# 02 — LAKA SEO Strategy Guide ## 1. Strategy thesis The best SEO strategy is not “produce more content.” It is to construct a measurable system that repeatedly does five things: 1. identifies valuable unresolved demand; 2. assigns each meaningful task to one intended canonical asset; 3. creates a superior information product for that task; 4. makes the asset discoverable, interpretable, credible, and connected; 5. learns from business outcomes and reallocates effort. The operating loop is: ```text OBSERVE → MODEL → GENERATE → FILTER → BUILD → CONNECT → PROMOTE → MEASURE → LEARN → ESCALATE OR STOP ``` --- ## 2. North-star model ```text INCREMENTAL ORGANIC CONTRIBUTION MARGIN = incremental qualified organic conversions × average contribution margin − attributable SEO operating cost ``` Rankings, impressions, clicks, links, and indexed pages are diagnostic states. They are not the final business result. Use a small executive scorecard: 1. Incremental organic contribution margin 2. Qualified organic conversions 3. Value-weighted non-brand clicks 4. Value-weighted target-cluster visibility 5. Eligible canonical coverage Everything else belongs to diagnostic dashboards. --- ## 3. Stage 0 — Define the business grammar Before keyword research, define: ```text WHO receives value? WHAT outcome matters? WHICH offer converts that value? WHAT counts as qualified? WHAT is one conversion worth in contribution margin? WHAT capacity limits growth? WHAT outcomes must not worsen? ``` ### Business object card ```text Business: Offer: Audience: Geography: Average contribution margin: Qualified conversion event: Capacity per month: Primary constraint: Primary guardrail: ``` ### IF / ELSE rules ```text IF value_event is undefined THEN do not use traffic as the north-star metric. IF contribution_margin is unknown THEN use a provisional value band AND mark all ROI outputs as forecasts. IF fulfillment_capacity is constrained THEN prioritize query clusters with higher lead quality OR higher margin rather than maximum traffic. IF several offers serve different audiences THEN create separate value models ELSE aggregate reporting will hide quality differences. ``` --- ## 4. Stage 1 — Establish measurement and eligibility ### Required measurement chain ```text QUERY → IMPRESSION → CLICK → LANDING PAGE → TASK EVENT → NEXT ACTION → QUALIFIED LEAD → SALE → MARGIN ``` ### Required technical chain ```text DISCOVERY → CRAWL → RENDER → CANONICAL SELECTION → INDEX → RETRIEVAL → SEARCH APPEARANCE ``` ### Baseline outputs - Search Console property and data access - analytics with defined organic conversion events - CRM or commerce value linkage - canonical URL inventory - crawl/index state inventory - query-to-URL table - branded/non-branded classification - country/device/search-appearance dimensions - baseline date and change log ### Gate ```text IF priority_pages cannot be measured OR conversion events cannot be trusted THEN measurement repair outranks content production. ``` --- ## 5. Stage 2 — Build the audience–task map Do not begin with a keyword tool. Begin with the user’s world. ### Audience dimensions ```text role industry organization size experience problem awareness solution awareness urgency budget risk tolerance geography language device/context accessibility needs ``` ### Task families ```text LEARN DEFINE DISCOVER DIAGNOSE COMPARE CALCULATE LOCATE VERIFY PLAN IMPLEMENT TROUBLESHOOT BUY USE MAINTAIN RENEW REFER ``` ### Journey states ```text UNAWARE PROBLEM_AWARE SOLUTION_AWARE OPTION_EVALUATION DECISION IMPLEMENTATION USE RETENTION ADVOCACY ``` The funnel is a graph, not a one-way pipe. Users can enter, leave, return, skip stages, or move backward. ### Output Create an audience–task matrix: | Audience | Condition | Task | Intent | Journey state | Value event | |---|---|---|---|---|---| ### Gate ```text IF a query cannot be associated with a plausible audience and task THEN classify it as uncertain AND do not publish from it yet. ``` --- ## 6. Stage 3 — Build the semantic demand graph ### 6.1 Start with concept families For every offer, map: ```text PARENT CONCEPTS SUBTYPES SYNONYMS / PARAPHRASES ENTITIES ATTRIBUTES ATTRIBUTE VALUES PROCESSES PREREQUISITES PROBLEMS CAUSES CONSEQUENCES SOLUTIONS ALTERNATIVES COMPARISONS OBJECTIONS RISKS LOCATIONS TIME STATES EVIDENCE TYPES QUESTIONS ``` ### 6.2 Add observed demand Sources may include: - Search Console queries; - customer calls and emails; - site search; - sales objections; - support tickets; - search suggestions and result patterns; - competitor visibility; - forums and communities; - first-party surveys; - paid-search terms; - analytics and CRM outcomes. ### 6.3 Separate semantic similarity from intent equivalence Two phrases can be semantically close but require different pages. Two phrases can look different but belong to one task. ```text "web design Calgary" "Calgary website designer" ``` Likely one service intent. ```text "web design Calgary" "how much does web design cost in Calgary" ``` Related topic, different task and likely different document role. ### 6.4 Cluster test ```text SAME PAGE IF: same sense AND same dominant task AND same answer structure AND same expected result class AND compatible next action SPLIT IF: different task OR different decision stage OR different local/product eligibility OR different format is necessary OR the combined