Existing work: open, inspect, judge
Direct URLs · status before criticismClient reactions below are predicted usability questions, not statements Péter actually made.
Proposal application
STAGING · protectedOpen proposal applicationhttps://kertujitok-product-stage-20260923.clientsflowdigital.workers.dev/panel/proposals/What was produced
Eight accepted scenarios, 15 downloaded documents and 26 rendered pages are recorded. Grid-to-PDF highlighting, settlements, payments and immutable history exist; the staged grid is currently read-only.
Why it is not enough yet
The demonstrated prototype allowed in-place cell editing and immediate affected-PDF updates. The new grid is read-only and sends editing to a separate Sheet, losing part of the demonstrated experience. The new application is also separate from the live panel. Sheets editing still needs safe row/formula guidance. First download also locks the original estimate, but the visible button sounds like an ordinary download.
A bemutatón itt át tudtam írni a cellát és láttam a PDF változását. Most miért kell másik ablakba mennem?Predicted client reaction · illustrative wording
Evidence: product/onboarding-evidence.md; product/pdf-tests/eight-pass-acceptance.json; cloudflare/proposals/app.js:74 · Finalizer: D1, D2, D3
Calendar and CRM
STAGING · protectedOpen calendar and crmhttps://calendar-stage-20260923.cfd-staging.com/panel/What was produced
23 API checks and 11 hosted journeys are recorded, with additional conflict/error/retry and mobile evidence. The calendar reuses CRM dates.
Why it is not enough yet
The calendar-only stage does not contain the proposal module. Shared-menu alignment remains unfinished. A search observation needs reproduction before calling it a defect. No integrated live-panel acceptance exists.
Ugyanonnan elérem az ajánlatokat is? Ha rákeresek egy ügyfélre, biztosan csak őt látom?Predicted client reaction · illustrative wording
Evidence: calendar/workers/visual-final.md; hosted-api-results.json; hosted-browser-results.json · Finalizer: D1, D4
Local maintenance landing
LIVE · publicOpen local maintenance landinghttps://kertujitok.hu/helyi-kertfenntartasWhat was produced
A focused mowing/leaf-collection page is published for Szigetszentmiklós and its 10 km area. The dropdown correction has a live receipt.
Why it is not enough yet
A working visible page does not prove enquiries are stored and measured. The complete production form-to-analytics chain and final route-wide production QA are still unproved.
Ha innen valaki ajánlatot kér, hol jelenik meg, és biztosan tudjuk, hogy ebből az oldalból jött?Predicted client reaction · illustrative wording
Evidence: product/public-production-release.json; product/qa/dropdown-production.json; marketing/production-test-resume.md · Finalizer: D5, D6
30-day report pack
COMPLETE · historicalOpen 30-day report packhttps://review.clientsflow.hu/kertujitok-30-napos-riport-2026-09-23-v1/What was produced
Four linked report sections, narration, native slides and a private Notion page are delivered. Its frozen window is 24 Aug–22 Sep.
Why it is not enough yet
It explains past results, not the live launch state. A client could mistake five mixed platform interactions for five real leads if the handoff collapses the distinction. Do not rewrite the frozen numbers to make current delivery look complete.
Rendben, de ebből mi működik most, és hány valódi érdeklődő érkezett?Predicted client reaction · illustrative wording
Evidence: marketing/workers/reporting/pack_receipt.json · Finalizer: D6, D7
Client onboarding guide
DRAFT · planned journeysOpen client onboarding guidehttps://review.clientsflow.hu/kertujitok-client-onboarding-2026-09-23-v1/What was produced
The initial guide was created before execution and includes a sanitized copy of the original frontend plus planned journeys.
Why it is not enough yet
Mockups and a historical prototype are still the teaching material. Final protected links, actual released screenshots and exact controls are missing. It is useful as a design brief, insufficient as a client manual.
Ezt a gombot nálam hol találom? A képen máshogy néz ki.Predicted client reaction · illustrative wording
Evidence: onboarding/index.html; product/onboarding-evidence.md · Finalizer: D7
Demonstrated highlight prototype
REFERENCE · not current appOpen demonstrated highlight prototypehttps://review.clientsflow.hu/kertujitok-proposal-highlight-2026-09-09-v1/What was produced
This is the exact September 9 reference required by Matt. It establishes the grid and PDF highlight interaction to preserve.
Why it is not enough yet
Its public availability makes it easy to confuse with the new persistent application. It must be labelled a demonstration, never presented as the final place to manage projects.
Ebbe írjam az új ajánlatot, vagy a másik felületbe?Predicted client reaction · illustrative wording
Evidence: original source: review-variants/highlight/site/ · Finalizer: D2, D7
Additional completed artifacts and old verification boards
Open the native Google Slideshttps://docs.google.com/presentation/d/1o0dtXeev3d5MGcpMPKLfJAGYjrMAWMiwPNcVB5A1cJQ
Open the private Notion reporthttps://app.notion.com/p/Kert-j-t-k-30-napos-PPC-SEO-s-CRO-riport-kertujitok-2026-08-24-2026-09-22-3e49cef986ad81c99085d5d316d6cd80
Open the product verification boardhttps://review.clientsflow.hu/kertujitok-product-verification-2026-09-23-v1/Historical progress surface. A green marker is not fresh production proof.
