The August Report

Three regimes. ₹14,900/day to 9 August. A collapse to ₹533 on the 10th when the balance ran dry — one trading day gone, roughly ₹32,000 of forgone delivery. Then ₹33,000–40,000/day to month end. The account showed "balance exhausted" again on the 31st.

Google's own clean-period data says 2.73×. The Shopify incremental method says 2.99×. They share no inputs. Landing within 10% of each other is the strongest evidence in this report.

Plan on 2.7–3.0× gross, 2.2–2.4× net. Not 146×.

Nothing has been corrected. The disputed days are excluded from every headline figure in this report and shown here in full.

Three bars agree. One does not. August bought 17% of the account's lifetime traffic for 15% of its lifetime cost — and recorded 37% of every conversion the account has ever booked. Remove the disputed campaign's August figures and the share falls to 24.9%, which is still high but no longer impossible.

The engine, drawn. Stand_14Jan booked 1,169.92 of its 1,231.55 conversions in four days — 95%, and 15% of the account's entire lifetime conversion count inside a single week. From 5 August the same campaign runs at a handful a day, like everything else.

Three-quarters of the month's conversions were booked before the fifth of the month. The curve is almost flat afterwards — which is what a normal month looks like.

fastrrv3 carries 75.8% of total sales and a 23.8% reversal rate — 3.6× the native Online Store. That one channel accounts for ₹2.31 Cr of the ₹2.41 Cr reversed. This is a checkout-quality question, not a marketing one.

Every line above 40% is a deity or animal pendant — Hanuman, Ganesha, Shiva, Tirupati Balaji, tiger, lion, elephant. They also top Shopify's sell-through at 100%: they sell fast and come back. That is a low-consideration cash-on-delivery impulse buy, not a jewellery purchase. The openable kadas that carry the business reverse at 7–15%.

Bottom-left is where you want to be: cheap and big. PMax Store and Brand_3june sit there. Search | Brand sits alone at the top — small and expensive. The two flagged campaigns are plotted on defective conversion data and are marked accordingly.

Gross sales peaked at ₹1.22 Cr in February and have not returned to it. August's ₹1.07 Cr is the second-best month of the year — bought with more than double February's media spend.

August at 9.82× is the floor of the year, against a 19.0× average and a prior low of 13.65× in November. Google-only — Meta spend was not supplied, so true blended MER sits below every bar shown. Read the shape, not the level.

Trailing twelve months ₹71.22 L against ₹8.18 L in the twelve before — an 8.7× increase in annual media investment. August alone is larger than the account's entire first fifteen months combined.

Two campaigns behave completely differently from the rest. Brand_3june wastes 11% of its search spend and PMax Store wastes 22% — against 86–94% everywhere else. Those are the two campaigns this report has been recommending all along, and the search data reaches the same verdict independently.

Bar length is spend, the number is CPA. Local and brand sit at ₹56 and ₹166 on 8% of the money. Core category — kundan, polki, choker, kada — takes ₹1.71 L at ₹440 with 92% of it converting nothing. Watches and apparel are the only genuinely off-catalogue themes, and together they are ₹10,124.

The remaining ₹4.4 L of zero-conversion spend is not a negative-keyword problem — it is a feed and structure problem, and Section 09 is where it gets fixed. Negatives alone will not touch it.

caveat Google names ₹5.64 L of the ₹9.25 L account. A further ₹2.53 L sits in "Other search terms" — queries below the privacy threshold, carrying 872 conversions. A quarter of the account's demand cannot be inspected at all.

Checkout tracking broke in February 2026 and has only just recovered. Tracked checkout sessions fell from 41.9% of real orders in December to 2.9% in February, sat between 2% and 4% until May, and climbed back to 57.0% in August — the best coverage the store has ever had. Orders across that window barely moved (2,060 → 1,970 → 1,815 → 1,642), so this was a measurement failure, not a demand collapse.

What it means for this review. Any conversion-rate trend quoted between February and May is meaningless. And August's headline improvement in checkout intent — reached-checkout up 136% on sessions up 53% — is confounded with the recovery in tracking itself. Both moved in the same month. Without the date the fix was deployed the two cannot be separated, so this report does not claim the traffic got better. It claims we finally started seeing it.

