01 / 12 · Marketing opportunity · business context
A Strong Brand With Room to Capture More Local Demand
Mozza has strong ratings, a loyalty programme and guests who search for it by name. The marketing question is narrower: can Search, Maps and branch pages bring additional high-intent local guests, profitably, branch by branch, and can we prove it?
- Reputation
- Loyalty
- Brand search
- Branch discovery
Evidence, confidence and presenter notes
Question this slide answers
What is the marketing opportunity, and what is it not?
Evidence
- PUBGoogle ratings of 4.5–4.9 across four branches (≈ 6.0K reviews, Oct 2026).Confidence: Medium.
- PUBHMA Privilege loyalty programme exists.Confidence: High.
- 3P“mozza by cocotte” ≈ 2.4K searches a month (Mangools, Oct 2026).Confidence: Medium.
- HYPAdditional profitable covers can be won from local, high-intent search: untested until the pilot.
Presenter notes
Thank you for the time. I am not here to audit your marketing. Mozza has a strong reputation and guests who search for it by name. The opportunity I want to test is narrower: local, high-intent demand around each branch, in the right language and at the right time, and whether it can be captured profitably and measured by branch. Everything on these slides is labelled by how sure we are, and nothing assumes that current marketing underperforms.
PUBPublic listing or page (dated)3PThird-party estimate, not analyticsHYPHypothesis to test
02 / 12 · Brand versus non-brand demand
Brand Demand Is Visible; the Larger Pool Is Local and Non-Brand
Guests who know Mozza search by name. A separate pool searches by area, mall and cuisine, in Thai and English. Keyword Planner ranges show that pool at the same or higher volume than brand searches. Ranges are estimates, never summed.
- Brand
- Local, non-brand
- Thai
mozza by cocotte
mozza emquartier
italian restaurant near me
pizza near me
italian restaurant bangkok
pasta bangkok
ร้านอาหารอิตาเลี่ยน ใกล้ฉัน
พิซซ่า
Google Keyword Planner, Bangkok, Oct 2025 – Sep 2026. Monthly search ranges (log scale), not additive, not Mozza analytics.
Evidence, confidence and presenter notes
Question this slide answers
Where does new demand sit compared with the demand Mozza already has?
Evidence
- 3PKeyword Planner ranges (Bangkok, all languages, Google Search, Oct 2025 – Sep 2026). Ranges are estimates, not additive.Confidence: Medium.
- OBSOn generic “Italian restaurant Bangkok” searches other restaurants appeared in the top results in our Oct 2026 snapshot (one device, one place and time).Confidence: Medium.
- 3PBrand-query position differs by source: first in our own test (one device), about position 2 in Mangools (Oct 2026), and position 14 for “mozza by cocotte” in a Semrush Position Tracking fragment dated Nov 5, 2026 whose location, device and language are not visible. Positions vary; we align one location, device and language before making any claim.Confidence: Low.
- DATAShare of reservations from brand versus non-brand search can only come from HMA's reports.
Presenter notes
Brand demand is visible and healthy. The second pool is guests with no name in mind: they search for a mall, an area or 'Italian near me'. Keyword Planner shows that pool at the same or higher range, but ranges are estimates and not additive. I am also flagging that brand-query positions differ between sources; I will not claim anything until we compare like for like. The question is how many of those guests Mozza already wins, and your data answers that.
3PThird-party estimate, not analyticsOBSOur own test (date and device stated)DATANeeds HMA information
03 / 12 · Branch-level local discovery
Local Discovery Happens Branch by Branch
Each branch has its own mall and its own Thai and English search terms, and volumes differ by branch. Branch pages, profiles and menus need to match how guests search for that specific location, in both languages.
