05What the history shows
Patterns found by re-reading the full history, not just the months I ran. Each is labelled: measured pattern, correlation, anomaly or seasonality.
Measured pattern55 Google "conversions", about 9 real leads
Map "Directions" taps were set as a main conversion, so Google bid for people who wanted directions, not a booking. Searches like "fun world bangalore timings" showed up as conversions.
The numbers
1–17 Oct Search: 3,782 impressions, 599 clicks (15.84% CTR), ₹5,229, 55 conversions of which 52 Directions. Real leads ≈9 (5 WhatsApp + 4 click-to-call). ₹95.07 per reported conversion vs ≈₹580 per real lead. GA4 phone_click has never fired.
So whatJudge Google on calls, WhatsApp and forms. Keep Directions as secondary. Fix website call tracking before reading any cost per lead.
Measured patternFewer people see the profile, more of them act
Profile views and search appearances fell year on year, but actions rose, so each view works harder. Almost all views are on a phone, mostly Google Maps.
The numbers
May–Sep 2026 vs 2025: actions +11.0%, directions +12.5%, calls +15.1%, views −30.5%, search appearances −47.1%. Actions per 100 views, Sep: 10.8 (2025) to 15.3 (2026). 414,975 views 1 May–11 Oct, 93.5% on phone; Maps mobile 74.8%. Menu views 0 every month; 2 booking clicks.
So whatThe phone Maps listing is the shop window. Visibility is the falling number: photos, weekly posts, a menu and a booking link come before more ad spend. Most of this period is before my changes (28 Sep), so it is a baseline, not a result.
AnomalyJanuary: 1,033 enquiries on ₹15k of ads
Jan was the biggest enquiry month of the year on the smallest ad spend of the first half. Q1 brought 84% of Q3's enquiries on 9.5% of Q3's spend. Each extra enquiry has cost more as spend rose, and Meta lead cost climbed from June to September.
The numbers
Jan 1,033 enq (18.5% of 5,588), spend ₹14,959. Q1 1,862 enq / ₹52,766 (₹28 each); Q2 1,512 / ₹1,72,419 (₹114); Q3 2,214 / ₹5,53,134 (₹250). Extra spend per extra enquiry: Q2 to Q3 ₹542; Q1 to Q3 ₹1,421. Correlation of spend with enquiries: r 0.22 (Jan-Sep), 0.61 (Mar-Sep). Meta cost per dated lead: Jun ₹131, Jul ₹635, Aug ₹311, Sep ₹1,070; Aug to Sep Meta spend +81% (₹85,733), leads 340 to 179. Aug to Sep total spend +98%, enquiries -5% (720 to 682), GA4 users 3,872 to 20,962.
So whatDemand was already there before the ads scaled. Find what January was (season, group bookings or a CRM backfill; no 2025 data to test) before reading Q3 as growth. Judge spend on cost per booking once source is tagged. Meta spend is booked at campaign end month, so monthly CPL is rough.
Correlation, not causationConversion fell in June, before the spend jump, then flat-lined
Conversion dropped from 32% in May to 20% in June on ₹20.9k of spend, a month before Google Ads began. Jul-Sep sat at 14-15% while spend ranged ₹1.15L-2.28L. Bookings stayed 90-157 a month while enquiries ranged 272-812, so conversion follows volume, not spend. The 14% headline also counts Jan-Feb enquiries that have no status.
The numbers
Conversion Mar 33.1, Apr 24.3, May 32.4, Jun 20.3, Jul 14.9, Aug 14.4, Sep 14.1%. Bookings 90/157/94/117/121/104/96. r(conversion, enquiries) -0.94; r(conversion, spend) -0.68 (n=7). Each extra 100 enquiries added about 5 bookings (slope 0.054). 788/5,621 = 14.0%; Mar-Sep 779/3,998 = 19.5%. June also lost enquiry type on 565 of 575 rows and Meta lead sheet starts that month.
So whatExtra leads were absorbed without extra bookings: look at response speed, routing and capacity before adding spend. Report conversion on tracked months only. Check whether June's CRM or lead-sheet change altered what counts as an enquiry.
