Landing pages for paid search

Sending paid traffic to your homepage is the most expensive habit in performance marketing, and it is almost universal.

The logic behind it is understandable. The homepage is the best-designed page on the site. It has been through the most rounds of review. It represents the whole business. Why would you send valuable traffic anywhere else?

Because the homepage is built for a visitor who arrived without a specific question. Paid traffic arrives with one. Someone who searched "marriage garden in garhwa with parking for 200" has told you precisely what they want, and a homepage answers by presenting twelve options and asking them to look for it.

The arithmetic is worth sitting with. A dedicated page will commonly convert at two to three times the rate of a homepage on the same traffic. That is not a marginal gain — it halves your cost per acquisition without touching a single bid, changing a single keyword, or spending an additional rupee on media.

There is a second reason, less obvious. Landing page experience is a component of Quality Score, and Quality Score affects both your ad rank and what you pay per click. A better-matched page does not only convert more of the traffic you buy; it makes the traffic cheaper to buy.

The test

For each of your top five campaigns by spend, write down the query someone typed and the page they landed on. Then read that page as if you were them, in the ten seconds they will actually give it.

Does it say the thing they searched for? Does it tell them what it costs, or what determines the cost? Does it make the next step obvious?

If the answer to any of those is no, you have found work worth more than anything happening inside the ad account.

The first job of the page is to confirm the visitor has arrived somewhere relevant. This has to happen fast — before scrolling, before reading, in the first second.

The mechanism is simple and widely ignored: use their words, not yours.

If they searched "banquet hall booking in south delhi", the headline should contain something very close to "banquet halls in south delhi", not "Premium Event Solutions for Discerning Clients". Your internal product name, your tagline and your positioning statement are all things you understand and they do not.

This is why intent-tiered ad groups matter beyond the ad account: an ad group organised around one theme can point at a page written for that theme. An ad group containing forty unrelated keywords cannot.

The three-part match

Message match is a chain and it breaks at the weakest link:

1. Query — what they typed
2. Ad — the headline they clicked
3. Page — the headline they land on

All three should carry the same core phrase. A visitor who clicked an ad promising "Same-day AC repair in Noida" and lands on a page headed "Our Services" has been, from their point of view, mildly misled — and they will leave.

Dynamic insertion, and its limits

Dynamic text replacement — swapping the page headline based on a URL parameter — lets one page serve many ad groups. It works, and for accounts with many similar variations it saves a great deal of build time.

It has a failure mode worth knowing. Because the parameter can be edited, the technique has been used to fabricate convincing screenshots of pages that never existed as published. Keep the substitution to a controlled list of values you have written, rather than echoing arbitrary URL content into the page. That is both a trust matter and a straightforward security one — echoing unsanitised parameters into a page is how cross-site scripting gets in.

A landing page makes an argument. Arguments have an order, and the order is not the one most pages use.

The order that works:

1. Am I in the right place? The headline, echoing their search. One sentence of clarification beneath it.

2. What exactly is on offer? Concrete and specific. "Banquet hall for 150–400 guests, in-house catering, parking for 80 cars" beats "world-class event experiences" in every measurable way.

3. What does it cost — or what determines the cost? This is the one businesses resist hardest, and it is the highest-leverage element on most pages.

The objection is always the same: pricing depends on too many variables, and publishing a number will lose deals. Both may be true. But the visitor has a budget question they cannot avoid, and if the page will not engage with it, a meaningful share of them leave to find someone who will.

You do not have to publish a price list. You have to engage with the question. "Most weddings here run between ₹1,200 and ₹2,000 per plate depending on menu and season" tells someone whether to continue, filters out people who were never going to buy, and costs you nothing but the enquiries you did not want.

4. Why should I believe you? Proof — and specific beats voluminous. One checkable fact does more than a wall of logos: a named certification with its ID, a real number, a photograph of the actual venue rather than a stock image of a venue.

The reason is straightforward. Generic claims are free, so they carry no information. "Trusted by hundreds of clients" could be written by anyone about anything. A certification number can be looked up.

5. What happens if I fill this in? Say it explicitly. "We will call you within one working day with availability and a quote" outperforms "Submit", because it removes the fear of what the form triggers. People do not fear giving you their number. They fear being called eleven times by a call centre.

Every field costs you completions. That is well established and it leads people to a conclusion that is only half right: make the form as short as possible.

The full picture is that field count trades volume against qualification, and which side of that trade you want depends on what happens next.

If you follow up every lead personally and your time is the constraint, more fields are correct. A form asking for budget range and timeline will produce fewer, better leads, and if you can only work fifteen enquiries a week you would rather they were the right fifteen.

If you have capacity to qualify on the phone, fewer fields are correct. Name and phone number, qualify in the call.

What is never correct is asking for information you do not use. Company size, job title, "how did you hear about us" — each costs completions and if nobody reads the answer, it is pure loss.