Keyword Planner monthly search ranges for the mall-name term of each branch (not additive).
| Branch | English term · range | Thai term · range |
|---|---|---|
| Siam Paragon | siam paragon restaurant · 1K–10K | ร้านอาหาร สยามพารากอน · 1K–10K |
| ICONSIAM | iconsiam restaurant · 1K–10K | ร้านอาหาร ไอคอนสยาม · 1K–10K |
| EmQuartier | emquartier restaurant · 1K–10K | ร้านอาหาร เอ็มควอเทียร์ · 1K–10K |
| Central Park | dusit central park restaurant · 1K–10K | ร้านอาหาร ดุสิต เซ็นทรัล พาร์ค · 100–1K |
| Central Chidlom | central chidlom restaurant · 100–1K | ร้านอาหาร เซ็นทรัลชิดลม · 1K–10K |
Siam Paragon
- English term · range
- siam paragon restaurant · 1K–10K
- Thai term · range
- ร้านอาหาร สยามพารากอน · 1K–10K
ICONSIAM
- English term · range
- iconsiam restaurant · 1K–10K
- Thai term · range
- ร้านอาหาร ไอคอนสยาม · 1K–10K
EmQuartier
- English term · range
- emquartier restaurant · 1K–10K
- Thai term · range
- ร้านอาหาร เอ็มควอเทียร์ · 1K–10K
Central Park
- English term · range
- dusit central park restaurant · 1K–10K
- Thai term · range
- ร้านอาหาร ดุสิต เซ็นทรัล พาร์ค · 100–1K
Central Chidlom
- English term · range
- central chidlom restaurant · 100–1K
- Thai term · range
- ร้านอาหาร เซ็นทรัลชิดลม · 1K–10K
Evidence, confidence and presenter notes
Question this slide answers
Which branches have which local demand, and in which language?
Evidence
- 3PKeyword Planner ranges for each branch's mall-name terms (Bangkok, Oct 2025 – Sep 2026).Confidence: Medium.
- OBSIn our Oct 2026 review the Central Park page sat at a temporary-looking address with another branch's title, and branch pages were mostly images with a booking button. Being re-verified.Confidence: Medium.
- DATAThe current branch list (including CentralWorld and any other venue) is to be confirmed with HMA.
Presenter notes
Local search is branch-specific. EmQuartier, ICONSIAM, Siam Paragon and Central Park all show their mall names in the same range; Central Chidlom is stronger in Thai than in English. Which branch to start with is a decision for your data: capacity, margin and the quality of the current branch page. I would not choose from keyword volumes alone.
3PThird-party estimate, not analyticsOBSOur own test (date and device stated)DATANeeds HMA information
04 / 12 · Google Ads and local high-intent acquisition
Google Ads: Capped, Local, High-Intent Search
A small Search test on one branch: mall and “Italian near me” intent, the branch page as landing page, separate Thai and English ad groups, a capped budget. No Google Ads activity was observable in our review. Performance Max only after Search proves value.
1Structure
What we would set up
Search only, one branch. Ad groups: mall or area, Italian near me, pizza, pasta; Thai and English separated; exact and phrase match.
How we would judge it
Impression share, click-through rate and search terms reviewed weekly.
2Landing page
What we would set up
The branch page with address, hours, menu and Reserve, never the homepage.
How we would judge it
Booking clicks per branch-page visit.
3Bidding
What we would set up
Maximise clicks with a CPC cap; move to conversions once reservation actions are tracked.
How we would judge it
Cost per booking click, then cost per confirmed reservation.
4Guardrails
What we would set up
Negative keywords (recipes, jobs), brand terms kept out of the non-brand campaign, budget cap and stop rule agreed before launch.
How we would judge it
Share of spend on irrelevant terms; spend against the cap.
Evidence, confidence and presenter notes
Question this slide answers
What would a minimal, low-risk Google Ads test look like?
Evidence
- 3PKeyword Planner top-of-page bid estimates for non-brand terms are mostly about €0.1–0.9 (≈ ฿4–35). A bid is a price per click, not a price per table.Confidence: Medium.
- OBSNo observable Google Ads footprint was identified for Mozza Bangkok or Hospitality Management Asia in Google's Ads Transparency Center during our review (Oct 2026).Confidence: Medium.
- HYPThe click-to-table rate, show rate and incrementality are unknown: the test exists to measure them. Cost per table = cost per click ÷ click-to-table rate.
- DATABranch capacity by hour, any existing Google Ads account and the media budget come from HMA.
Presenter notes
This is deliberately small. One branch, Search only, capped, landing on the branch page. Bids for these terms are cheap per click on paper, but a bid is a price per click, not per table; what matters is the click-to-table rate, which nobody knows yet. I would not run it at all unless the branch has seats to fill and we can see reservations by source. Performance Max comes later, if Search proves value.
