# Csupi: melyik Ads- és SEO-munka éri meg most?

Created: 2026-10-06
Source: dated provider evidence in the October 6 monthly report, verified September 29 to October 3 releases, and later source-traceable history synthesis.
Schema: 2
Action plan ID: 5b778a8e-a90a-4f30-903d-cf3c08d7c806
Client: Autószállítás Non-Stop Autómentés Budapest 0-24h
Review URL: https://review.clientsflow.hu/csupi-ads-seo-strategia-2026-10-06-v1/
Decisions doc: csupi-ads-seo-strategia-2026-10-06-v1
Review rule: This tailored review has zero actionable future tasks. Completed work is evidence. Future improvements are optional proposals, not dispatch authorization. Comments and voice discussion do not execute them.

## Objective and settled scope

A useful strategy answers which next deliverable could connect marketing effort to suitable completed work. Google Ads is read-only. No budget increase, targeting move, production deployment or future optimization is executed by this review. No strategy refinement rounds. This is a single grounded review surface with comments and an optional read-only voice explanation. No client emails, private contract fees or raw conversations are published.

## 1. Recommendation

Elsőként a megkereséseket kapcsoljuk össze az elvégzett munkákkal és azok közvetlen költségével. Ezután az októberi, teljes kampányidőszakból készítsünk döntési anyagot. A meglévő szolgáltatásoldalakat kövessük, mielőtt új oldalakra vagy nagyobb hirdetési költésre vállalkozunk.

An alternative is to prioritize visibly weak CPC immediately. That may reduce traffic costs while also reducing valuable enquiries. Outcome evidence is therefore the stronger starting point. A commercial minimum fee, acceptable pickup/destination area and capacity remain business inputs, not inferred marketing settings.

## 2. Evidence, with limits

Google Ads, September 6 to October 5 versus August 7 to September 5, Budapest: spend 123,487 versus 111,546 HUF, clicks 142 versus 174, average CPC 870 versus 641 HUF. Nine imported phone-link signals are phone intent, not verified connected calls or jobs. Previous zero was affected by measurement interruption. Attribution of earlier QA to later imports is unresolved. No guessed test subtraction.

Search Console, September 4 to October 3 versus August 5 to September 3, Pacific: organic clicks 190 versus 174, impressions 17,055 versus 19,321, CTR 1.11% versus 0.90%, average position 10.1 versus 11.9. Deep-garage page 19 clicks, motorcycle page 11. Positive demand indicators, not causal SEO uplift or revenue. September 30 and October 3 releases were too late to explain the full comparison.

GA4 uses a separate September 5 to October 4 window. Paid homepage: 45 sessions, 34 engaged, seven with phone interaction. Engagement is not a qualified enquiry. Tiny town samples and differing platform windows cannot establish geography profitability.

Generic uses a verified 50 km Budakeszi presence radius. Town-query campaigns use Hungary presence targeting. A query containing a town does not establish where the vehicle is located. Buda has only one click in the compared window. No broad geographic move is justified by that sample alone.

## 3. Completed work and its purpose

- September 29: verified 21 requested RSAs, 400 enabled local exact keywords, seven-day schedules and three paused automatic keyword-removal subscriptions. Purpose: preserve approved copy, reach and operating hours. Configuration is not delivered spend or work volume.
- September 30: 15 approved copy rows across six existing pages, service-card links, mobile/review presentation and manual carousel. The 58 held copy rows remain held. Purpose: clearer, approved service information and contact paths.
- October 3: 38 phone ContactPoint schema repairs, historical sitemap redirect, robots cleanup, existing optimized image restoration and one contrast correction. Purpose: accurate machine-readable service information and a complete working package.
- October 6: current policy/geo readback and comparative report. All 21 active ads approved/eligible. Purpose: establish what actually needs a fix and make the decision evidence reviewable.

Four visual follow-ups remain. Job attribution, qualified enquiries, revenue and marketing return have not been established. No retrospective hours are invented.

## 4. Optional deliverables and why

### A. Enquiry-to-job decision table

Existing empty CSV template, with no invented rows. Record service, actual pickup and destination, quoted/accepted fee, outcome, rejection reason, revenue and direct cost when known. Separate proven, caller-reported and unknown source. Names and phone numbers are unnecessary. Deliverable: a complete agreed observation window and a summarized suitable-job/direct-margin table. Why: distinguish poor commercial fit from poor marketing. Dependency: client-owned outcomes and commercial preferences. Margin is revenue minus direct cost, not total business profit.

### B. Closed-October Ads decision brief

Deliverable: read-only campaign and query comparison, Generic/local separation, spend, phone intent, joined known job outcomes, recommendation and exact evidence that would change it. Why: October is the first full month after the September 29 settings. Budget ceilings and low delivery are separate facts. No guaranteed spend floor. Missing outcomes yield conditional recommendations, not a fabricated return figure. No bid/budget/geo changes run automatically.

### C. QA and phone-intent reconciliation ledger

Deliverable: event date, session/test marker when available, GA4 source evidence, Ads conversion-action role/import date and confidence, with unresolved rows explicit. Why: older QA and later imports must not be combined. History confirms manual testing mode was explicitly temporary. A fresh October6 homepage/script read matches the current dist temporary tracking-consent.js; closing that temporary state needs a separately scoped measurement decision, not automatic execution. The dated September 23 snapshot has 17 labeled QA events and three secondary conversions. These are separate from the later nine primary imports. Neither subtracting 17 nor treating nine as jobs is valid.

### D. Existing-service SEO review at 30 and 60 days

Deliverable: page/query tables for deep-garage, motorcycle and misfuel services, consistent completed windows, deployment dates, and small evidence-led content/internal-link proposals. Why: these services have concrete demand signals, while late releases have had little observation time. Success evidence combines persistent useful demand with known enquiry quality. No bulk new pages or assumed ranking causation.

### E. Small contact-path CRO patch proposal

Deliverable: exact issue/screenshot inventory, chosen fixes and fresh desktop/mobile proof for contact or service comprehension. Four known findings concern desktop widget/card overlap, mobile value alignment, a sideways photo and an unsupported icon. Why: remove evidenced friction cheaply. It is a proposal. Whole-site visual acceptance is not clean, and no conversion uplift is promised.

## 5. What would change the recommendation?

If suitable jobs repeatedly have good direct margin and capacity exists, propose a bounded paid-media experiment with exact affected settings and an observable stop condition. If local short jobs are never acceptable, propose a query/message filter grounded in actual requests. If they are acceptable above a minimum fee, clarify that commercial offer first. If useful organic service demand persists, improve those existing pages. None of these options is launched by this document.

## 6. Review route

Read the recommendation, compare the evidence and limits, inspect completed work, then assess the optional deliverables and the business questions. Comments explain intent. Voice is read-only. Neither creates approval or executes a plan.