Conversion Rate Optimisation
A site that takes six seconds to load, hides add to cart below the fold, or breaks checkout on an older phone does not have a hypothesis to test. It has defects to fix. We fix those first, at any traffic level, and test afterwards if your volume supports it.
From £1,000/month, or a one-off audit from £1,500.
Three-month minimum on the programme, then rolling.
They have different requirements, and most brands need far more of the first than they have been sold.
01
Load speed, script weight, page layout, where the buy button sits, how the journey flows, whether checkout works on a four year old Android. These are defects measured against known behaviour, not hypotheses. They need no traffic volume to justify, because you are not proving anything. You are removing something that is demonstrably in the way.
02
Anything genuinely uncertain: a new page layout, a different offer, a changed value proposition. Here you do need volume, because the only way to know whether a new idea is better is to measure it against the old one at significance. Below a few hundred orders a month, that takes longer than it is worth.
Treated as one service, the two get sold together at any volume. A brand doing 200 orders a month ends up paying for experiments it will never be able to read, while a six second load time costs it money the entire time the tests are running.
Everything here is structural. None of it needs a test to justify fixing, and none of it depends on your traffic volume.
Shopify app bloat is the single most common one. Every app you install adds JavaScript to every page, including the ones it is not used on. Most stores are carrying scripts from apps that were uninstalled years ago.
Once the structural work is done, testing is how you answer the questions that are genuinely open. What it can tell you depends entirely on your order volume, and that is arithmetic rather than opinion.
Low volume is not a reason to do nothing. It is a reason not to buy a testing retainer, which is a different thing, and the distinction is worth knowing before you sign one.
A 15% cut in cost per click improves one channel. A 15% improvement in conversion rate improves every channel at once, including the traffic you are not paying for. It compounds against organic search, email, marketplace referrals and direct, and it keeps working after you stop spending.
It also changes what you can afford to bid. Once the site converts better, campaigns that were marginal become profitable, and the ceiling on paid acquisition moves up. That is the real reason conversion work belongs alongside media rather than in a separate project with a separate supplier.
01
Script audit across every app and tag, deferred or removed where it is not needed on that page, images properly sized and compressed, render-blocking resources moved. Speed is the one improvement that lifts every page, every device and every traffic source at the same time.
02
Most traffic, usually the worse build. Interaction delay, layout shifting as content loads, buy buttons that need precision to hit, sticky elements covering the thing you are trying to tap. Desktop is usually fine and is not where the money is.
03
Where the buy button sits, how many steps stand between interest and payment, what the navigation buries. This is layout and flow rather than design, and it is decided by how people actually move through the site rather than by preference.
04
Shipping cost shown early rather than at the end, guest checkout available, the payment methods your customers expect. Fewer people start checkout when the cost is honest upfront, and far more of them finish.
05
Sizing, materials, delivery date, returns, compatibility with something they already own. Buyers leave to find the answer and do not come back. This is usually content work, not design work.
06
Site-wide conversion rate moves when your traffic mix moves, which makes it useless for judging whether the site improved. We report by device, source and new against returning, because that is where a real problem is actually visible.
A UK bags and luggage brand, client since May 2021. Figures are their own platform data.
Year before we started
First full year
Total since
Google Shopping was rebuilt and Meta prospecting added, but the conversion work is what let the spend scale. Doubling the rate at which visitors bought meant every campaign could support a higher cost per click, which opened up traffic that had previously been unaffordable.
That is the pattern worth understanding. Conversion work rarely produces a headline of its own. It changes what the rest of the account is capable of.
One client outcome under specific conditions, not a forecast for your brand. Cumulative revenue since engagement.
Weeks 1 to 2
Funnel analysis by device and source, session recordings on the drop-off steps, speed and Core Web Vitals, checkout walked through as a customer. Volume assessed so we both know what is testable.