3PThird-party estimate, not analyticsOBSOur own test (date and device stated)HYPHypothesis to testDATANeeds HMA information
05 / 12 · Multilingual targeting
Thai and English First, Arabic Only if Data Supports It
Thai and English demand are both visible in keyword data. Arabic is a hypothesis: no demand data has been collected and the international booking path has not been tested. Thai branch pages come first; Arabic follows evidence.
1
Thai
Dedicated Thai branch pages and Thai ad groups: demand is visible
2
English
Branch pages built for local and tourist intent: demand is visible
3
Arabic
Test only if booking or search data shows demand
Language decisions follow booking and Search Console data
Evidence, confidence and presenter notes
Question this slide answers
Which languages justify investment, and on what evidence?
Evidence
- 3PThai terms such as “ร้านอาหารอิตาเลี่ยน ใกล้ฉัน” sit in the 1K–10K range and “พิซซ่า” in the 10K–100K range (Keyword Planner).Confidence: Medium.
- OBSIn our Oct 2026 review the Thai version of the main site displayed English content. Being re-verified.Confidence: Medium.
- HYPArabic-language demand and a tested international booking path are unproven: no Arabic data has been collected.
- DATAThe language and country mix of reservations can only come from HMA.
Presenter notes
Thai and English are both evidenced. Arabic is an open question, not a plan. This demo site shows how Thai and Arabic pages would work, but I would only invest in Arabic if your booking data or Search Console shows international demand. Thai native review of any copy comes before anything goes live.
3PThird-party estimate, not analyticsOBSOur own test (date and device stated)HYPHypothesis to testDATANeeds HMA information
06 / 12 · SEO and Local SEO architecture
Structure Guests and Search Assistants Can Read
Branch pages with address, hours, directions and menu text; Business Profile links to the right branch page; clean branch URLs; structured data; links between branch, menu and language pages. Our observations date from October 2026 and are being re-verified.
OBSObserved (our review)
- Main food and breakfast menus are image-only PDFs
- Branch pages mostly images plus a booking button
- Some profile links point to the homepage; one public detail differs between sources
3PEstimated (third-party tools)
- Two tools estimate organic visits about 7–8 times apart
- A crawler reports 82% site health: 6 errors, 73 warnings (counts only)
DATATo validate with HMA data
- Search Console queries and pages by branch
- Business Profile calls, directions and website clicks
- Which pages currently rank for each branch query
Evidence, confidence and presenter notes
Question this slide answers
What search structure would make each branch easy to find and to choose?
Evidence
- OBSPublic-site and listing review, Oct 2026 (nine observations): image-only menus, branch-page structure, profile links, address difference, listing prices. Notes supplied by the project owner; re-verification pending.Confidence: Medium.
- 3PSemrush (Nov 4, 2026) and Ahrefs (Oct 2026) estimate Mozza's monthly organic visits roughly 7–8 times apart; neither is Mozza analytics.Confidence: Low.
- 3PSemrush Site Audit, Nov 5, 2026: health 82%, 6 errors, 73 warnings, 58 pages crawled. The issue list has not been reviewed.Confidence: Medium.
- DATASearch Console and Business Profile data are needed to size any of this.
Presenter notes
The structure question is simple: can a guest or an assistant find the right branch, read the menu, and reserve? Our review suggests branch pages and menus carry little readable text, but I am labelling that as our observation, to be re-verified. The traffic tools disagree with each other, so decisions should rest on your Search Console, not on a model. This is a prioritisation exercise, not a list of faults.
OBSOur own test (date and device stated)3PThird-party estimate, not analyticsDATANeeds HMA information
07 / 12 · Customer journey and reservation conversion
From Search to Seat: Each Step Can Be Tested
Discover, choose the branch, read the menu, reserve, visit. Each step can be tested without assuming a problem. International guests need a booking path tested end to end, including phone-number formats, before anything is claimed.
1Discover
What guests need
A clear, local result for the area, mall or cuisine they searched.
How we would test it
Local-intent impressions and clicks by branch (Search Console).
2Choose a branch
What guests need
Address, hours, how to get there, one clear Reserve action.
How we would test it
Branch-page views, directions requests, booking clicks.
3Evaluate the menu
What guests need
Readable dishes and prices before deciding.
How we would test it
Menu-page engagement, then booking clicks.
4Reserve
What guests need
A short path, in their language, with their phone number format.
