Google Ads can tell you which campaign produced the most leads. It cannot tell you which campaign produced the most customers, because the moment a lead becomes a customer happens in your CRM, which Google cannot see. The same is true of the city, the device, the phone make and the hour of day.
This CRM holds both halves. Marketing → Audience Intelligence puts them together: it takes the leads that closed, compares them against the ones that did not, and tells you what the winners had in common.
What it actually compares
Every lead that arrives through a landing page carrying your tracking script brings a description of the visit with it — where in the world the person was, what they were using, how they were connected, how long they spent reading, what time it was where they were. None of that requires a permission prompt or a cookie banner; it is what every browser already tells every website it visits.
The screen groups your leads by each of those, then shows one number per group that matters more than the rest: the win rate against your own average. Not against an industry benchmark, which would be meaningless for your business — against the rate you close at.
Reading the lift figure
A segment reading ×2.1 better means leads from it close at a little over twice your overall rate. That is the number to act on, and it is expressed as a multiple rather than as percentage points on purpose: "+18 points" means something completely different to a business closing 4% and one closing 40%, and would be read wrongly by at least one of them.
Why some segments show no verdict at all. A segment needs at least 8 leads before it is ranked, and a "stop spending here" needs 25. Below that it is listed with its real numbers under "too early to judge" and nothing else. This is deliberate. With three leads and one win, the win rate is 33% and it means nothing — but anyone who reads "Nashik converts 3x better" will move budget on it, lose money, and never trust a number on the screen again. Being told there is not enough data yet is the feature.
The four things to do with it
- Location targeting. The fastest lever in most Indian accounts. Cities differ from each other far more than devices or hours do, and Google Ads lets you bid on them separately within the same campaign.
- Device bid adjustments. If desktop closes at twice the rate of mobile and you are paying the same for both, you are funding the wrong half. Both Google and Meta let you adjust this in campaign settings.
- Ad scheduling. The hours shown are in the visitor's timezone, not yours, because that is how ad schedules are set. "Our best leads come in at 9pm" is useless advice if the 9pm belonged to someone in Dubai.
- Landing page weight. If most of your customers arrive on mobile data rather than wi-fi, page weight is a conversion problem, not a technical nicety.
“Your actual customer” — and why it is not your traffic
The profile block shows the most common answer among the deals you won, not among all your leads. The gap between those two is often the most valuable thing on the page.
A typical example: most leads arrive on budget Android phones, and most customers turn out to be on iPhones or desktops. Both facts are true. The first describes the traffic you are buying; the second describes the people who pay you. When they disagree, the traffic is the thing to change.
What it needs from you
- Mark leads Won. Everything here is built on that one action. Nothing can be compared until deals are being closed in the CRM rather than in somebody's head. Below three won deals in the window, the screen says so and offers no recommendations at all.
- Record deal values. Win rate alone treats a ₹8,000 sale and a ₹2,00,000 sale as the same event. With values recorded, the revenue column often tells a different story from the win-rate column — and the revenue one is usually the one to follow.
- Use the landing page script. Leads typed in by hand or pushed through the API have no visit behind them, so they are excluded. The screen tells you what share of your leads it is working from rather than quietly reporting on a third of them as though it were everything.
Location detail, and the honest limits of it
City-level reporting depends on resolving the visitor's IP address, which your installation's administrator switches on. With it off, everything else on the screen still works — device, operating system, browser, connection, language, hour and behaviour all come from the browser itself — and location falls back to country, from the visitor's timezone.
Where coordinates are shown they are the centre of a city, never a precise position. We do not ask browsers for GPS access: that fires a permission prompt on your landing page, in front of a stranger who came to fill in a form, and it costs you conversions to learn something the IP already approximates well enough to choose a city to advertise in.
One honest caution
This screen tells you what correlated with closing, not what caused it. If your best city is also where your best salesperson lives, the city is not the reason. Treat a strong segment as a hypothesis worth funding and then check the next month's numbers — which is exactly what the window selector at the top is for.