In this article
Start with a definition and a baseline you can trust
A conversion is a visitor who did the thing the page exists for: called, sent a WhatsApp, filled the form, booked, requested a quote. Decide which of those count, and count all of them — a site judged only on form fills is being judged on a tenth of its enquiries in most Indian businesses, where the phone does the work. That means call tracking or a site-only number, a WhatsApp link that is measured, and a form that records its source.
Then the baseline: for the last month, enquiries per hundred visitors, by landing page and by traffic source. Brand searches convert well and tell you little; visitors from a relevant search or an ad landing on a service page are the ones whose rate you are trying to move. If the analytics cannot give you this, the first fortnight of the project is making it do so. Improvement claimed without a baseline is a feeling.
Message match, proof, forms and calls to action
Message match: the page must say, in its first screen, what the visitor came for. A visitor who searched “laser cutting Bengaluru” and lands on a home page about “engineering solutions” leaves; one who lands on a laser-cutting page with the headline saying so stays. Check every landing page against the search or ad that feeds it, and rewrite the first screen to match. Proof: near the point of decision, not at the bottom — a named customer, a review with a name, a photograph of the real work, the certification that matters in your trade. Forms: name, phone, one line about the need; every extra field costs visitors, and the form must say what happens next. Calls to action: one primary, in words that describe what the visitor gets — “Get a quote today”, “Book the free audit” — repeated where the page has just made its argument, and a tappable number and WhatsApp for those who will not type.
These four account for most of the gap between a one-percent page and a four-percent page, and none of them needs a redesign.
Run experiments in priority order
List the changes, rank them by how many visitors they touch and how cheap they are, and make them one at a time, giving each two weeks or a few hundred visitors before judging. Speed and the visible number first, because they touch everyone. Then the first-screen message on the pages with the most relevant traffic. Then proof placement, then the form, then the call to action wording. Formal split-testing tools are for sites with thousands of visitors a week; a small business compares this fortnight with the last, on the same page and source, and keeps what worked.
Judge each change on enquiry quality as well as count: a shorter form that doubles enquiries but halves the share that become customers has moved the work to the sales team. Track enquiries that became quotes, and quotes that became orders, alongside the conversion rate.
A sequence that works for most SMEs
Week one: measurement — call tracking, the WhatsApp link, form sources, the baseline by page. Week two: speed on a phone, the number at the top, the WhatsApp link everywhere. Weeks three and four: the first screen of the three landing pages that carry the most relevant traffic, rewritten to match what brought the visitor. Weeks five and six: proof moved beside the decision, and the form cut to three fields with a “what happens next” line. Then a fortnight of watching, and the next change.
Owner, marketer or agency, the responsibilities are the same: someone measures, someone writes, someone changes the page, and someone decides at the end of each fortnight. In most small businesses that is two people.
Worksheet · use it here or print it
Conversion experiment backlog
One experiment at a time, each written before it runs. Fill the baseline first — without it, every result is an opinion.
Order the backlog by message match, proof, form and call to action — in that order — because that is the order in which changes usually pay.
What you enter stays in this browser and is not sent to us.
Mistakes, and when it is not the website
The mistakes: measuring form fills only; changing five things at once; copying a competitor’s page instead of matching your own traffic; adding a pop-up or a chatbot, which usually lowers the rate; and optimising a page whose visitors are the wrong people, which no page can convert. A safeguard: if brand-search visitors convert well and everyone else does not, the problem is upstream — the ad, the search terms, the listing — not the page.
Measure the programme by enquiries per hundred relevant visitors, enquiry-to-quote rate, and cost per enquiry, month on month. This is conversion-rate optimisation as we do it for small businesses: the measurement first, the four fixes, and the fortnightly changes with their numbers — and the free audit will tell you, from your own analytics, which of the four is losing you the most.
Questions owners ask
What is a good conversion rate for a small-business website?
Between one and three enquiries per hundred visitors from relevant sources is typical for a service business; a page with clear message match and proof can do better. Compare your own pages against each other and against last month, not against an industry figure.
Do I need an A/B testing tool?
Not below a few thousand visitors a week. Compare fortnights on the same page and source. Tools are for deciding between two versions at scale; a small business needs to find and fix obvious faults first.
How long before results show?
Speed and the visible phone number show within days. Message and proof changes show in a fortnight. Give the whole sequence two months before judging it.
Will a shorter form bring worse enquiries?
It can, which is why enquiry quality is tracked alongside count. Usually a three-field form with a good “what happens next” line brings more and equally good enquiries, and qualification moves to the first call.
Should we redesign instead?
Only if the structure cannot carry the audience. Most conversion problems are the first screen, the proof, the form and the number — fixable on the existing site in weeks.
What does GullySales do?
The measurement setup, the baseline, the four fixes written and made, the fortnightly change programme with its numbers, and a monthly review. Scoped in the free audit and priced in writing.