Open the calendar verification boardhttps://review.clientsflow.hu/kertujitok-calendar-verification-2026-09-23-v1/Historical progress surface. A green marker is not fresh production proof.
Open the marketing verification boardhttps://review.clientsflow.hu/kertujitok-marketing-verification-2026-09-23-v1/Historical progress surface. A green marker is not fresh production proof.
KPC decision evidence: 251 advice rows imported with zero rejects and zero cited-figure mismatches in the saved receipt. The raw Stage 3 dashboard contains private account/query evidence and was deliberately not hosted. /Users/agency/Documents/Agty/KERTÚJÍTÓK/.tmp/overnight-20260923/marketing/workers/kpc/ads-control/reports/stage-3-20260923T231819-68beff25.html
Google Ads execution lessons: the exact-session research is complete and belongs in the finalizer input, rather than a new research task. /Users/agency/Documents/Agty/KERTÚJÍTÓK/.tmp/overnight-20260923/marketing/workers/knowhow/report.md
Usability critique: what to fix and why
P1 = affects trust, task completion or promised delivery. P2 = important clarity or polish. Evidence gaps are labelled separately from observed defects.
P1 Source-confirmed prototype regressionThe demonstrated table can no longer be edited
Evidence: The September 9 prototype creates input/textarea controls in table cells and regenerates the PDF on edits. The staged renderSheet builds text-only cells and points to an external Sheet for editing. Highlighting remains, direct editing does not.
Client consequence: The client saw an edit-and-see workflow. A read-only replacement is a loss of demonstrated behavior even if the underlying accounting improved. The earlier read-only choice was an implementation decision, not a user requirement.
Required improvement: Restore in-place editing backed by the same authoritative Sheet. Show save/error/conflict states and verify edited cells, Sheet readback and the affected regenerated PDF together. D2
P1 Confirmed gapOne client task spans separate builds
Evidence: The proposal and calendar proofs belong to isolated stages. Neither is evidence that the current production panel supports the complete journey.
Client consequence: A collection of working modules still leaves the client without the promised single entry point.
Required improvement: Release the combined panel and prove Blog → Ajánlatok → Naptár → CRM with one login, including mobile. D1
P1 Observed workflow riskRoutine row editing exposes internals
Evidence: The written workflow asks users to insert full rows, copy E/H formulas and avoid the hidden Q identifier. Duplicate IDs reject import, so silent corruption is not being alleged.
Client consequence: Péter must remember spreadsheet mechanics unrelated to quoting a job. A harmless copied row can become a support request.
Required improvement: Preserve Sheets but make normal row edits safe, validate import errors in plain Hungarian and show the exact repair. D2
P1 Source-confirmed usability gapDownload also establishes the original
Evidence: The button reads “PDF mentése és letöltése”. Its handler locks the original estimate before generating/storing the first file. The visible primary action does not explain that state transition.
Client consequence: A client may believe they are testing a download while committing the baseline against which later work is compared.
Required improvement: Make draft/first issue state clear before the action. Explain recovery when save/download fails after the original has been locked. D2
P2 Observed visual frictionThe spreadsheet dominates first use
Evidence: The staged screenshot carries the original image/header area and a large grid beside a scaled PDF. Five financial totals and several actions compete above it.
Client consequence: A first-time user has to infer whether to edit here, open Sheets, refresh, or download. Familiarity with the original workbook helps, but hierarchy can improve.
Required improvement: Keep the demonstrated grid and highlight. Make the next action, source freshness and edit location obvious; ensure preview can be read at useful size. D2
P2 Recorded defect / unconfirmed behaviorCalendar finish is inconsistent
Evidence: Shared menu entries lack neighboring icons/inset. Separately, an earlier Kanban query left all records visible; that behavior is an observation pending reproduction.
Client consequence: Uneven navigation makes additions feel bolted on. If search truly fails, finding a specific client becomes unreliable.
Required improvement: Align menu styles. Reproduce search with known matches/no-matches, then fix only the confirmed behavior. D4
P1 Verification gapTracking has no complete live proof
Evidence: The saved-lead runtime and GTM20 are published, but the genuine production submission, D1 row and same-ID GA4 diagnostic event have not been witnessed end to end.
Client consequence: The client cannot rely on reported lead counts or campaign comparisons yet.
Required improvement: Prove the real chain with safe synthetic data and notification suppression; keep Ads attribution distinct. D5
P1 Provider-dependent gapConfigured campaigns are not serving proof
Evidence: At 23 Sep 21:53 UTC, five campaigns were enabled at 5700 Ft/day; maintenance at 800 Ft/day was paused. Billing/destination review and actual impressions remained unresolved.
Client consequence: The client asked for ads to run. Enabled settings and a live landing page alone do not deliver that outcome.
