Fusion · refined goal contract

Verification and Constraints

COMPLETE · 14 / 14 VERIFIED · PARENT ACCEPTED

Plan better campaigns and make the 280,000 Ft monthly service credible through specific choices, price qualification, measurement and clear decisions.

Sol 6.1 High · separate pinned taskRevised 1 October 2026Desktop only

Execution authorised. Matt asked for a fresh separate pinned Sol 6.1 High task after refinement. The old preparation-only Launch gate no longer applies. No campaign outcomes are claimed.

Goal and finish line

End goal: Give Matt a convincing Hungarian sales-call and WhatsApp package to discuss restarting Márti’s PPC management.

Done means: A branded hosted presentation, campaign/measurement/test register, exact 200k and 350k media options, an honest editable model, sample reports and unsent message/call copy.

14 / 14execution criteria verified

4useful deliverables

280k Ftmonthly management, media additional

Review in this order

Scope and why the large application builds were removed.
Campaign choices, qualifying journey and measurement.
Deliverables and their observable checks.
Actual execution evidence and delivered links.

Exact review constraint: “Only do QA for desktop and only optimize for desktop. You must not do smaller screen size QA.” Smaller screens will be untested.

Scope that supports the sales conversation

Keep	Use a representative example	Defer application engineering
Campaigns, audience hypotheses, creative angles, destinations, 200k/350k allocation, tests, measurement and reports.	Lead register, premium proposal, follow-up ownership and one-week organic plan. Márti publishes, agency provides edited images and copy directions.	Full CRM portal, proposal generator and drag-and-drop content-calendar application. Future scope discussion, no silent monthly commitment.

Budget discipline: about 2,000 Ft/day is 60,000 Ft in a 30-day month for a module’s total media spend. The campaign library is a choice of opportunities. The launch plan must concentrate spend and show tests and reserve within the total. Fee plus media is 480,000 or 630,000 Ft before explicit extras.

Price curiosity and weak perceived structure come from Matt’s reported feedback. The package must show a better method without blaming Márti or guaranteeing orders. Source evidence distinguishes confirmed preferences, proposals and unknown facts.

Six opportunities, one affordable first launch

M01 · Pool and diningShared RS Barcelona + Belgian Fusion Tables destination. Price/space fit and showroom enquiry.

M02 · Family collection2 pool + 2 ping pong + 2 foosball + 1 shuffleboard. Browse first, visit without choosing a SKU.

M03 · FoosballFriends, play and useful furniture. Verify the folding SKU. Persona is a creative hypothesis, not wealth targeting.

M04 · Ping pongSport and active hobbies first, dining function extra. Account-verify targeting options.

M05 · Pool enthusiastsChecked skill Reel and product-first alternative, usually creative angles inside M01.

M06 · HINOMISeparate chair journey and economics. Selectable, not automatically funded.

For both media budgets, the executor recommends which campaigns launch, which wait, and the observed result that releases reserve or changes allocation. Holiday examples require confirmed offers and delivery/assembly cutoffs.

What is measured and how it changes decisions

Advert and visit

Campaign/ad IDs, UTMs and click identifiers where available.

Unique enquiry

Successful submit or recorded real contact. Deduplicate by lead ID.

Qualified or pending

Price acknowledgement, product/space fit, timing and reachable contact.

Visit and proposal

Booked vs attended, proposed vs accepted, next action and owner.

Won, lost or unknown

Order contribution and loss reasons, with sales lag and dated cohorts.

Next spend decision

Continue, change, stop or collect more evidence. Inconclusive is valid.

The deliverable specifies event/source, denominator, collection method, owner, cadence and missing-data rule for every metric. Manual lead-register reporting works first. Platform feedback is conditional on verified implementation and volume. A click, cheap lead or platform claim does not prove a sale.

Evidence and client fit

Five exact Codex sessions extracted, validated and read in full. Seven available full meeting texts read, with 61 source-checked quotations. The latest five recovered dates are 15 September, 31 July, 29 July, 14 May and 25 February. This does not prove coverage of the actual last five meetings. July archives revise the earlier Diagonal restriction and establish the PDF-first price baseline and delivery-ownership concern.

