Conversion rate optimisation

'Conversion rate optimization' draws 301,000 searches a month in India at difficulty 61 — one of the largest terms in the entire category, and mostly informational, which tells you how many people are trying to learn it rather than buy it.

The commercial logic is simple: doubling conversion rate has the same revenue effect as doubling traffic, and is usually faster and cheaper. On most sites we audit, the enquiry form is the single largest leak.

We do not start with a test. We start with evidence: session recordings, form-field drop-off, scroll and click behaviour, search console query-to-landing-page mismatch, and where possible five to ten minutes on the phone with recent enquirers asking what nearly stopped them.

Testing without that produces a long list of inconclusive experiments. Most Indian B2B sites do not have the traffic volume for statistically valid A/B testing anyway, which makes qualitative research the honest method rather than a compromise.

Forms asking for information the business does not need at that stage. Phone numbers that are not clickable on mobile. No indication of what happens after submitting. Pricing hidden when the buyer's first question is price. And page speed, which is a conversion problem as much as a ranking one.

The standard CRO playbook assumes ecommerce traffic. Split-test the headline, run it to significance, keep the winner. On a B2B site with a few hundred sessions a month and a conversion rate of two or three per cent, that method does not work, and it is worth being exact about why.

Detecting a 20 per cent improvement on a 2 per cent baseline takes on the order of several thousand sessions per variant before the result means anything. A site with 500 sessions a month would be waiting most of a year for one answer, by which time the market, the traffic mix and the page have all changed. Anyone promising A/B-tested gains at that traffic level is either running tests they stop as soon as they look good, or not running them at all.

So on sites below roughly 10,000 sessions a month we do not sell testing. We fix defects, we make the offer legible, and we read session recordings and enquiry transcripts — qualitative evidence that does not need statistical power to be valid.

Whether the form actually works. This is not a hypothetical. On this site in September, a spam honeypot field named “website” was being auto-filled by Chrome for real visitors — the form showed the success message, the analytics event fired, and the enquiry was silently discarded. Two lead events with no enquiry behind them is how we found it. We now log every screened submission instead of discarding it. We check this on every site we take on, because a broken form beats every conversion idea we could otherwise offer.

Whether the tracking is honest. A conversion that fires on page load, a thank-you page reachable by refresh, a lead counted twice by two systems. Optimising against a false number is worse than not optimising.

Friction that has no purpose. Required fields nobody needs to answer to start a conversation. Phone-number validation that rejects an international format. A file upload that fails silently on mobile.

And whether the page answers the question. Most B2B pages describe the service and never state what happens next, what it costs to start, or who is on the other end of the form.