Funnel experimentation and A/B testing from Gully Sales finds the steps where your enquiries leak away, turns the fixes into testable hypotheses, and runs them one at a time so you learn what worked and what only looked like it did.
Gully Sales Private Limited works with small and medium businesses across India. Every engagement begins with a free audit of your current funnel.
In one paragraph
Funnel experimentation and A/B testing from Gully Sales replaces opinion-led changes with evidence. We measure how many people pass each step of your funnel, find where the largest drop-offs sit, write hypotheses that explain them, and run controlled tests one at a time until you can say which change moved the number and which did not.
The problem
Your team is not careless about improvement. The homepage has been changed three times this year. The enquiry form was shortened because someone read that short forms convert more. The ad creative was swapped because sales said the leads felt weak. Every one of those changes was reasonable on the day it was made. The trouble is that they all happened in the same few weeks, traffic moved for its own reasons too, and now nobody can say which one helped. So the next argument about the funnel gets settled the way the last one was, by whoever speaks with the most conviction in the meeting.
Why it persists. Testing looks like a large-company activity. It sounds as though it needs heavy traffic, a specialist tool and a statistician, and most Indian SMBs have none of the three. So the funnel gets improved by redesign instead: everything changes at once, the result is judged by feel, and the exercise repeats a year later. The quieter truth is that a modest funnel can still be tested. It needs fewer tests, chosen more carefully, each one bold enough to show a difference.
If it stays unresolved. Spending grows faster than results, because the only lever anyone trusts is more traffic. Every rupee added at the top still has to pass through the same leaking steps. Over two or three years that becomes an acquisition cost your margins cannot carry, and a funnel nobody understands well enough to repair quickly when the market turns.
What changes
In the first weeks
In how the work runs
In sales and marketing
In what management can see
Over the longer term
We control the baseline, the hypotheses, the test design, the build and the honest reading of the result. Traffic volume, seasonality and market conditions decide how quickly a test can be called, and some tests will show no difference at all. A test that ends flat is still a result worth owning.
Who it is for
What Gully Sales does
Funnel baseline
We define the funnel in your own terms, step by step, from the first visit through to the enquiry, order or qualified conversation. Then we measure how many people reach each step and how many move on, over a period long enough to be representative rather than flattering.
Friction analysis
We work through each losing step and look for the reason people stop: a form asking for information too early, a page that answers a different question than the ad promised, a price that appears without context, a mobile layout that hides the action, a follow-up that arrives two days late.
Hypotheses and prioritisation
Each piece of friction becomes a written hypothesis in plain language: what we think is happening, what we would change, and what we expect that change to do to a specific number. Every hypothesis is then scored on the traffic it touches, the money behind it and the effort to build it.
Copy and UX improvements
We write the variant, not just the brief. That means rewritten headlines and body copy, reordered page sections, shortened or resequenced forms, clearer buttons, and layouts adjusted for the devices your visitors actually use.
Experiment design and running
For each test we decide what changes, who sees it, how the traffic is split, how long it runs and what result would count as a decision. We build it in your existing website and campaign tools, check it on desktop and mobile before launch, and watch it while it runs.
Conversion reporting
Every month you receive a report showing the funnel against its baseline, the tests that ran, what each one showed, the decision taken, and what is queued next. Flat and negative results are reported in the same detail as positive ones.
Implementation roadmap
Winning variants are made permanent, losing ones are removed cleanly, and everything learned is turned into a sequenced plan of what to change outright, what to test next and what to leave alone for now.
How it runs
Free audit and funnel definition
We walk your funnel as a customer does, from advertisement or search result through to enquiry or order. We agree what counts as a step, what counts as a conversion, and which parts of the journey are in scope for this work.
Baseline and tracking check
We measure volume and conversion at each step over a representative period, and check that the tracking behind those numbers is sound. Where a step is not measured or is measured twice, we say so plainly rather than reporting a figure we do not trust.
Friction analysis
We examine the steps losing the most people and identify why they lose them, using the funnel data, the pages themselves, what your sales team hears from enquirers, and how the journey behaves on a phone.
Hypotheses and prioritisation
Each cause becomes a written hypothesis with an expected effect on a specific number. We score every hypothesis on traffic touched, revenue at stake and effort to build, then sequence them into a backlog.
Variant build and launch
We write the copy, prepare the layout or form variant, set the split and duration, agree the decision rule, and build the test in your own tools. Everything is checked on desktop and mobile before any traffic reaches it.
Run and read
The test runs for the agreed period without interference. We monitor it for technical faults but not for early winners, then read it against the rule set beforehand and write down what it showed, including when it showed nothing.
Roadmap and handover
We consolidate what the round taught, update the backlog, sequence the changes worth making outright, and train your team or agency to design and read the next tests using the same standard.
Ways to work with us
Funnel diagnostic
A one-time baseline, friction analysis and prioritised test backlog. You receive the map and the ranked queue, and your own team or agency runs the experiments from there.
