Method
Most reporting hands you a platform dashboard taking credit for itself. This is the method we use instead — what we instrument, how the range is built, where it fails, and what it cannot see.
A single number implies a precision the data does not support. Ad platforms each claim the same conversion, cookies expire, WhatsApp conversations happen off-site, and phone calls leave no click.
So we report a floor and a ceiling, and we tell you which assumptions move each end.
A report that says "₹4.2 lakh of revenue, attributed" is easier to sell and impossible to defend. One that says "between ₹3.1 and ₹4.6 lakh, and here is what widens it" survives contact with your accounts.
The instruments
Status is what is running in production today, not what is planned. Two of these are built and switched off, and they say so.
Every link we publish carries a normalised source, medium and campaign, written to one scheme rather than whatever each platform defaults to.
Without one scheme, the same campaign appears as four rows and no total is correct. This is unglamorous and it is the foundation everything below stands on.
Short links resolve through our own redirect, which tallies the click server-side before forwarding — no personal data, no cookie required, normalised before it is stored.
Click counts that depend on a browser executing script under-count on exactly the devices Indian buyers use most. Counting at the redirect does not care about ad blockers, tracker blocking, or a locked-down browser.
A WhatsApp click is recorded against the page and location it came from, tallied daily, with no message content and no phone number stored.
In India this is frequently the highest-intent action on the site, and in most agency reporting it is invisible — the visitor leaves for another app and the session simply ends.
Conversions sent from our server rather than the browser, so that a blocked pixel does not silently delete a real sale from the record.
Built and switched off. It needs platform credentials and the open data-protection review to close first. Until then, browser-side reporting is what it is, and the floor of the range reflects that.
One question after checkout — how the customer first heard of you — captured as stated attribution alongside the measured kind.
It is the only instrument here that sees channels no tracking can: a recommendation, a WhatsApp group, an offline conversation. Blocked because it fires after an order, and orders cannot complete until payments are configured.
The range
Two numbers and the distance between them. The distance is the honest part.
01
Only conversions we can trace end to end: a tagged click, our own recorded redirect, and an enquiry or order that followed it. Nothing inferred. This number is defensible in front of a finance team and it is always too low.
02
The floor plus conversions that plausibly belong to the channel but cannot be proven — an untagged direct visit shortly after a campaign, a branded search that did not exist before the campaign ran, a WhatsApp enquiry from a page the campaign drove traffic to.
03
Reported, not hidden. A wide gap is information: it usually means a channel is doing work the instrumentation cannot see, and it tells you where instrumenting next is worth the money.
The limits
Every agency will tell you what it measures. This is the half that is usually left out, and it is the half worth reading before you engage anyone.
Growth intelligence
Measurement is only worth something if you can read it and act on it.
This is the shape of the report: a floor and a ceiling for every channel, and an honest row for what could not be attributed at all.
| Channel | Range | Relative scale |
|---|---|---|
| Google AdsTraced clicks at the floor; assisted paths at the ceiling | 18 – 31 | |
| Meta AdsWidest range here — browser-side reporting is the weakest link | 9 – 22 | |
| WhatsAppCounted at our own redirect, so the range is narrow | 11 – 14 | |
| Direct and organicIncludes anyone who heard of you somewhere we cannot see | 6 – 17 | |
| Not attributedThe residual. Reported, never redistributed to flatter a channel | 8 – 24 |
Illustrative example of the report format — not a client's data
What the range is for
01
Measure
Spend, clicks and enquiries joined to one record.
02
Read the range
Where the floor and the ceiling disagree is where to look.
03
Change one thing
One change at a time, or the next reading explains nothing.
04
Check it moved
Against the same method, not a new one.
Failure modes
These are not hypothetical. Each one has a routine behind it, because a method is only as good as what it does on a bad week.
Version 1.0 · Updated 2 September 2026
This page is versioned deliberately. When the method changes — a new signal, a retired assumption, a failure mode we did not anticipate — the version changes with it and the previous statement is not quietly replaced.