Prioritise suitable showroom or consultation opportunities, with genuine, borderline and unsuitable feedback.
Lead seasonal creative with family, interiors, dining and play, with model-specific delivery confirmation.
Check actual mechanisms, source colors and product facts before review. Use atmospheric interiors, the official logo and normal Hungarian capitalization.
Reuse faithful manufacturer content and diagrams, prepare consistent drafts and focus client input.
Show the PPC owner, monthly work, preserved learnings and handover. Use one coordinating contact and concise decision updates.
Allocate by qualified outcomes. Facebook remains an option, with no assumed fixed split or Buda-only buyer rule.
July permission supersedes the old Diagonal exclusion. Verify current stock and offer facts normally.
Start with the PDF price-list baseline and compare maintained guide-price context later. Name who updates prices and judge affordability/qualified outcomes. Sequential visitor counts do not establish causal proof.
Keep HINOMI as a separate two-chair journey with verified names and local showroom value.

Private transcript extracts, opinions/friction analysis and internal margin modelling stay outside this public board. The client deck reflects the relevant preferences respectfully. The source inventory links the private files locally.

Work and proof

Only useful deliverables · criterion evidence must be real

D1 · Choose campaigns and customer journeys

Current problem: Too many product ideas for limited spend.
Exact change: Six selectable modules, a recommended launch subset and two deeper funnels.
Visible in: Campaign menu and detailed campaign appendix.
Representative test: Follow both budget selections to the funded audience, advert, destination and qualified outcome.

3 / 3

Show observable checks and evidence

D1.1 Each of M01–M06 specifies need/product, priority and launch prerequisite, platform and objective, practical targeting settings versus unverified audience suggestions, campaign/ad-set/ad hierarchy, initial creative variants, landing/CTA, qualification, spend basis and primary outcome. The recommended launch subset per budget explains why other modules wait and when they enter.
Six complete selectable modules and justified launch subset. Verified content: View delivered content
D1.2 Two deeper journeys, pool/dining and family collection, show message → representative advert → annotated landing wireframe → price acknowledgement/fit questions → showroom or qualified enquiry → next contact and measurement. M02 has exactly **two pool, two ping pong, two foosball and one shuffleboard** slots. Visitors can request a viewing without choosing a SKU. Placeholders are explicit and brief a specific missing asset.
Two deep journeys and exactly2+2+2+1 collection slots. Verified content: View delivered content
D1.3 Black Friday and Christmas options contain audience hypothesis, copy angles, proposed offer mechanism, budget within the chosen scenario, timing backwards from confirmed cutoffs, and a fallback without unverified discount/arrival claims. M05 includes an interesting pool-skill Reel storyboard and a product-first alternative, without fake product capabilities or incorrect physics. HINOMI is selectable but uses separate economics and landing intent.
Seasonal options, two M05 storyboards and separate HINOMI journey. Verified content: View delivered content

D2 · Measure quality and decide the next spend

Current problem: Platform leads hide affordability and sales progress.
Exact change: Trace unique leads to sales stages, define staged tests and reconcile both budgets.
Visible in: Measurement, test register, budget comparison and editable illustrative model.
Representative test: Trace one sample lead, calculate zero sales and apply win, loss and inconclusive test outcomes.