Shopify reports a 0.01% conversion rate and flags a warning on it. Do not let anyone quote that number. Its native funnel counted 68 completed checkouts against 1,960 actual orders, because 96.5% of orders complete in third-party checkouts that emit no events. The true rate is 0.466% — and it fell 24% as the scaled traffic arrived.

Below every listed rival on top-of-page rate and under 10% impression share. The search terms show the other half: MR Jewels wins ₹9,777 of competitor-brand traffic at ₹87 a conversion while its own brand term in its own postcode costs ₹117 a click and returns nothing. source conflict Three exports give 61.82%, 61.89% and 54.53% top-of-page; none is picked here.

Two peaks a day — a midday working window and a 9–11PM evening window that is the strongest block of the week. The 3–5AM trough carries 2.4% of impressions. Wednesday runs 40% ahead of Monday.

Mobile is 93.5% of ad spend and 96.3% of conversions — and 95.1% of the store's 3.56 million sessions. Desktop is 4.5%. This is a phone business end to end. TV screens cost ₹51.32 a click and returned nothing — exclude.

Karnataka is 9.6% of Indian sessions. Bengaluru alone is 9.0%. Maharashtra sends 1.5× Karnataka's traffic; Hyderabad, Mumbai and Delhi together send 2.6× Bengaluru. The showroom serves one city out of a demand base spread across every metro — and 7.4% of sessions come from outside India entirely, led by 97,561 from the United States.

Bars are on separate scales because the ranges differ by four orders of magnitude — read the mismatch, not the heights. Google paid is the widest session bar and one of the shortest order bars in the account.

Only 708 of 21,015 orders carry a referring channel. The other 96.6% arrive with no referrer, because fastrrv3 and the other headless checkouts strip it. This table is a sample, and a biased one — it over-represents whatever passes a referrer cleanly.

So the honest reading is not "Google paid produced 26 orders". It is: among the orders Shopify can see, paid traffic converts an order of magnitude worse than organic, and Google paid is the worst of them. That is a signal to investigate, not a verdict to act on alone.

It is also consistent with two independent findings in this report: the 2.99× incremental ROAS from Section 02, and the 88.5% of search spend that converts nothing from Section 10. Three different methods, all pointing the same way.

Four in five customers are new. Repeat purchase peaks at 12.6% in month one and falls to about 1% by month six. This is an acquisition treadmill, not a retention business — which is precisely why the 19.8% reversal rate and the cost of acquiring a first order matter more here than in a subscription model. Every rupee of reversal is a customer acquired and lost in the same transaction.

kwikchat converts 55× better than Google paid on a twentieth of the traffic. A chat widget is out-selling the entire paid search budget in the attributable sample. Worth an hour of somebody's attention. Separately: 33 referring channels with more than 100 sessions produced zero orders across 1,47,519 sessions.

Meta drives 2.26 million sessions — more than Google paid — and ₹35.7 L of attributable sales. Without the spend figure none of it can be judged, and the blended MER in Section 07 stays Google-only. This is now the single most valuable missing number after gross margin.

₹94,273 of new campaign spend — 10.4% of the month — none of which existed in July.

The three new store groups all landed at ₹20–29 CPA. The two inherited groups sit at ₹287–367.

One of the four scale decisions was sound. S_Tofu absorbed a 513% click increase and its CPC fell — that campaign has genuine headroom. The other three were steered by numbers that Section 03 has now disproved.

Hold spend flat at ₹9.0–9.5 lakh. August proved the account can absorb money, not that it should. Fix mix and reversals, then scale into October and November — ₹11–13 L and ₹13–15 L respectively estimate. At 2.99× incremental gross, October's extra ₹3 L should return roughly ₹9 L gross — but only if the reversal rate does not climb with volume, which is exactly what it did in August.

Reversals cost more than advertising. The 146× ROAS is fiction; the real number is 2.7–3.0× gross. Spend rose 143% and sales rose 22%. The ads are pointed at pendants that come back instead of kadas that stay sold, and 88.5% of named search spend converts nothing. Four in five customers are new and Shopify can name the source of only 3.4% of orders — an acquisition treadmill running half-blind. And two numbers still decide everything: gross margin, and Meta spend.