Correlation, not causationCalls fell May to August; ads did not lift Maps
Calls per direction request fell from 28% in April to 9% in September. Calls fell 55% from May to August, before and during the first Google ads, and only recovered 11% in September. Google's own store report credits 18% of Jul–Sep directions to ads. The profile sent thousands of website clicks but only 2 booking-link clicks. An earlier read of September was a part month and has been corrected.
The numbers
Calls/directions Apr 27.9%, May 23.6%, Jun 16.7%, Jul 13.3%, Aug 9.9%, Sep 9.3%. Google spend Jul/Aug/Sep ₹33,001/9,362/36,941; directions 7,538/7,303/8,565; calls 1,004/721/797 (full months, read 17 Oct). Ads store report 4,087 directions of 22,199 (18.4%), 224 store visits, Jul–Sep.
So whatCall loss is not a demand or ad-spend problem. Test the phone line, turn on chat routed to WhatsApp, and judge Performance Max on enquiries, not taps. A booking engine link would give the profile a booking path that is not a reseller.
Measured patternRevenue follows ticket size; Meta money sits away from where CRM enquiries sit
Average booking ranged ₹16.6k to ₹57.7k by month, so won value moves with ticket mix, not booking count. July was 45% of Q3 value; July to September ticket fell 20% and bookings 21%. Sales wins 32% of closed leads by count but 68% by value, because lost leads are small. Meta put 63% of spend on corporate and weddings; the CRM types those as 2.4% of enquiries.
The numbers
Avg booking Mar ₹57,710; Apr 24,514; May 16,634; Jun 21,270; Jul 50,534; Aug 35,638; Sep 40,506. Jul ₹61.1L of Q3 ₹1.37Cr. Aug to Sep: spend +98%, bookings -8%, value +5%, value per ad rupee 32.2x to 17.0x. Won 788 vs lost 1,657 (32%); won ₹2.69Cr vs lost ₹1.27Cr (68%); avg won ₹34.1k, avg lost ₹7.7k. Meta: corporate ₹2,47,530 + wedding ₹1,92,519 = 63% of ₹6,99,015; rooms + day-outs ₹33,259 (4.8%). CRM types: rooms + day-outs 3,599 (64%), corporate 54 + wedding 82 = 136 (2.4%); 884 untyped. Meta sheet leads: corporate 308, wedding 221, birthday 407 vs CRM-typed 54, 82, 154.
So whatJudge months on value and cost per booking by product. The CRM types look incomplete (June, 884 untyped), so reconcile the Meta lead sheet to the CRM before moving budget by product. Value by enquiry type is not in the sources.
Measured patternOne executive is 57% of wins and 57% of open pipeline value
Exec. A handles 28% of leads and books 57% of wins. Win rate on closed leads runs 58% for Exec. A and 14% for Exec. D, who lost 702 of 1,362 leads (42% of all lost). Exec. H has 0 wins from 450 leads. 13 February birthday rows hold 43% of open value. Opens carry no value on 158 leads.
The numbers
Exec. A 1,567 handled (28.5%), 447 won (56.7%), ₹1.22Cr won (45%), open ₹8.13Cr (57% of ₹14.29Cr). Closed-lead win rate: GM 75%, A 58%, F 50%, E 43%, B 40%, C 23%, G 15%, D 14%, H 0%. A+B+D = 71% of leads, 85% of wins. Exec. E: 18 wins at ₹1.35L average, 158 of 200 open with ₹0 value. Open 2,609 = 46% of enquiries; 567 have no status. 13 rows ₹6.2Cr (₹47.7L each) leaves clean open pipeline near ₹8.1Cr. If D matched A's 28.5%: +272 wins, about ₹79L at D's ₹29.2k ticket (assumes equal lead quality).
So whatRouting and hygiene, not media: rebalance load, review the D and H queues, quote or close the ₹0 opens, delete typo rows. One person is a single point of failure for most bookings. Lead assignment rules are not in the sources.