Practical rules

• Ask for the phone number or the email, not both, unless you genuinely need both to make contact. This single change lifts completion rates noticeably.
• Never make a field required unless you cannot proceed without it. An optional field that most people fill in is strictly better than a required one that some people abandon.
• Validate helpfully. "That doesn't look like a valid phone number — please include the area code" beats a red border with no explanation.
• Do not use a CAPTCHA unless you have an actual spam problem. They are a meaningful conversion tax and a genuine accessibility barrier. A honeypot field plus a timing check catches most automated submissions invisibly.
• Multi-step forms can outperform single long ones, because the first step is small and commitment builds. They also let you capture partial data. Worth testing rather than assuming.

The question that decides the design

Not "how do we get more leads?" but "what is the smallest amount of information that lets us have a useful first conversation?" Ask for that. Nothing more.

Paid traffic is less patient than organic traffic. Organic visitors chose you from a results page and have some investment in the outcome. Paid visitors clicked an ad, often on mobile, often on a poor connection, and have no loyalty whatsoever. A slow page loses them at a higher rate than the same page loses organic visitors.

Core Web Vitals give you the targets:

• Largest Contentful Paint (LCP) — under 2.5 seconds. When the main content appears.
• Interaction to Next Paint (INP) — under 200 milliseconds. How quickly the page responds when touched. This replaced First Input Delay in March 2024 and is stricter.
• Cumulative Layout Shift (CLS) — under 0.1. How much the page jumps as it loads.

CLS deserves particular attention on landing pages, because the failure it describes is somebody tapping a button that moves as they tap it. On a page whose entire purpose is one action, that is not a cosmetic problem.

Where the time usually goes

In rough order of how often each is the culprit:

• Images that are far larger than displayed. A 3MB hero image scaled to 800px wide. Serve appropriately-sized images in a modern format.
• Third-party scripts. Chat widgets, heatmap tools, analytics, tag managers, ad pixels. Each is defensible and collectively they are usually the largest cost. Audit what is actually loading.
• Web fonts. A page that renders invisible text while waiting for a font has an LCP problem that is entirely self-inflicted.
• Unused CSS and JavaScript, particularly from page builders that ship the whole framework regardless of what the page uses.

Measure with real-world data — Chrome User Experience Report data in PageSpeed Insights, or your own field measurements — rather than lab scores alone. A lab test on a fast connection will tell you the page is fine when your actual visitors on mobile data are waiting six seconds.

Most landing page tests are worthless, not because testing does not work but because they are run in ways that cannot produce a reliable answer.

The three ways tests go wrong

Stopping when it looks good. Watching a test daily and stopping the moment significance appears is the most common error in the discipline. Random variation will produce apparent significance at some point in almost any test if you keep checking. Decide the sample size and duration before you start, then do not look at the result until you get there.

Testing things too small to matter. Button colour tests are popular because they are easy. They almost never produce a change large enough to detect without enormous traffic. Test the offer, the headline, the price presentation, the form length, the page structure. Test things that could plausibly move the number by 20%, not 2%.

Not enough traffic to conclude anything. This is the constraint that quietly invalidates most small-business testing. Detecting a 20% relative improvement on a 3% baseline conversion rate needs roughly a thousand visitors per variant. Detecting a 5% improvement needs tens of thousands.

If your landing page receives 400 visitors a month, you cannot A/B test it meaningfully. That is not a failure of will — it is arithmetic. Use a sample size calculator before running a test, and if the required sample exceeds what you can gather in a month, do not run the test.

What to do instead when traffic is low

Low-traffic pages are not un-improvable. They are un-testable, which is different.

• Make larger changes and judge over longer periods. A rebuilt page compared month over month is crude, confounded by seasonality, and still more informative than an underpowered A/B test.
• Watch session recordings. Twenty recordings of real visitors will tell you more about where a page fails than a statistically meaningless test. You are looking for hesitation, rage clicks, scroll-backs, form abandonment mid-field.
• Ask the people who convert. "What nearly stopped you getting in touch?" is one question in the follow-up call and it routinely surfaces things no analytics tool would.
• Fix the things that are obviously wrong. Message match, form length, a missing price. You do not need a test to justify these; you need them done.

Heatmaps, honestly

Heatmaps are good at showing where attention goes and bad at explaining why. A cold area below the fold might mean the content is uninteresting or that the section above it answered the question and the visitor converted. Treat them as a source of hypotheses, never conclusions.

When a page underperforms, work through this in order. As with the account itself, the order matters — fixing a lower item while a higher one is broken means doing it twice.

1. Is the traffic right? Before blaming the page, check the search terms driving it. A page converting at 0.5% may be a fine page receiving irrelevant traffic. This is a structure problem masquerading as a page problem, and it is the most common misdiagnosis there is.

2. Does the message match? Query, ad and page headline carrying the same core phrase.

3. Is the offer clear and is the price question engaged with? Most pages fail here.

4. Is the form asking too much? Count the fields. Justify each one.

5. Is the proof specific? Replace one generic claim with one checkable fact.

6. Is it fast, particularly on mobile? Measure with field data.

7. Only now, test. And only if you have the traffic to conclude anything.

The reason for this order is that steps 1 through 6 are things you can reason about and fix in an afternoon, with high confidence that they are improvements. Step 7 is expensive, slow, and only worth doing once the obvious work is done.

Most pages never needed a test. They needed somebody to read them as a stranger would, notice that the price question goes unanswered and the form asks for a company name nobody uses, and fix those two things.