A shared operating model, not a handoff
Brand-agency AEO partnerships fail in predictable ways. The agency builds a strategy the brand can't approve fast enough to act on. The brand makes structural edits that break the exact elements designed to earn recommendations. Data lives in one place and gets summarized into a slide deck for the other. By week eight both sides are frustrated and neither can explain why the work isn't moving.
This playbook prevents that. It's a shared operating model for brands and agencies running AEO together, on one premise: both sides see the same data, agree on the same structure, and own clearly separated responsibilities from day one. It uses LBOX as the shared operating system — the same subject, the same Answers & Citations data, the same benchmarks — throughout the engagement.
Ardent context: a mid-market B2B SaaS company. Its in-house team (Content Lead, Brand Manager, VP Marketing) engaged an AEO agency to run strategy and production while Ardent kept control of brand voice, final approvals, and internal reporting. Two competitors were already recommended in AI answers for Ardent's core questions; Ardent was in none. Ardent's Share of Answer moved from 4% to 29% in 90 days.
Who it's for: brand-side Content Leads, Brand Managers, VP Marketing, and Digital leads who need to see exactly what the agency is doing and how to read the shared data; and agency-side Account Managers and Strategists who need a clear structure for running a brand client through 90 days without losing control of the content architecture or the timeline.
The shared operating principle: neither side can run this alone. The agency brings AEO strategy, LBOX expertise, and content architecture. The brand brings subject-matter authority, internal approvals, and the business context that makes content credible. It works only when both hold their role without crossing into the other's lane.
The metric everything reports to: Share of Answer
Share of Answer (SoA) = the share of answers, on a frozen set of real buyer questions, in which the engines recommend the brand — not merely mention it — over n runs per question × engine, with a confidence interval. Anchor on recommended, not mentioned; score unbranded questions; attribute movement with difference-in-differences against a control. Both sides read SoA off the same LBOX dashboard.
How the partnership model is structured
Three phases, each with defined outputs from both sides. The phase structure mirrors the agency and brand playbooks, but every deliverable is split into two columns: what the agency produces and what the brand provides or approves.
| Phase | Timeframe | Agency primary role | Brand primary role | Key output |
|---|---|---|---|---|
| 1 · Foundation | Days 1–30 | Strategy, subject setup, question mapping, briefs | Access, approvals, content inventory, stakeholder alignment | Shared Question Priority Map + signed-off briefs |
| 2 · Execution | Days 31–60 | Production, technical rollout, SoA tracking | Brand review within SLA, publishing access | Published content, schema, first movement report |
| 3 · Optimization | Days 61–90 | SoA audit, remediation, displacement | Approval of remediation, Q2 roadmap sign-off | 90-day report + Q2 roadmap |
The shared LBOX workspace
Before week one, the agency configures the LBOX subject and grants the brand team view access. This is non-negotiable. Partnerships where the brand only sees summarized slide decks consistently underperform those where both sides look at the same dashboard.
Brand-side access enables: the Brand Manager and VP Marketing verify the SoA baseline independently; the Content Lead reviews Sources directly before approving briefs; the VP pulls the monthly report without waiting for a presentation; both sides flag anomalies (unexpected competitor recommendations, sentiment shifts) in real time.
Agency workspace responsibilities: configure the subject (brand domain, 2–3 competitor domains, 20–30 questions); grant brand-side view access before the week-one kickoff; set up tracked pages as content publishes; generate and share the monthly white-labeled report at the start of each month.
Role map — who owns what
The most common brand-agency failure is role ambiguity. This table is the definitive split.
| Responsibility | Agency | Brand |
|---|---|---|
| LBOX subject setup & configuration | Owns | View access |
| Question mapping & prioritization | Leads | Reviews & approves |
| Content brief development | Leads | Reviews for brand accuracy |
| Content production | Leads | Reviews for voice; approves within SLA |
| Technical (AEO Audit, schema, indexing) | Owns | Provides CMS access |
| SoA tracking & reporting | Owns | Receives & reviews |
| Internal stakeholder reporting | Provides data | Owns |
| Final publication decision | Recommends | Approves |
| Remediation decisions | Diagnoses & recommends | Approves within 24 hours |
| Q2 roadmap | Leads | Approves & funds |
Phase 1 · Foundation (Days 1–30)
Objective: establish a shared SoA baseline in LBOX, align on question priorities, and produce a signed-off brief package before any production. Neither side moves to Phase 2 without completing both columns.