Test programme
We design, build and run an agreed number of experiments across a defined period, with a results note for each and a monthly conversion report showing the funnel against its baseline.
Embedded experimentation support
We work alongside your marketing team or agency, setting the standard for how tests are designed and read, and reviewing every result before a change is made permanent.
Launch testing
A new landing page, offer or form is baselined and tested around its launch, so the change is judged on evidence rather than on how the first week happened to feel.
Why Gully Sales
We test the whole funnel, not only the page.
Conversion is lost in the advertisement, on the landing page, inside the form and in the hours after the enquiry arrives. We test wherever the loss is largest, including the follow-up steps most testing work never touches.
The decision rule is agreed before the test runs.
How long it runs and what result would count as a win are written down before launch. That single discipline removes most of the arguments that follow a test, and most of the wrong conclusions.
We work with the traffic you actually have.
Indian SMB funnels rarely carry heavy volume. So we run fewer tests with bolder differences between versions, and we say honestly when a step does not have the traffic to settle a question.
You are not asked to buy an experimentation suite.
Tests are built in your existing website, forms, campaign platforms and CRM. You are not asked to buy an experimentation suite in order to answer questions your current stack can answer.
Flat results are reported as clearly as wins.
A test showing no difference saves you from an expensive redesign built on a hunch. We report it in the same detail, because knowing what does not move the number is worth what you paid for it.
We understand what happens after the click.
Gully Sales works across marketing, sales, customer success and revenue operations, so a funnel test can follow the enquiry into how quickly it is answered and whether it ever became a customer.
Where it applies
Industrial manufacturer
Multi-location clinic
Direct-to-consumer e-commerce
Professional services firm
Education and training institute
Proof
Kalessi Bathroom Tiles, e-commerce, Australia
Questions buyers ask
We agree the primary conversion with you before anything is measured. For most businesses it is a qualified enquiry, a booked appointment or a completed order, not a page view. The baseline is then measured from your own data over a period long enough to cover a normal weekly cycle. If your tracking cannot support a number honestly, we tell you that instead of quoting one.
There is no universal threshold, because it depends on your current conversion rate and how large a change you are testing. What we can say is that a step reaching a few hundred people a month can usually support a bold test, while one reaching a few dozen cannot. We check this at the audit and tell you plainly which steps are testable and which are not.
The baseline, friction analysis and prioritised backlog are the first phase, and the timing depends on how much history your tracking holds. After that, each experiment runs for a period fixed in advance rather than a period we can promise you now. We give a written schedule after the audit, once we have seen your volumes and how quickly changes can be published.
Access to your website, analytics, campaign accounts and CRM. An hour from whoever owns marketing, and a conversation with the people answering enquiries. Publishing access, or a developer who can apply changes. And one agreement that matters more than the rest: no other change to a step while a test on it is running.
Against the baseline recorded before we started. Each test is judged on the conversion it was designed to move, over the period agreed before launch. The programme is judged on conversion at each step, cost per enquiry, how many tests were designed and decided, and how much of your funnel is now reliably tracked. Every figure carries its measurement period.
We do not run your advertising campaigns, rebuild your website or replace your CRM as part of this work. We do not promise a particular lift, because no honest testing practice can. Deep statistical modelling and personalisation engines sit outside this service too. Where a test needs work beyond scope, we say so and quote it separately rather than absorbing it quietly.
Conversion rate optimisation improves your website experience, and much of it is judged on craft and good practice. This service is narrower and stricter: it measures the whole funnel, forms hypotheses about specific losses, and settles them with controlled tests. The two work well together, and many clients start here to find out which conversion work is worth commissioning.
We report it in the same detail as a win, then decide together whether the hypothesis was wrong, the variant too timid, or the step too small to settle the question. Flat results are useful. They stop money going into a redesign built on an assumption, and they usually sharpen the next hypothesis considerably.
Yes, and it often works better that way. Your agency knows the site and can publish changes quickly. We bring the baseline, the hypotheses, the test design and an independent reading of the result. That separation is useful, because the people who built a page are rarely the right people to judge whether it should be replaced.
Usually none. Most funnels can be tested with your existing website platform, analytics, forms, campaign accounts and CRM, and we prefer to work inside what you already pay for. If a testing tool would genuinely help at your volume, we say so, explain what it adds, and leave the purchase decision with you.
Then we fix that before testing anything. A test read from unreliable data is worse than no test, because it produces confident conclusions that are wrong. The tracking check is part of the baseline phase, and it names every gap. Some are quick to close, some need developer time, and we tell you which is which.
Yes. In many Indian SMBs the largest single loss sits between the enquiry arriving and someone answering it. We measure that step like any other and test response speed, message wording and who makes the first contact. It is often the change with the shortest path to revenue, and it costs nothing in media spend.
Talk to us
The assessment is free and carries no obligation. You get a plain reading of where your funnel loses people, whether your volumes can support testing, and what we would test first.