answer would become confusing ``` ### Output A canonical intent registry: | Cluster ID | Representative query | Task | Intent | Variants | Exclusions | Intended URL | Status | |---|---|---|---|---|---|---|---| --- ## 7. Stage 4 — Quantify opportunities ### Opportunity scoring model ```text OPPORTUNITY SCORE = ( demand evidence × business fit × task-value fit × information advantage × conversion value × attainability × confidence × reuse potential ) ÷ ( effort × risk × maintenance burden × time to learning ) ``` Use 0–5 values or calibrated probabilities. Do not pretend the result is exact; it is a transparent prioritization model. ### Existing-opportunity multiplier Add a multiplier when an existing page already has: - impressions; - positions near useful visibility; - links; - conversions; - brand recognition; - stable indexing; - strong internal graph position. ```text IF existing_asset_has_signal THEN prefer improve/merge/reposition BEFORE creating a new asset, unless intent ownership is fundamentally wrong. ``` ### Priority bands ```text P0 — eligibility or measurement failure blocking value P1 — existing high-value opportunity with low/moderate effort P2 — core commercial or decision cluster P3 — supporting informational or authority asset P4 — experimental or speculative demand P5 — reject, defer, or monitor ``` --- ## 8. Stage 5 — Choose the information product A page should be designed as a product that completes a task. ### Information-product selector | Task | Preferred product forms | |---|---| | Define | concise definition, examples, boundaries | | Learn | guide, lesson, demonstration, video | | Compare | table, matrix, benchmark, decision tree | | Calculate | calculator, estimator, formula, worksheet | | Diagnose | checklist, test, decision tree, audit | | Locate | verified directory, map, location page | | Plan | roadmap, template, calendar, framework | | Implement | SOP, code, configuration, download | | Verify | evidence review, source map, methodology | | Choose | scored criteria, trade-offs, recommendations | | Buy | service/product page, proof, scope, process, action | | Maintain | monitoring guide, alert, checklist, dashboard | ### Information advantage Approve at least one defensible advantage: ```text firsthand experience OR proprietary data OR original test OR better synthesis OR better visualization OR better interaction OR more current verified information OR stronger local evidence OR clearer decision support OR useful tool/template OR accessibility advantage OR faster task completion ``` ### Gate ```text IF the proposed page merely restates common information AND has no meaningful experience, evidence, utility, or synthesis advantage THEN redesign the asset OR reject it. ``` --- ## 9. Stage 6 — Write the page specification Every page brief must define: ```text Cluster ID Audience Task Intent Journey state Canonical URL Page type Expected result formats Direct answer Required entities and relations Required attributes Evidence plan Media plan Internal links in Internal links out Next action Primary metric Guardrails Refresh trigger Owner ``` ### On-page logic ```text IF title accurately represents the task AND distinguishes the result AND creates a truthful reason to click THEN test it against current CTR expectations. IF title change improves clicks BUT lowers qualified conversion rate materially THEN revert or refine. IF page has impressions but poor rank THEN diagnose intent, format, information advantage, authority, and internal graph BEFORE repeating keywords. IF page ranks but does not convert THEN inspect audience quality, answer-to-offer continuity, and next action. IF page converts but lacks impressions THEN prioritize discoverability, internal links, authority, and search appearance. ``` --- ## 10. Stage 7 — Design the internal information graph ### Four edge classes 1. **Understanding edges** — definition, prerequisite, evidence, example 2. **Choice edges** — comparison, alternative, trade-off 3. **Journey edges** — next step, assessment, service, product 4. **Authority edges** — supporting research to core asset ### Hub rule Create a hub when it helps a user choose a path across a coherent set of tasks. Do not create a hub solely to insert links. ### Supporting-asset rule ```text IF an adjacent task has independent demand AND can provide a complete useful answer AND supports a valuable canonical asset THEN create supporting_asset AND link contextually toward the relevant core asset. ELSE include the answer as a section OR do not create it. ``` ### Orphan rule ```text IF page_is_priority AND no crawlable internal link exists THEN add at least one contextually useful link OR remove the page from the priority set. ``` --- ## 11. Stage 8 — Build external evidence and prominence Authority is not only “link quantity.” Treat it as independent evidence and graph prominence. ### Authority program components ```text LINKABLE_REASON + RELEVANT_PROSPECT + EVIDENCE_ASSET + RELATIONSHIP / OUTREACH + PLACEMENT + RETENTION + BUSINESS IMPACT ``` ### Linkable reasons - original data; - benchmark; - free tool; - template; - definitive reference; - newsworthy change; - expert collaboration; - unique local resource; - visual explanation; - case study; - correction or replacement for a broken resource. ### Rules ```text IF a page is purely commercial AND difficult to