Week 1 — Kickoff, subject setup & baseline
Agency: configure the subject (brand domain, 2–3 competitors, 20–30 questions); run the baseline AI Visibility scan (Share of Answer, Share of Voice, citation frequency across the full engine panel); grant brand view access before the kickoff call ends; walk the brand team through Answers & Citations and Sources during the call — both sides should read the data independently by end of week one.
Brand: provide the full content inventory (every URL, date, topic); confirm the 2–3 competitor domains (the brand often knows the landscape better); confirm CMS access for the technical contact (without it, schema and indexing stall in Phase 2).
Deliverable: shared subject live with brand view access confirmed; baseline SoA documented; competitor landscape reviewed by both sides.
Ardent: the agency configured 28 questions across three clusters. The kickoff included a live walkthrough of Answers & Citations showing Competitor A recommended in 11 of 28 questions and B in 8. Ardent's VP saw the data directly and called it "the clearest picture of our AI visibility gap we've ever seen." CMS access was provided by end of day.
Week 2 — Question mapping & prioritization
Agency: build the Question Priority Matrix using Topics, Query Fanout, and Prompt demand; segment into Quick Wins, Core Targets, Stretch Goals; present with the LBOX data visible as the source, not a standalone spreadsheet.
Brand: review the matrix and flag questions that don't align with current business priorities, product lines, or seasonal focus; add high-value questions the agency missed (brand teams know the exact phrasing buyers use).
A critical note on approval: brand teams sometimes deprioritize Quick Win questions as too narrow. That's a mistake — Quick Wins earn recommendations fast, which moves SoA, which creates internal proof before the harder Core Target work pays off. Explain this with the Prompt demand data during the week-two review.
Deliverable: approved Question Priority Matrix (3-tier). Both sides sign off before brief development.
Ardent: in a 45-minute working session with Topics open, the Content Lead flagged one Quick Win as seasonally irrelevant (product being sunset) and replaced it with one from internal sales-call data. Final matrix: 7 Quick Wins, 12 Core Targets, 9 Stretch Goals.
Week 3 — Content brief development
Agency: build briefs for the first wave of Quick Wins (5–7), each by examining the currently cited pages in Sources; specify the target question, the 40–60 word self-contained opening answer, the H2/H3 from Query Fanout, and evidence requirements; share with a written explanation of why each structural element exists.
Brand: review for factual accuracy and voice within 48 hours; do NOT alter structural elements (opening answer, H2/H3, evidence) without first raising it with the AM; flag issues with a specific note, not "this doesn't sound like us."
The brief review protocol — the most friction-prone handoff. Brand reviewers instinctively want to soften the opening answer, swap technical headings for brand language, and remove competitor comparisons; all three reduce the chance of a recommendation. Put a one-paragraph note atop every brief explaining what each element does. When pushback comes, open Sources together and show the cited competitor's opening paragraph. The comparison closes most arguments.
Deliverable: approved brief package (5–7). Both sides sign off before production.
Ardent: the Brand Manager asked to lead with product positioning instead of a direct answer; the AM opened Sources, showed Competitor A's cited page, and the original structure was approved within 24 hours.
Week 4 — Phase 1 review & production planning
Agency: generate the 30-day report as the foundation of the review; present the SoA baseline, competitor Share of Voice, and the matrix to the VP; confirm the Phase 2 plan (schedule, review SLAs, publishing workflow, technical timeline).
Brand: VP reviews and formally approves before Phase 2; confirm the review SLA in writing — 48 hours for drafts, 24 hours for remediation (delays in brand review are the single most common cause of Phase 2 compression); confirm the cadence — biweekly working reviews plus a monthly executive summary from the LBOX report.
Deliverable: Phase 1 review complete; production plan approved; SLAs confirmed in writing; Phase 2 begins day 31.
Ardent: the VP reviewed Ardent's 4% SoA against Competitor A's 31%, approved the plan, confirmed the 48-hour SLA by email, and asked for the monthly report sent directly to her inbox so she could track progress independently.
Phase 2 · Execution (Days 31–60)
Objective: produce and publish optimized content, ship the technical layer, and run the first SoA check against day zero. The 48-hour review SLA is not bureaucracy — it's the mechanism that keeps the timeline intact.
Week 5 — First-wave production
Agency: assign writers and manage the workflow; before assigning, return to Sources and verify the cited pages haven't changed since week three; enforce brief adherence — flag and correct the common drift (softening the opening answer into a brand narrative) before brand review.
Brand: review drafts within 48 hours for accuracy, voice, and missing subject-matter context; do NOT review for structure (opening answer, H2/H3, evidence, tables) — that's the agency's domain; if a structural element feels wrong, raise it with the AM, not the writer.
