver it in a 15-minute conversation, not an email.
- Add an input-metrics row per team to the Module 09 scorecard (reviews requested, profiles corrected, pitches sent) so each team's shipping is visible next to the outcomes.
Every surface AI reads has a named owner and a delivered brief, each team's goal is written in their language, and the scorecard shows input metrics beside outputs.
- PRO TIP
Lead with what they get: CS gets review volume for their NPS story, PMM gets consistent positioning, PR gets a measurable outcome for placements. The brief is a gift, not a tax.
- WATCH OUT
A surface assigned to a department is unowned. 'Customer Success' is nobody; 'Dana in CS' is an owner.
- NO BUDGET?
Team of three? One person can hold several surfaces, but write the map anyway. The gaps you cannot cover become the honest case for hiring or for hiring us.
OPTIONAL / AI CO-PILOTRun this lesson with AI, the prompt, sources and expected output+
Choose the business priority, interpret exceptions, protect confidential data, challenge weak evidence, and approve the final decision.
Clean, classify, compare, calculate, and draft rows from supplied evidence. AI may surface patterns; it does not own strategy.
Linear / Asana / ClickUp, plus performance inputs from GSC + audit + citation tools.
| Surface | What AI reads from it | Owning team | Named owner | Brief delivered | Campaign goal | Input metric |
|---|---|---|---|---|---|---|
| G2 profile + reviews | Category, sentiment, recency | Customer Success | Head of CS | Yes | Sentiment on "support quality" | 8 reviews / month |
CLAUDE / CHATGPT PROMPTAnalyze the evidence without outsourcing the decision+
Remove personal or confidential data. Attach the named exports, explain every column, and tell the model when the dataset was collected.
You are assisting a human B2B SaaS SEO and AEO operator with Module 10: Run the 90-day execution cadence. Current lesson: Assign every surface an owner Objective: AI reads surfaces owned by six different teams. The program only scales when each surface has a named owner, a brief, and a goal that was already theirs. Required artifact: 90-day execution board BUSINESS CONTEXT I WILL PROVIDE - Product, category, target market, pricing model, sales motion, and geography - The priority customer segment and the commercial outcome for this 90-day cycle - Operating backlog exported from Linear / Asana / ClickUp - Performance inputs exported from GSC + audit + citation tools - Definitions for any internal fields, stages, scores, and abbreviations TASK 1. Map every surface AI reads to the team that controls it: product and pricing pages to PMM, docs to Product, reviews and customer quotes to Customer Success, partner and directory listings to Partnerships, press and community to PR or founders, call transcripts to Sales. 2. Set two goal levels: channel goals (cohort win rate, organic plus AI-referred traffic, self-reported attribution) for the program, and campaign goals (visibility on one attribute, citation of one page, sentiment on one objection) for initiatives. Never tie a single campaign to revenue. 3. Give each owning team a brief that serves a goal they already have, and report input metrics (pages updated, profiles corrected, pitches sent) beside the outputs so movement can be explained. REQUIRED OUTPUT Return a table using these exact columns: Surface | What AI reads from it | Owning team | Named owner | Brief delivered | Campaign goal | Input metric. For every recommendation, cite the source row, URL, call note, or data point that supports it. Add a confidence column in your analysis: High, Medium, or Low. List missing evidence separately instead of guessing. Finish with a section named HUMAN DECISIONS REQUIRED. RULES - Do not invent search volume, revenue, customer statements, product capabilities, or competitor facts. - Do not treat correlation as causation. - Preserve contradictory evidence and explain why it conflicts. - Do not make the final priority or publishing decision. Prepare the evidence for a human owner. - Use this example only as a format reference, not as evidence: Customer Success owns review velocity with a brief that supports their NPS goal; SEO acts as conductor, supplying briefs and tracking, and owns none of the surfaces outright.
Weekly and monthly reviews are scheduled. A named human owner must verify this before the lesson is complete.
Customer Success owns review velocity with a brief that supports their NPS goal; SEO acts as conductor, supplying briefs and tracking, and owns none of the surfaces outright.
The workspace shows a live 90-day board: the quarter objective, the prioritized backlog, the capacity plan with dependencies, and the close-out review. Notice the 20% capacity held in reserve. That is not slack; that is the budget for everything you will discover once real work starts.
90-day search operating system
Sequence diagnosis, production, distribution, engineering, and measurement around dependencies and team capacity.
Capacity by workstream
A roadmap fails when content, engineering, experts, and distribution are planned independently.
Check each item only after the artifact meets the standard. Progress saves on this device; use the academy-home backup to move it elsewhere.