How we would test it
Booking-click rate, click-to-confirmed rate, completion by language.
5Visit
What guests need
Confirmation and directions that match the branch.
How we would test it
Confirmed reservations and seated covers by branch (HMA systems).
Evidence, confidence and presenter notes
Question this slide answers
Where could the path from search to reservation be strengthened, and how would we know?
Evidence
- OBSReservations are completed on an external booking platform (Oct 2026 review).Confidence: Medium.
- HYPFriction at any step is a hypothesis: no funnel data has been seen.
- DATAFunnel and completion data by branch, and the booking platform's reports, come from HMA.
Presenter notes
Five steps, each with one thing to test. I am not asserting that guests drop out anywhere; I am saying each step is measurable and cheap to improve if it turns out to matter. The international path, phone-number formats and languages, deserves a hands-on test before we discuss it further.
OBSOur own test (date and device stated)HYPHypothesis to testDATANeeds HMA information
08 / 12 · Attribution, GA4, GTM and measurement
First, See What Is Measured Today
Before any media spend we need to see what exists: GA4 and Tag Manager, Search Console, Business Profile insights and booking reports, by branch. Then we agree events, UTM conventions and one branch-level report. We assume nothing about what is or is not in place.
HYPEvents to agree (examples)
- Reservation start or booking click
- Directions and call
- Messaging (WhatsApp, LINE)
- UTM on every booking link
DATASources to review with HMA
- GA4 and Tag Manager (if any)
- Search Console
- Business Profile insights
- Booking platform reports by branch
OBSLimits we will state
- No exact branch or source attribution until the booking provider's source fields are known
- Modelled store visits only support, never lead
Evidence, confidence and presenter notes
Question this slide answers
What do we need to measure, and what can attribution honestly tell us?
Evidence
- HYPEvent names and conventions are a proposal to agree with HMA's team.Confidence: Medium.
- DATAWe have not seen HMA's analytics, Search Console, Business Profile or booking reports, so the current set-up is unknown. It does not show that anything is missing.
- OBSMetrics are kept in separate layers: Maps and profile interactions, booking clicks, confirmed reservations, seated covers and revenue.Confidence: High.
Presenter notes
I will not assume anything is broken. HMA may already have much of this. The first job is a short read-only review of what exists, by branch. Then we agree a handful of events and a UTM convention, and one report. The honest limit: attribution can only go as far as the booking provider's reporting allows; where it cannot, we use a baseline and a comparison branch.
HYPHypothesis to testDATANeeds HMA informationOBSOur own test (date and device stated)
09 / 12 · Incremental seated covers and contribution
Success Means Incremental Covers and Contribution
We measure incremental seated covers and contribution after media, commissions and relevant operating costs, against a baseline. Clicks and rankings are leading indicators only. No result is promised: if the data cannot show a clear benefit, we say so.
- Incremental revenue
- −Media
- −Commissions
- −Relevant operating costs
- =Incremental contribution
Supporting metrics, in order
- Interactions
- Booking clicks
- Confirmed reservations
- Seated covers
Evidence, confidence and presenter notes
Question this slide answers
Which number decides whether marketing paid back?
Evidence
- HYPMethod: incremental, capacity-capped, net of costs, against a baseline and a comparison branch where one exists. No outcome is promised.Confidence: High.
- DATAAverage spend per cover, party size, margin, commissions and capacity by hour can only come from HMA. No figure from this deck is a Mozza number.
Presenter notes
Incremental means what would not have happened anyway. We compare with the branch's own baseline and, where possible, another branch. Contribution is after media, any booking-platform commission and the operating costs that extra covers bring. Thresholds are agreed before launch so nobody moves the goalposts. I make no promise about the result.
HYPHypothesis to testDATANeeds HMA information
10 / 12 · A controlled pilot
A Controlled 90-Day Pilot
One branch, 90 days, compared with a baseline. Weeks 1–4 prepare with no media spend; weeks 5–13 test; day 90 is a decision point: go, adjust or stop. Branch, hours and budget are set with HMA and depend on capacity and measurement.
Agree and baseline (weeks 1–2)
Prepare: tracking, branch page, profile, Thai page (weeks 1–4)
Test: capped local Search, weekly optimisation (weeks 5–13)
Week
- Go
- Adjust
- Stop
Conditional on operational capacity and measurement feasibility.