The review protocol in practice: the agency sends each draft with a two-paragraph cover note — what the piece is designed to do, which structural elements are non-negotiable, and what specific feedback is needed. This keeps reviews focused.
Deliverable: first drafts reviewed and approved, ready for publication.
Ardent: the Content Lead reviewed all five drafts within 36 hours, flagged two factual corrections, and approved the structure without changes; the AM updated both within 24 hours.
Week 6 — Publication & indexing
Agency: coordinate publication with the brand's CMS; if the agency publishes, confirm the brand reviewed the live page before indexing; submit URLs for indexing; add live URLs to tracked pages with dates noted.
Brand: provide final publication approval, or publish directly if it retains CMS control; do NOT delay for extra rounds of editing — a piece approved in week five publishes in week six.
The structural-edit trap: the most damaging move at publication is a new stakeholder (legal, product, a senior leader) seeing the content for the first time and requesting opening-answer changes. Prevent it in week four: "Who else will see this before it publishes?" Get them into the brief review in week three, not the publication review in week six.
Deliverable: all Quick Wins published; URLs tracked; indexing requested.
Ardent: the Head of Product flagged a factual correction (accepted) and an opening-answer change (declined, with the Sources comparison); the piece published on schedule.
Week 7 — Technical audit, schema & AEO review
Agency: run the AEO Audit on all published pages, resolve crawlability issues within 72 hours; implement schema (FAQPage / HowTo / Article / Product); verify canonicals and indexing.
Brand: provide CMS access if the agency implements directly; review/approve metadata or canonical changes within 24 hours.
LBOX workflow: the AEO Audit checks whether AI crawlers can reach and extract each page. Share results directly from LBOX, not as a summary — if a page has a crawl issue, show the brand the specific flag so they understand what's being fixed and why.
Deliverable: audit results shared with the brand; schema implemented; indexing/crawlability report.
Ardent: the audit flagged one page with a render issue blocking extraction; the agency showed the SEO Lead the flag in LBOX; it was fixed in 48 hours and confirmed by re-running the audit.
Week 8 — First SoA check & Core Target briefs
Agency: run the first check in Answers & Citations (filter to ~14 days since publication); check SoA against baseline; open Sentiment and flag any unfavorable framing; begin Core Target briefs.
Brand: attend the biweekly working review and read the data directly in LBOX, not from a slide; review Core Target briefs within 48 hours.
What to share at week eight: share Answers & Citations live during the review, not a screenshot. Walk through every new recommendation — which question triggered it, which engine, and what the answer actually says. The answer text is the most valuable data point; it reveals what the engine extracted and why.
Deliverable: first SoA movement report; Core Target briefs delivered.
Ardent: the week-eight check showed the first two recommendations (Perplexity, Google AI Overviews); SoA moved 4% → 11%. The AM walked the VP through both live in Answers & Citations; she asked to see the actual answer text, approved the Core Target briefs the next day.
Phase 3 · Optimization (Days 61–90)
Objective: analyze the data at scale, remediate underperformers, pursue displacement, and build the Q2 roadmap together.
Week 9 — Full SoA audit & gap analysis
Agency: run the full audit across all questions and engines using Answers & Citations, Sources, Topics, and Sentiment in sequence; build the Gap Analysis Report (expected vs. actual, difference-in-differences vs. control); identify underperformers (live 3+ weeks, zero recommendations); diagnose from the actual answer text.
Brand: review the Gap Analysis within 48 hours; flag underperformers where you have additional subject-matter context; do NOT request a full rewrite before the agency has diagnosed the specific structural cause.
LBOX workflow: share the full audit directly during the week-nine review — Answers & Citations, the Sources competitor comparison, and Sentiment, together, both sides on the same data at the same time.
Deliverable: full SoA audit + gap analysis, shared with the brand.
Ardent: recommendations on 4 of 5 Quick Wins and 2 of 6 Core Targets; the one underperformer's cited competitors all included a price comparison it lacked; the brand confirmed they were comfortable including pricing, which unblocked the remediation immediately.
Week 10 — Remediation & structural updates
Agency: build remediation briefs from the answer-text diagnosis and the Sources comparison; submit with a specific note on what changed and why, referencing the data that drove it; re-submit for indexing and update tracked pages with remediation dates.
Brand: approve remediation within 24 hours (targeted structural fixes, not rewrites); do NOT use the review as an opportunity for a broader refresh — stay on the specific fix.