Required improvement: Refresh eligibility/billing/approval and current impressions. Complete independent work; return the exact provider blocker if one remains. D6
P1 Budget-control gapThe monthly limit needs a real control
Evidence: 6500 × 30.4 = 197600 Ft is a planning estimate. The historical account configuration had no hard account budget cap. Current spend and all enabled campaigns must be included.
Client consequence: An estimate presented as a guarantee could violate the agreed 200000 Ft monthly ceiling.
Required improvement: Document and verify an all-account stop/adjust control using actual spend-to-date and remaining allowance. D6
P1 Observed documentation gapThe guide cannot yet teach the shipped app
Evidence: It openly labels itself a first planned version. Its screenshots are mockups and the earlier prototype; real final controls and access steps remain absent.
Client consequence: A polished guide can still send the client down the wrong click path.
Required improvement: Replace placeholders with the final live interface, test the guide with a first-time reviewer and show exact limitations. D7
P1 Freshly confirmed content gapPublished business details need completion
Evidence: The current public /aszf and /adatkezelesi-tajekoztato responses both contain [kitöltendő] for registered office, tax number and registration number.
Client consequence: Placeholder legal/business fields undermine trust and signal unfinished delivery. This is a content completeness observation, not a legal compliance opinion.
Required improvement: Recover verified existing business facts and replace only supported fields. Escalate only any irreducible missing fact. D5
P2 Independent usability judgmentThe report needs a handoff bridge
Evidence: The report opens four equally weighted sections: results, competitors, next period and decision situations. Each is useful, but none answers what is available to Péter today.
Client consequence: The client must translate a historical report into a current operational decision.
Required improvement: Keep the report intact and add a concise current-status bridge in the handoff, with one recommended reading order. Do not reopen the already agreed six-service choice. D7
P2 Independent usability judgmentDaily calendar actions compete
Evidence: Recorded upcoming-view screenshots show four actions repeated on reminder cards: open lead, done, reschedule and clear date.
Client consequence: A busy operator has to scan the same choices for every reminder. This is friction, not proof the actions fail.
Required improvement: Improve the primary/secondary action hierarchy if the first-time walkthrough confirms hesitation. Preserve every action and avoid a speculative calendar redesign. D4
End state
Péter can open the existing panel, reach Blog, Ajánlatok and Naptár from the hamburger menu, edit proposal inputs once through the demonstrated editable grid backed by Google Sheets, understand and download accurate immutable PDFs, track completed work separately from requested and paid amounts, and manage dated CRM reminders without duplicate entry. The live public forms have proved saved-lead measurement. Google campaigns cover all six agreed services within the combined monthly budget, with actual serving proved or its precise external blocker honestly recorded. One final Hungarian guide teaches the actual released journeys using real screenshots and direct links. Technical release and Péter's own acceptance remain distinct.
Confirmed requirementProposed methodUnproved result / open fact
End goal system
flowchart TD
S["<b>Google Sheets</b><br/>One input source"]:::confirmed --> P["<b>Existing Client Panel</b><br/>Proposals and calendar<br/>CRM and blog"]:::confirmed
P --> Q["<b>Proposal documents</b><br/>Highlight and history"]:::confirmed
P --> C["<b>CRM reminders</b><br/>One record, date-only workflow"]:::confirmed
W["<b>Public enquiries</b><br/>Real saved leads"]:::confirmed --> M["<b>Measured acquisition</b><br/>Live chain still unproved"]:::open
A["<b>Six Google services</b><br/>200000 Ft/month"]:::confirmed --> W
B["<b>Billing and approval</b><br/>Eligibility unresolved"]:::open --> A
P --> G["<b>Client guide</b><br/>Actual released journeys"]:::confirmed
classDef confirmed fill:#E8F5E9,stroke:#2E7D32,color:#173B25
classDef assumption fill:#FCE4EC,stroke:#C2185B,color:#5B1430
classDef open fill:#FFFFFF,stroke:#94A3B8,color:#475569,stroke-dasharray:5 5
Route to the end state
Scroll this diagram horizontally to read the full route at a comfortable size.
flowchart LR
P["<b>Separate stages</b><br/>Proposal and calendar"]:::confirmed --> R["<b>Review contract</b><br/>Then explicit Launch"]:::confirmed
M["<b>Partial live marketing</b><br/>Serving unproved"]:::open --> R
R --> F["<b>Finalize and combine</b><br/>Preserve working parts"]:::assumption
F --> T{"<b>Evidence passes?</b><br/>Final-build acceptance"}:::open
T -- No --> F
T -- Yes --> L["<b>Safe live release</b><br/>Rollback and live checks"]:::open
L --> H["<b>Client-ready handoff</b><br/>Guide and truthful status"]:::open
L -- External blocker --> E["<b>Partial delivery</b><br/>Owner and exact resume point"]:::open
classDef confirmed fill:#E8F5E9,stroke:#2E7D32,color:#173B25
classDef assumption fill:#FCE4EC,stroke:#C2185B,color:#5B1430
classDef open fill:#FFFFFF,stroke:#94A3B8,color:#475569,stroke-dasharray:5 5The finalizer may progress product and marketing in parallel. One owner serializes shared production release. External Ads billing must not stall independent product work.
