LAB LIVE
SCANNING416 CAMPAIGNS
ANOMALIES15 FLAGGED
AVG LIFT+4.40× ROAS
Restaurants · Hospitality

Restaurants, measured

at covers — not clicks.

Local Search and Maps as managed channels. Reservation-system attribution closed end-to-end. Google Business Profile run as part of the channel mix, not a side project. Direct account access where MCC breaks.

Reservation → cover
ConversionClosed loop
Managed
GBPWeekly cadence
Direct
AccessMCC bypass when needed
RESTAURANT · SAMPLE
EU
Reservations / month
342
+62%
CAC / cover
€4.20
−51%
GBP impressions
84K
+118%
Channel mixcover-weighted
Local
Search
GBP
Meta
Local
38%
Search
28%
GBP
22%
Meta
12%
Customer journey · attribution pathmeasured
01
Local query
'restaurant athens taverna'
02
Map result
GBP profile · reviews + photos
03
Site visit
Menu + price + booking CTA
04
Reservation widget
gclid passed via URL
05
Booking confirmed
CRM record · cover_id
06
Cover served
EOC fires on actual attendance
Sample dashboard · data anonymized◆ pattern from our practice

Where restaurant accounts go wrong.

  • 01

    PMax pulls junk traffic into a local-intent business

    Restaurants run Performance Max because someone said it works. PMax interprets 'restaurant near X' as commerce intent and serves your ad to people in another country. Click cost goes up, reservations stay flat.

  • 02

    Reservation system silos the conversion

    TheFork, OpenTable, SevenRooms, Quandoo — third-party reservation widgets that don't pass the gclid. GA4 sees session_end where the booking actually happens. Channel attribution evaporates.

  • 03

    Google Business Profile is treated as a side project

    Posts irregular, photos outdated, opening hours wrong half the year. Maps and Local Pack signal degrades. Paid Search competes with itself instead of complementing organic local presence.

  • 04

    Multilingual menu, single-language ads

    Tourist-heavy locations need targeted creative in resident-language and traveller-language. Most accounts run one set of ads and wonder why CPC differs across districts.

Eight pieces tuned to local intent.

  • 01

    Local Search architecture

    Cuisine + neighbourhood + intent queries mapped properly. Branded vs generic split. tCPA targets per language variant. PMax kept off accounts where it would distract.

  • 02Loop closed

    Reservation system attribution

    gclid passed into reservation widget URL params. Captured in CRM or reservation platform export. Daily reconciliation against actual covers — not against form_submit.

  • 03

    Google Business Profile as a managed channel

    Weekly posts, photo refresh cycles, Q&A monitoring, review response cadence. GBP insights pulled into the same dashboard as Ads.

  • 04

    Map insights and competitor positioning

    Local Search ranking tracked across cuisine-relevant queries. Competitor profile changes monitored. Promotion windows (lunch, brunch, prix-fixe) timed.

  • 05Multilingual

    Resident vs traveller targeting

    Different creative, different language, different reservation flow. Geo-targeting tuned per district. Traveller intent during conference seasons separated cleanly.

  • 06

    Delivery channel coordination

    If the restaurant runs delivery (Glovo, Deliveroo, Wolt), creative and budget split managed to avoid cannibalising direct reservations or dine-in.

  • 07

    Direct customer-level access

    Where MCC access fails (we've seen this — Symposio, EVA Instruments), we operate via direct account access. No PERMISSION_DENIED surprises.

  • 08

    Reservation-by-channel dashboard

    Daily covers by source. CAC per cover by channel. ROAS calculated against actual revenue per cover, not ticket value at reservation.

Foundation, search, operate.

  1. 01

    Restaurant audit

    Phase 01 · 5 working days

    Ads, GA4, GBP, reservation system. Map the channel mix, score gclid-to-reservation integrity, surface the GBP gaps.

    • Channel mix reality check (paid + organic + GBP)
    • Reservation gclid bridge integrity map
    • GBP completeness and competitor positioning
    • Resident vs traveller demand split
  2. 02

    Foundation

    Phase 02 · Weeks 1–2

    GBP brought to operational standard. Reservation attribution bridge wired. PMax retired or contained as appropriate.

    • GBP refresh: photos, posts, Q&A
    • Reservation attribution live end-to-end
    • PMax decision (contain / retire)
    • Channel group correcting Local vs Paid Search
  3. 03

    Search architecture

    Phase 03 · Weeks 2–5

    Search rebuilt by intent and language. Resident vs traveller separated. Delivery channels coordinated.

    • Search architecture by cuisine × neighbourhood × intent
    • Multilingual creative variants
    • Delivery vs dine-in budget split
    • Promotion calendar tied to demand windows
  4. 04

    Operate

    Phase 04 · Ongoing

    Weekly GBP post cadence, monthly photo refresh, ongoing Search optimisation, seasonal promotion management.

    • Weekly GBP posts and review responses
    • Monthly photo and Q&A refresh
    • Ongoing Search and bidding optimisation
    • Seasonal promotion management

Common questions.

  • Probably yes for a neighbourhood restaurant doing well on word of mouth. The economics start working at 2+ locations, or for destination restaurants where tourist demand is meaningful. Below that, GBP optimisation alone is usually enough — we'll tell you that during the audit.

  • We don't manage the platform accounts directly — those are operational, not media buying. We coordinate budget and creative across dine-in paid ads and delivery so they don't cannibalise each other in reporting.

  • Most can, with one engineering ticket on their side. If yours genuinely can't, we work with what's available — UTM passing, server-side reinforcement, time-window correlation. Imperfect attribution beats no attribution.

  • If you mean Instagram-driven brands without paid spend, no — we're a performance marketing lab, not a social agency. If you mean restaurants where Instagram is part of the demand mix alongside paid, yes — we treat the organic signal as input to channel decisions.

  • We've solved this exact problem at Symposio. Where MCC write access throws PERMISSION_DENIED, we operate via direct account access with login_customer_id omitted. Documented in our practice, not improvised.

Open for clients

Send the account and the reservation system.

Read-only access. Five working days. Written diagnostic with channel mix and reservation-attribution health.

No long contracts · Free audit before any work