Deliverable: remediated pages re-published; tracked pages updated with dates.
Ardent: the underperformer's new opening added the price context every cited competitor included; the Brand Manager approved within 18 hours; within ten days it earned a recommendation in ChatGPT.
Week 11 — Competitive displacement & stretch goals
Agency: in Sources, filter for questions where a competitor is cited, sort by frequency; build displacement briefs for the 3–5 highest-value questions where the brand has a credible authority claim and the cited page has structural weaknesses; begin production.
Brand: review displacement briefs with particular attention to factual depth — displacement pieces must be demonstrably more comprehensive than the competitor page, and the brand's subject-matter knowledge is the advantage; approve within 48 hours.
LBOX workflow: share the Sources competitor analysis during the week-eleven review — which pages are cited, at what frequency, on which engines — then open each cited page and walk through its structural weaknesses. This makes the opportunity concrete and builds brand-side investment.
Deliverable: displacement brief package (3–5); competitor recommendation-share analysis from Sources.
Ardent: Sources showed a competitor cited across all engines for a query answered by a thin 900-word overview; seeing it directly, the Content Lead said "we have a much better answer to this — let's build it," and approved the brief the same day.
Week 12 — 90-day report & Q2 roadmap
Agency: generate the full LBOX export (SoA day 0 vs. 90, Share of Voice, recommendation events by engine, competitor share change, sentiment trend); build the 90-day report and Q2 roadmap; present with the data visible as the primary source.
Brand: VP reviews the report and presents to internal leadership; approve or adjust the Q2 roadmap (question targets, production volume, SoA projections); confirm the engagement for Q2 or adjust scope.
The Q2 conversation: the 90-day report is the retention moment. Let the data carry the argument — an SoA that moved from 4% to 29% in 90 days, shown directly in the platform, is more persuasive than any narrative. Build the Q2 roadmap from Topics and Query Fanout, showing exactly which opportunities remain and the projected trajectory if the work continues.
Deliverable: 90-day performance report; Q2 roadmap with targets and projections; engagement continuation confirmed or adjusted.
Ardent: by day 90 SoA reached 29%, up from 4%; 8 recommendation events across 28 questions; Sources showed Competitor A had lost 2. The VP approved a Q2 expansion (12 new questions per quarter, a 55%+ SoA target by month nine), attributing the decision directly to seeing the SoA trend line during the review.
The complete partnership deliverable map
| Week | Phase | Deliverable | Agency owner | Brand owner | LBOX surface |
|---|---|---|---|---|---|
| 1 | Foundation | Shared subject + baseline audit | Account Manager | VP Marketing (view access) | AI Visibility scan, Answers, Sources |
| 2 | Foundation | Approved Question Priority Matrix | Strategist | Content + SEO Lead | Topics, Query Fanout, Prompt demand |
| 3 | Foundation | Signed-off brief package | Strategist | Brand Manager + Content Lead | Sources, Query Fanout |
| 4 | Foundation | Phase 1 review + production plan | Account Manager | VP Marketing (sign-off) | 30-day Report |
| 5 | Execution | Reviewed first-wave drafts | Writers + AM | Content Lead (48h SLA) | Sources |
| 6 | Execution | Published pages + indexing | AM / SEO | Brand Manager (final approval) | Tracked pages |
| 7 | Execution | AEO Audit + schema report | SEO / Strategist | SEO Lead (CMS access) | AEO Audit |
| 8 | Execution | First SoA report + wave-2 briefs | Account Manager | VP Marketing (biweekly review) | Answers & Citations |
| 9 | Optimization | Full SoA audit + gap analysis | Strategist + AM | Content Lead (48h review) | Answers, Sources, Topics, Sentiment |
| 10 | Optimization | Remediated pages re-published | Writers + AM | Brand Manager (24h SLA) | Sources, AEO Audit |
| 11 | Optimization | Displacement brief package | Strategist | Content Lead (48h review) | Sources, Topics |
| 12 | Optimization | 90-day report + Q2 roadmap | AM + Strategist | VP Marketing (sign-off) | Full Live View export |
Shared monitoring cadence
| Cadence | Agency action | Brand action | LBOX surface |
|---|---|---|---|
| Weekly | Check Answers & Citations, SoA, Share of Voice | Review weekly report email | Answers & Citations, Share of Answer |
| Biweekly | Working review with the brand | Attend; review live data | Full Live View |
| Monthly | Generate and share white-labeled report | VP reviews independently | 30-day report |
| Day 90 | Full export for the performance report | Attend 90-day review; approve Q2 | Full Live View export |
The six rules of a high-functioning partnership
1 · Share the LBOX workspace on day one. Not a slide deck — the actual subject, with brand view access before the kickoff ends. Shared data means faster decisions, fewer disputes, better results. This is the operating foundation.
