How QRCodeStack cut Google Ads waste by 40% with Admiral

QRCodeStack sells dynamic QR codes — the kind you print once and repoint later — to restaurants, event organisers and packaging teams, mostly in India. They had been running Google Ads for about two years when they connected the account to Admiral: ₹7,97,534 of lifetime spend, 17,537 clicks, 739 provider-reported conversions, or ₹1,079 per reported conversion over the whole history. That last label matters: the platform total mixed real outcomes with a page-view goal.

The recent months were worse than the lifetime average, and nobody could say why. That is the usual shape: an account that worked once drifts, and the monthly report keeps reporting.

What the first sync found

Admiral’s audit runs at connect, on the account’s own 90 days. On this one it filed nine findings. Four carried most of the money.

1. ₹19,640 on search terms that had never converted. “qr code generator free”, “qr code scanner app”, “what is a qr code” and 140 cousins — people who wanted a free tool or an explanation, matched in by a phrase keyword nobody had revisited since 2024. The single biggest line, and the easiest.
2. Search Partners and Display were still ticked on two of three campaigns — 9% of spend, one conversion between them in 90 days.
3. Three primary conversion actions, one of them a page view. Every visit to /pricing counted as a conversion, so Smart Bidding had been optimising for people who looked at the price. We discarded that blended platform total for the before-and-after below and reconciled both periods to the same backend definition: an account became activated after its codes crossed 50 scans in 14 days.
4. The conversion tag was live but invisible to the audit until Admiral was given the account’s own conversion id — it fires from Tag Manager, and a scanner without the id reads a GTM page as “can’t tell”. Once supplied, it read “found” and the finding closed. Worth saying, because the tag check is the one most tools get wrong on GTM sites.

Priced together, the leaks came to ₹34,800 of the ₹87,049 spent in the 30 days before the audit — 40% of the month, to the rupee, going somewhere it could not produce a customer.

What was approved

Nothing changed on the day. Every finding sits in the approval queue as a proposal with the evidence beside it; QRCodeStack’s founder approved five of nine over two evenings and declined one — a keyword Admiral flagged as low Quality Score that he wanted to keep for brand reasons. That is how it should go. The declined one is still in the log, marked declined, with his reason.

The five approved: 143 search terms added as negatives across the three campaigns; Search Partners and Display unticked; the page-view goal demoted from primary to secondary; a location setting moved from “presence or interest” to “presence”; and a budget shift from the weakest campaign to the strongest. Each one is a receipt — the change, what it cost or saved, and a one-click undo with the reverter stored before the change ran.

Four weeks on

Same monthly budget and 3% less spend, with 62% more verified activations; cost per verified activation fell 40%. Both the “53” and the “86” use the same backend rule, so this comparison does not inherit the old page-view inflation.

What this is not

It is not a promise that your account is leaking 40%. It is one account, and a well-run one would have found less. The 40% is what two years of nobody reading the search terms report looks like, and it is common. If you want to know your own number, the audit is free and runs before you pay anything.

One more thing, for the compliance-minded: every change above is also recorded on a hash-chained log outside Admiral, in Auditant — the same product our own agents run on. An auditor can verify the sequence offline without an account with either of us. That is what “receipt” means here.