Evidence, confidence and presenter notes
Question this slide answers
What exactly would we do, how long would it take, and how can HMA exit?
Evidence
- HYPPilot design: a test branch against its own baseline, plus a comparison branch where one exists. The branch is not chosen.Confidence: Medium.
- DATAFeasibility depends on HMA's capacity data and on what reservation reporting can show.
Presenter notes
Small and reversible. The first month is preparation with no media spend. Media starts only once tracking is verified. At day ninety we decide together: go, adjust or stop. I would only start if the branch has seats to fill in the hours we target and we can measure reservations well enough to judge the result. If either is false, I would tell you and not start.
HYPHypothesis to testDATANeeds HMA information
11 / 12 · Decision criteria
Decision Criteria Are Written Before Launch
Four gates, agreed with HMA in weeks 1–2 and written down before any media starts. The form is fixed now; the thresholds are set with your numbers. Platform migration is not part of the pilot.
| Gate | Rule (thresholds agreed with HMA in weeks 1–2) |
|---|---|
| Start | Capacity exists in the targeted hours AND basic reservation counts by branch are available AND HMA approves the branch, hours and budget cap. Otherwise: do not start. |
| Go | Net incremental contribution after media, commissions and relevant operating costs meets the agreed threshold, in slots with spare capacity, with data quality passed. |
| Adjust | Positive early signals but too few events or incomplete data: extend by up to 30 days with agreed fixes. |
| Stop | Net incremental contribution below the threshold after the agreed learning period, no measurable difference from baseline, or data cannot support a conclusion: spend stops at once. |
Start
- Rule (thresholds agreed with HMA in weeks 1–2)
- Capacity exists in the targeted hours AND basic reservation counts by branch are available AND HMA approves the branch, hours and budget cap. Otherwise: do not start.
Go
- Rule (thresholds agreed with HMA in weeks 1–2)
- Net incremental contribution after media, commissions and relevant operating costs meets the agreed threshold, in slots with spare capacity, with data quality passed.
Adjust
- Rule (thresholds agreed with HMA in weeks 1–2)
- Positive early signals but too few events or incomplete data: extend by up to 30 days with agreed fixes.
Stop
- Rule (thresholds agreed with HMA in weeks 1–2)
- Net incremental contribution below the threshold after the agreed learning period, no measurable difference from baseline, or data cannot support a conclusion: spend stops at once.
Evidence, confidence and presenter notes
Question this slide answers
How will we decide to start, scale, adjust or stop?
Evidence
- HYPDecision form proposed for agreement; no numeric threshold is assumed.Confidence: High.
- DATAThresholds, minimum event counts and the budget cap are set with HMA after discovery.
Presenter notes
The point of this slide is that the rules exist before the result does. I do not have your margins, so I am not proposing numbers. In the first two weeks we set thresholds with your finance and operations people. If the pilot does not clear the bar, we stop and the foundation work stays with you.
HYPHypothesis to testDATANeeds HMA information
12 / 12 · The ask
The Next Step: A 20-Minute Executive Meeting
Twenty minutes to hear your marketing priorities, which branch and hours matter most, and which data already exists. Then we decide together whether a pilot is worth scoping. We ask for no data access before that conversation. Contact: Wadhah Belhassen.
- Executive meeting
- Scope (only if worthwhile)
- 90-day pilot
Evidence, confidence and presenter notes
Question this slide answers
What is the smallest decision we are asking for?
Evidence
- PUBContact details are the ones already published on the Strategy page.
Presenter notes
My ask is small: twenty minutes, whenever suits you. I would like to hear which branches and hours matter to you, what you already track, and where you would not want me to go. I am not asking for any access or data before that. If it makes sense afterwards, I will scope a pilot with you. If it does not, we have both learned something.
PUBPublic listing or page (dated)
Appendix
Supporting detail, shared after the discovery meeting. Technical and measurement detail lives here, not on the slides.