2 · Confirm SLAs in writing before Phase 2. 48 hours for content review, 24 for remediation — the minimum to keep 90 days on track. Written, not verbal. When a review runs late, the AM references the written SLA.
3 · Separate brand review from structural review. Brand checks facts and voice; agency owns the opening answer, H2/H3, evidence, and schema. When a reviewer crosses into structure, the AM opens Sources and shows the cited competitor page. That conversation ends quickly.
4 · Get all stakeholders into the brief, not the publication. The most expensive mistake is a new stakeholder appearing at publication with structural changes. In week four the AM asks who else will see the content; every named stakeholder joins the brief review in week three. One conversation at the brief stage is worth ten at publication.
5 · Read the answer text, not just the score. SoA tells you whether you're winning; the actual answer text in Answers & Citations tells you why. Both sides read it at every biweekly review — it reveals which data point triggered a recommendation and what a page is missing.
6 · Build remediation into the plan, not the post-mortem. Some pages won't earn a recommendation first time — expected. Answers & Citations gives the diagnosis; Sources gives the fix. Build remediation into Phase 3 from week four and treat it as a normal step.
The partnership advantage: brands running AEO in-house move faster on approvals but slower on strategy; agencies faster on strategy but slower on brand context. A well-structured partnership eliminates both bottlenecks. When both sides hold their lane and share the same data, the 90-day model performs at its ceiling.
Phase exit checklists — dual sign-off
Neither side advances with open items. These are the go/no-go gates.
Phase 1 exit (Day 30). Agency confirms: subject configured (brand domain, competitors, 20–30 questions); brand view access granted and working; baseline SoA saved as the day-zero benchmark; Question Priority Matrix finalized and brand-approved; brief package approved by Brand Manager + Content Lead (min. 5); Phase 2 plan confirmed. Brand confirms: content inventory provided; competitors confirmed; CMS access granted; matrix reviewed and signed off; briefs reviewed and approved; 48h/24h SLAs confirmed in writing.
Phase 2 exit (Day 60). Agency confirms: all Quick Wins published and tracked; AEO Audit complete, crawlability resolved within 72 hours; schema on all pages; first SoA movement report delivered; Core Target briefs approved and in production; biweekly check-in complete with SoA shared via LBOX. Brand confirms: all drafts reviewed within the 48h SLA; publication approvals provided without delay; CMS access maintained; biweekly reviews attended with the data read directly.
Phase 3 exit (Day 90). Agency confirms: full SoA audit complete across all questions and engines; underperformers diagnosed, remediated, re-published; displacement package delivered; 90-day report finalized with LBOX data as primary source; Q2 roadmap delivered; subject updated with the Q2 set. Brand confirms: Gap Analysis reviewed within 48 hours; remediation approved within the 24h SLA; displacement briefs reviewed and approved; 90-day report presented to leadership; Q2 roadmap approved and continuation confirmed or adjusted.
Red flags — escalate to VP Marketing (brand) and agency leadership when
- Zero recommendations at week 8 — read the actual answer text before concluding; if unclear after diagnosis, call an emergency strategy review with both sides present.
- Brand review delays exceeding 72 hours — late reviews compress Phase 3; escalate and reset the SLA in writing.
- Structural changes to published content requested post-publication — teach with Answers & Citations and Sources before approving.
- AEO Audit failures unresolved within 72 hours — a technical blocker; escalate to the agency SEO lead and the brand's technical contact simultaneously.
- New stakeholders requesting changes at the publication stage — get them into the brief process immediately and document the delay and its cause.
Final note
The partnership model is the most powerful version of the 90-day framework, and the most fragile — more moving parts, more stakeholders, more places for friction. The structure here manages that complexity.
The brands that get the most from an agency engagement are not the ones that hand over control and wait — they stay in the workspace, hold their review SLAs, protect the structural elements the agency designed, and show up to every biweekly review ready to decide. The agencies that retain brand clients past 90 days are not the ones with the most polished decks — they share the data directly, explain every decision with platform evidence, and treat the brand as a partner in the work, not an audience for a presentation.
Align your team and your agency on this model before day one. Configure shared LBOX access in week one. Run Phase 1 together before any content is produced. The 90 days go faster than they look, and the decisions made in the first two weeks determine whether the final two produce a Q2 roadmap or a post-mortem.