cite THEN create a genuinely useful evidence asset OR supporting resource that can earn attention without disguising its purpose. IF a prospect is irrelevant OR placement is paid/sponsored without appropriate treatment OR outreach claim is misleading THEN reject the opportunity. IF acquired link disappears THEN classify cause AND recover only when the relationship and value remain valid. ``` Measure qualified referring domains, target-page lift, branded demand, assisted conversion, and retained placements—not tool scores alone. --- ## 12. Stage 9 — Technical eligibility and search appearance ### Technical priority order ```text 1. Accessibility to users and crawlers 2. Correct status and destination 3. Rendered main content 4. Canonical ownership 5. Index eligibility 6. Internal discovery 7. Duplicate/facet control 8. Sitemap freshness 9. Mobile and real-user performance 10. Structured data eligibility 11. International/local correctness ``` ### Technical decision rules ```text IF URL should appear in Search THEN it must be accessible AND return an appropriate successful response AND expose indexable content AND not carry conflicting exclusion signals. IF page must disappear for users and search THEN use removal or redirect logic appropriate to the business case, not robots.txt alone. IF duplicate URLs must remain accessible THEN use coherent canonical signals AND consistent internal links AND sitemap preference. IF content is generated by JavaScript THEN validate rendered content and crawlable links. IF structured data type is not relevant to visible page content THEN do not add it. IF Core Web Vitals are poor THEN prioritize fixes by real-user exposure and business impact rather than chasing a perfect laboratory score. ``` --- ## 13. Stage 10 — Match the next action to the journey ### Transition grammar ```text UNAWARE → understand problem PROBLEM_AWARE → explore solution class SOLUTION_AWARE → compare approaches OPTION_EVALUATION → inspect proof, scope, cost DECISION → transact or contact USE → implement successfully RETENTION → renew, expand, repeat ADVOCACY → review, refer, cite ``` ### Nudge rule ```text NEXT_ACTION must be: logically adjacent AND proportionate to trust AND useful without coercion AND measurable ``` Examples: ```text guide → related guide OR diagnostic comparison → calculator OR assessment calculator → saved result OR consultation service page → qualification form customer guide → support action OR upgrade case study → relevant service page ``` --- ## 14. Stage 11 — Experiment with LAKA ### Experiment ladder ```text BASELINE: observe and diagnose MINOR: change presentation or a single local variable MAJOR: change the asset’s answer, evidence, or format STRUCTURAL: change cluster ownership, templates, graph, or data flow PARADIGM: change the solution class or create a new information product ``` ### Escalation logic ```text IF baseline data is unreliable THEN do not escalate; repair measurement. IF a minor test fails AND diagnosis indicates a content/format problem THEN move to Major. IF several pages share the same failure THEN move to Structural. IF competitors all solve the wrong problem OR no existing result class can complete the task THEN consider Paradigm. IF expected incremental value < expected cost THEN stop, defer, or monitor. ``` --- ## 15. Stage 12 — Continuous allocation At a fixed review cadence: 1. update demand and value data; 2. detect query, URL, and conversion shifts; 3. recalculate opportunity scores; 4. separate noise from sustained change; 5. continue, expand, revise, reverse, or retire; 6. record what the system learned. ### Portfolio allocation A balanced portfolio may include: ```text 40% existing-page and conversion improvements 25% core commercial/decision assets 15% supporting authority assets 10% technical/structural work 10% paradigm experiments ``` This is a starting heuristic, not a universal rule. Reallocate according to measured marginal value, constraints, and learning velocity. --- ## 16. Failure diagnosis tree ```text NO IMPRESSIONS? → discovery / crawl / index / demand / cluster mismatch IMPRESSIONS BUT LOW VISIBILITY? → intent / format / information value / internal graph / authority VISIBILITY BUT LOW CTR? → title / snippet / result type / brand / mismatch / SERP crowding CLICKS BUT LOW TASK COMPLETION? → weak answer / poor format / slow or confusing experience / wrong query TASK COMPLETION BUT LOW NEXT ACTION? → journey discontinuity / weak offer / wrong CTA / trust gap LEADS BUT LOW QUALIFICATION? → wrong audience / misleading promise / broad intent / form design SALES BUT LOW MARGIN OR RETENTION? → offer economics / fulfillment / customer fit / expectation mismatch METRICS IMPROVE BUT BUSINESS DOES NOT? → proxy optimization; redefine the primary outcome ``` --- ## 17. Strategy completion test The strategy is ready only when every priority cluster answers: ```text Who is the user? What are they trying to do? Why now? What observed demand supports it? Which queries represent the task? Which single asset owns the intent? What form best completes the task? What makes the asset meaningfully better? How will users and crawlers find it? What independent evidence supports it? What is the next logical action? What event creates business value? How will success and failure be distinguished? What is the LAKA escalation path? ```