A. Evidence register
Every statement on the slides, with its label, source and confidence.
| Statement used | Label | Source / date | Confidence | Slide |
|---|---|---|---|---|
| Ratings 4.5–4.9, ≈ 6.0K reviews, four branches | Public listings | Google Maps, Oct 2026 | Medium | 1 |
| HMA Privilege loyalty programme | Public page | Current | High | 1 |
| “mozza by cocotte” ≈ 2.4K searches a month | Third-party estimate | Mangools, Oct 2026 | Medium | 1 |
| Brand, non-brand and Thai search-volume ranges; mall terms by branch | Third-party estimate | Keyword Planner, Bangkok, Oct 2025 – Sep 2026 | Medium | 2, 3, 5 |
| Other restaurants in top results on generic Italian searches | Our own snapshot | Oct 2026 | Medium | 2 |
| Brand-query position differs by source | Our test / Mangools / Semrush fragment | Oct – Nov 2026 | Low | 2 |
| Non-brand top-of-page bids ≈ €0.1–0.9 | Third-party estimate | Keyword Planner, Oct 2026 | Medium | 4 |
| No observable Google Ads footprint in Ads Transparency Center | Our own review | Oct 2026 | Medium | 4 |
| Image-only menus, thin branch pages, Thai page showing English, profile links, address difference | Our own review (owner-supplied notes) | Oct 2026, re-verification pending | Medium | 3, 5, 6, 7 |
| Organic-visit estimates ≈ 7–8× apart; Site Audit 82%, 6 errors, 73 warnings | Third-party estimates | Semrush Nov 2026; Ahrefs Oct 2026 | Low / Medium | 6 |
| Branch list, funnel, capacity, margin, language mix, analytics set-up | Needs HMA information | — | Open | 3–10 |
B. Measurement methodology
Implementation detail, relevant only once HMA agrees to scope a pilot.
- KPI ladder (never blended): Maps and Business Profile interactions → booking clicks → confirmed reservations (provider-reported) → seated covers and revenue (point of sale) → modelled store visits (supportive only, if the account is eligible).
- Incrementality: a pre-period baseline of 4–8 weeks; difference-in-differences against the branch's own baseline and, where one exists, a comparison branch; optionally targeted versus untargeted slots within the branch.
- Tracking (only after HMA agrees to scope): GA4 key events (reservation start or click, directions, call, WhatsApp, LINE); a Tag Manager container review; UTM on outbound booking links; Business Profile performance export; provider reservation exports by source and branch; covers and revenue by hour.
- Formula: incremental confirmed reservations × show rate × party size = incremental covers, capped by available seats in the targeted slots; × spend per cover × contribution margin = contribution before costs; net incremental contribution = that − media − provider commissions − relevant incremental operating costs.
- Cost per table = cost per click ÷ click-to-table rate: the rate is unknown until measured.
- No exact branch or source attribution is promised until the booking provider's source fields are known.
- No example figure in this deck is a Mozza number.
C. Information requested from HMA
Requested at or after the executive meeting. Nothing is requested before it. Aggregated counts only; no guest personal data.
- Tier 1 (to scope a pilot): branches in scope; average check and party size per branch; margin definition and any commissions; covers or occupancy by hour, weekday versus weekend, 8–12 weeks; reservations by source and no-show rate; current marketing spend by channel and branch, and any Google Ads account; which analytics and booking reports exist and who owns them.
- Tier 2 (once scope is agreed): read-only access to Search Console, GA4 and Tag Manager, and Business Profile insights; reservation-platform exports.
D. Risks and assumptions
| # | Risk / assumption | Mitigation |
|---|---|---|
| 1 | The chosen branch may be near capacity in the targeted hours | Choose the branch and hours from HMA's own data; do not start if capacity is not there |
| 2 | Reservation reporting may not show source or branch | Use a baseline plus a comparison branch; judge on confirmed reservations and covers; state the limit |
| 3 | Incrementality may be low (guests would have booked anyway) | Judge incremental, not attributed, results; holdout where appropriate |
| 4 | Third-party tools can mislead (traffic estimates differ; brand positions differ by source) | Not used for decisions; replaced by HMA's first-party data |
| 5 | A small sample over 90 days | Pre-agreed minimum event counts; “Adjust” extends by up to 30 days |
| 6 | Seasonality or events distort the read | Compare with the baseline and a comparison branch; record events |
| 7 | Our own observations are dated and single-device | Re-verify and repeat before any external claim |
| A | Assumes HMA wants to explore additional covers on weekdays and in quieter hours | Confirm in the meeting |
Independent growth proposal by Wadhah Belhassen. Not an official Mozza / HMA document. Concept demo; third-party figures are estimates, not Mozza analytics.