5 / 5

Show observable checks and evidence

D2.1 The metric dictionary covers landing visits, raw unique enquiries, reachable/contacted, qualified leads, showroom booked/attended, proposals, won/lost and attributed revenue/contribution. Each has definition, denominator, event/source, collection method, owner, update frequency and missing-data rule. Qualified means explicit affordability after a price indication, product/use and space fit, plausible purchase horizon and reachable contact. Pending qualification, duplicates, uncontactable, lost and unknown remain visible. Contact clicks/form starts are diagnostic proxies, not actual enquiries. Show a realistic starting spreadsheet/lead register without requiring a new portal.
Metric dictionary and realistic synthetic lead register. Verified content: View delivered content
D2.2 Measurement specification maps website successful-submit events and IDs/UTMs to lead records, deduplication and sales status. It distinguishes website recording, tag firing, platform acceptance, attribution and optimisation eligibility. CRM/offline feedback to Google/Meta is a conditional post-restart implementation with actual account/objective/volume prerequisites, readback checks and a manual fallback. Do not claim current integration or promise it causes better leads. Unknown/unattributed records and overlapping platform claims have a reconciliation policy. Track cohorts by enquiry date and allow sales lag.
Successful-submit, deduplication, attribution and conditional offline path. Verified content: View delivered content
D2.3 Both 200k and 350k scenarios sum exactly, identify active campaign/modules, initial daily planning equivalents, retargeting eligibility, experiment allocation, reserve release, and reallocation conditions. Include a focused default and one meaningful substitution option, not many equally weighted menus. Explain 30/31-day and platform pacing limits. The 350k option may deepen a winner rather than multiply campaigns. Each allocation names the decision it funds.
Exact200k/350k media plus280k fee, reserves and substitution. Verified content: View delivered content
D2.4 At least four staged comparisons cover price transparency/qualification, creative angle, showroom vs enquiry CTA and landing choice. Each specifies single primary comparison, primary KPI, guardrails, practical effect worth acting on, spend/time/sample planning, sales lag and win/loss/inconclusive actions. At these budgets, distinguish observational optimisation from randomised A/B evidence. Do not split into simultaneous underfunded tests or claim significance after a fixed arbitrary lead count. Include worked numbers showing what changes next: creative variant, destination, sales response or budget. Cost-stop rules are economic decisions, not statistical proof.
Four staged tests with twelve worked decisions, sample/lag limits. Verified content: View delivered content
D2.5 Editable illustrative economics model and a separate private comparison use explicit contribution, FX/date, qualification and closing assumptions. Include media, the 280k fee, extra costs, zero sales and 90-day lag. Separate ROI from ROAS, round break-even orders up, validate zero/invalid inputs and avoid double-counting qualification. State acquisition cost and sales-maturity windows separately: one funded-month cohort includes one month’s costs with sales observed up to 90 days later, while three months of continued acquisition include all three months of media, fees and extras. Never compare three months of results to one month’s costs. No reliable ROI forecast is claimed before evidence exists.
Editable fictional calculator, separate private assumptions and cost/maturity windows. Verified content: View delivered content

D3 · Present a credible monthly working method

Current problem: Restarting must show how the service will work.
Exact change: Owners, creative approvals, 30/60/90 days and consistent decision reports.
Visible in: Main Hungarian deck, weekly/monthly examples, proposal and organic appendix.
Representative test: A reviewer can explain the first week, responsibilities, assets, measures and next-month decisions.

3 / 3

Show observable checks and evidence

D3.1 Main deck follows an accessible sales story: reported problem → proposed method → initial campaigns → qualification and measurement → budget/economics → first 30/60/90 days → responsibilities, creative process and reporting → concrete restart discussion. Keep the fee **280,000 Ft/month**, media additional. Show what monthly work supports that fee, the PPC owner/role to confirm, Matt as coordinator, preserved historical learnings and the first-week handover/baseline checks. Expertise or capacity claims require actual evidence. Scope quantities/cadence are explicitly proposals, not silently agreed obligations. Show a bounded creative process: approved product facts/assets → angle/format brief → copy/image adaptation → agency checks → Márti approval → post-restart publishing → sales feedback → next variation. Design uses existing Fusion off-white/deep-green/sage/restrained-gold references, familiar official logo and normal Hungarian capitalization.
Hungarian main story, monthly work, ownership and first-week handover. Verified content: View delivered content
D3.2 A weekly short pulse and monthly decision report use one internally consistent labelled synthetic dataset. Include campaign/period and spend by platform, enquiries, qualified/borderline/unsuitable/pending/unreachable, qualification rate/CPQL, showroom/proposals/wins/losses, mature vs immature cohort, unknown sales outcomes, tests, decisions, named owners and next dates. Include the 280k fee separately. Demographic or click diagnostics are secondary and source-labelled. Do not fabricate revenue when the fixture has no won order or treat a forecast as a result. Reports explain what should stop, continue or change.
Weekly/monthly reports reconcile one zero-sale synthetic fixture. Verified content: View delivered content
D3.3 Agency versus Márti responsibilities, the PPC handover/continuity step and response-time assumptions are proposed plainly. Provide one sanitised premium proposal example reusing the late-August design source, a lead follow-up worksheet with next-action ownership, and a representative one-week organic plan with edited-image placeholders/copy directions and product-page destinations. Márti manually publishes. Recorded follow-up activity does not establish poor selling or neglect, so frame improvements as a shared process. Portal/calendar are future optional implementation ideas only, without a separate build deliverable or implied inclusion in 280k. Describe addendum scope points as non-binding discussion items, not a signed contract.
Sanitised proposal, next-action worksheet and one organic week. Verified content: View delivered content

