Watch what happens on a mid range Android phone on 4G, which is how most of your traffic actually arrives.
Second zero, they tap your ad. Second one, white screen. Second two, still white, and a hero image the size of a small film is on its way. Second three, the header appears. Second four, they are back in Instagram.
You paid for that click. They never saw the page.
Around 53% of mobile users abandon a site that takes more than three seconds to show something, and every additional second costs roughly 7% of your conversions. In 2026 mobile pages load about 22% slower than desktop ones, 3.1 seconds against 2.5, which means the version most people see is the slow one and the version you check on your laptop is not.
I keep telling people in other posts to fix conversion before buying more traffic. This is the post where I say what that actually means.
What is on the table
The median dedicated landing page converts at about 4.02%. The top quarter reach 11.45% or better.
Same traffic, same auction, nearly three times the customers.
And a general page on your website converts around 2.35%, a bit over half what a dedicated page does. Which produces the first and most common own goal: sending paid traffic to your homepage. A homepage is built to serve everybody, so it persuades nobody in particular, and the person who clicked an ad about one specific service arrives at a menu.
One ad, one promise, one page. If you run three services, you need three pages.
Speed, because in India it is the binding constraint
Most CRO advice treats speed as a technical footnote. Here it is the main event, because the median visit is a mid range phone on a mobile network in a moving vehicle, not a MacBook on office wifi.
Three things cause nearly all of it.
Images nobody compressed. A photographer sends 4MB JPEGs, somebody uploads them as they are, and the page now weighs more than the app store listing for a game. Resize to the dimensions actually displayed, convert to WebP, and you will usually halve the page on your own.
Scripts loading before content. Analytics, a chat widget, a heatmap tool, two pixels and a font, all fetched before anybody sees a word. Move what you can after the content renders and delete what nobody has opened in six months.
A hero video. It is the single most expensive decoration on the Indian web and I have never seen one earn its weight.
Test it properly: your phone, mobile data, not the office connection, and ideally a handset that is two or three years old. If the page is usable in under two seconds on that, you are ahead of nearly everybody in your category.
Match what the ad promised
The visitor’s first moment is not spent reading. It is spent checking they are in the right place.
So if the ad said “same day AC repair in Koramangala” and the page headline says “Excellence in climate solutions since 2011”, you have introduced doubt at the exact moment you could least afford it. They will not puzzle it out. They will go back.
The fix is boring and it works. The H1 repeats the promise from the ad, close to word for word. Then the proof. Then the action.
If you cannot write a headline that matches the ad, the problem is the ad.
Cut the form until it hurts
The numbers here are stark enough that I will just give them to you.
Three field forms convert at around 10.1%. Nine field forms drop to about 3.6%. Getting down to five fields or fewer roughly doubles conversion, and each field past five costs something like 20% to 30%.
So the rule is not “keep it short”. It is: ask only for what you need to have the next conversation, and get the rest on the call.
For most Indian service businesses that is a name and a phone number. Possibly the city. That is it.
Email is frequently the field that quietly damages you, because people give you a fake one rather than abandon, and now your data is wrong as well as your conversion rate being lower. Budget dropdowns, “how did you hear about us”, company size, job title: every one of those is a question you could ask a human being who is already talking to you.
The CTA is probably not a form at all
This is where imported playbooks fail hardest in India.
A large share of your visitors do not want to fill in anything. They want to press a green button and type a message. So put WhatsApp where a thumb naturally rests, which is the bottom third of the screen, as a sticky bar that stays visible while they scroll.
Tap to call next to it, because the other half will ring you.
Three things to get right. The number has to be one somebody actually monitors, the WhatsApp link should carry a prefilled message so you know which page it came from, and somebody has to reply within minutes rather than the next morning.
One warning. These enquiries are the easiest to lose track of, because they happen in an app rather than on your website. That is the tracking hole worth closing before you start judging which pages work.
Proof, above the fold, of the kind people here believe
Not a stock photo of a handshake.
A Google rating with the review count. A named customer in the same city, ideally one the visitor might recognise. A number from real work: jobs completed, years running, response time you actually hit. A photograph of the actual team or the actual premises.
The specific beats the impressive. “Trusted by leading brands across India” persuades nobody. “148 installations in Whitefield since 2021, rated 4.7 on Google” is smaller and considerably more convincing.
Things not worth your morning
Button colour. It has been tested to death and the effect is noise next to everything above.
Carousels. Visitors see the first slide and nothing else, so you have hidden most of your page behind an animation.
Long copy versus short copy as a general question. The right length is however long it takes to answer the objection, and that is a question about your buyer rather than a rule.
Before lunch
Four things, about an hour.
Open your main landing page on your own phone, on mobile data, and count the seconds until you can read the headline. If it is over three, that is your entire morning and everything else can wait.
Then compare your top ad headline against the page H1, side by side. Make them match.
Count your form fields. Delete every one you could ask about on the phone instead.
Then add a sticky WhatsApp and call bar to the bottom of the page.
None of this costs media budget, and all of it multiplies whatever you already spend. Which is why it comes before working out how much to spend, not after.
My notes on performance marketing cover the traffic side once the page is worth sending people to. If you’d like a second pair of eyes on your own page, drop me a line on email, WhatsApp or LinkedIn and we can have a quick chat. I’m contracted full time so this isn’t a pitch.