D4 · Deliver a shareable persuasion kit

Current problem: Internal planning alone is difficult to present.
Exact change: One public review surface and grounded unsent sales language.
Visible in: Hosted deck, WhatsApp/call kit, private evidence index and this board.
Representative test: Anonymous desktop access, actual interactions and postdeployment section screenshot QA.

3 / 3

Show observable checks and evidence

D4.1 Publish assembled client-facing HTML through Comment HTML Review Host, preserving unrelated mirror pages and existing document identity on revisions. Verify anonymous HTTP access, all delivered links/assets, desktop interaction/calculator behaviour and comment/rewrite persistence in a disposable test document. Complete desktop-only Luna section screenshot QA with actual postdeployment captures, two independent fresh Luna xhigh partitions, Luna Max fixes if needed and fresh QA of the resulting build. Capture every visible section and important expanded state, plus multi-section composition. Do not do smaller-screen checks. No automatic user-facing opening.
Anonymous live access, 14 source-matching assets, calculator checks, persisted disposable comment/rewrite and two fresh Luna xhigh desktop partitions accepted. 95 actual postdeployment captures. Smaller screens untested. View delivered content
D4.2 Unsent Hungarian WhatsApp invitation and after-call follow-up, call opening and concise objection answers address poor affordability/lead quality and structure without blaming the client. They point to the delivered URL, explain the specific changed method and ask for one concrete restart discussion. No guarantees, invented purchase results, unapproved offer or binding commitment.
Unsent invitation, follow-up, call opening and objections. Verified content: View delivered content
D4.3 Final contract cards accurately separate planned, built, checked, published and sent states. Every passed criterion links actual content or decisive evidence. The private source index links every session gist, transcript extraction/validation, preference profile, campaign/test documents, model, copy kit and verification report. Explicitly disclose meeting coverage, unknown facts and desktop-only testing. Parent accepts actual artifacts and proofs, not a child status alone.
Honest states, criterion proofs and private complete source index. Verified content: View delivered content

Execution and recovery

One separate pinned Sol 6.1 High executor owns the kit and its verification. Execution finished. Parent accepted the delivered content, budget reconciliation, live build identity and desktop evidence. Messages remain drafts and campaigns remain inactive.

Authority: reversible output work and non-sensitive public publication are authorised. Client sends, paid campaign activation, binding commitments and spending beyond the explicit limit are separate actions. No new security layer or automatic preview opening.

Client-facing presentation Public model and fixture checks Published execution verification Stable contract

Delivery states and remaining launch inputs

Planned: campaigns and implementation. Built, checked and publicly published: the sales kit and its working examples. Sent: no messages. Activated: no campaigns. Business outcomes: none claimed.

Desktop only, 1440 × 1000, DPR 1. Smaller screens untested by explicit instruction. Seven available full meeting texts and 61 source-checked quotations inform the kit. Latest five recovered dates do not prove actual last-five meeting coverage.

Before campaign activation, confirm current account/tracking state, SKU stock/prices/delivery, tax basis, named PPC specialist/capacity, mature qualified close rate/order contribution and separate chair economics.