Weeks 2 to 6
Script and app audit, speed fixes, mobile layout, checkout flow, the content gaps stopping purchases. No testing yet, deliberately. Running an experiment on a known defect wastes weeks proving something you already knew, and the defect keeps costing you money while it runs.
Weeks 6–12
Only now, and only on the questions that are genuinely open. Hypotheses drawn from your funnel data rather than a list of best practices. Tests run to significance rather than to a reporting deadline, and losing tests reported as clearly as winning ones. If your volume does not support testing, this stage is more structural work instead, and we will say so.
Ongoing
Every conversion improvement changes what campaigns can afford to bid. If we also run your paid media, that adjustment happens automatically. If we do not, you get the numbers to hand to whoever does.
A one-off audit if you want the diagnosis and will act on it yourself. An ongoing programme if you want it run.
Conversion work on its own is worth having. It is worth more sitting next to the media spend it makes affordable, which is why most brands who start here end up on the programme.
It works when
It does not when
Yes, and often more than for a high-traffic brand, because low-traffic sites tend to have had less attention paid to them. Speed, layout, flow and checkout are all fixable at any volume and none of them require a test. What you should not buy at low volume is a monthly testing retainer, because you will spend a year failing to reach significance while paying for the privilege.
Testing is the best known part of it, and it answers open questions well. What it does not do is fix defects. If your add to cart sits below the fold on mobile and twenty apps are loading scripts on every page, no experiment is needed to establish that, and running one first costs you months. Worth asking anyone you speak to how much of their proposal is testing and how much is fixing.
A good deal of it. Script weight, third-party tags, image handling, interaction delay and layout stability are engineering problems with commercial consequences. We do this work rather than recommending it and leaving you to find someone, which is why it sits alongside the tracking and measurement work rather than in a separate design project.
Fixes land within weeks and their effect is usually visible quickly. Tests take as long as your traffic dictates, which the audit will tell you. Anyone promising a specific percentage improvement by a specific date has not looked at your numbers.
No. Design and development are quoted separately and often by someone else. A redesign changes everything at once, which means you learn nothing about what worked. Improvement and redesign are different projects and conflating them is how brands lose conversion rate while spending money to gain it.
Many do, and we report them the same way as wins. A programme that only ever reports winners is either not testing properly or not telling you everything. The losses are how you learn what your customers do not care about, which is worth knowing before you build more of it.
Whichever suits your platform and volume. The licence is paid directly by you rather than marked up through us. At lower volumes we will often recommend not paying for one at all.
Yes, and often that is the sensible arrangement. We specify and prioritise, they implement, we measure. If you would rather we implement on Shopify directly, that is included.
Vaseem Shaikh has spent fifteen years running paid media and ecommerce accounts for consumer brands. Before that he spent five years researching artificial intelligence, genetic algorithms and machine learning at IIM Ahmedabad, work he has published and now applies directly to how experiments are designed and read.
That background is the reason results here are reported at significance rather than at a convenient moment, and why losing tests are reported alongside winning ones.
He leads a team of specialists across paid media, content, retention and analytics. You speak to him during the audit, and he stays involved afterwards.
Google Ads Advisors is an invited research community Google consults on products and reporting ahead of general release. It is not a partner tier or a certification.
Vaseem Shaikh
Founder, Signal Over Noise
Where visitors drop out, what each drop-off is costing, whether you have the volume to test, and a ranked list of what to fix first. About five working days, no charge.
You keep it either way. Hand it to your developer if you want to.
Start a project
Tell us about your brand. You get a written audit of your advertising, site and marketplace position, ranked by what each fix is worth.
No charge. About five working days. We reply within one working day, Monday to Friday, 9:00 to 18:00 UK.
London, UK
71–75 Shelton Street
Covent Garden, London, WC2H 9JQ
Ahmedabad, India
DTC & ECOMMERCE GROWTH AGENCY
Services
Agency
We usually reply within a few hours
Our team will be in touch with you shortly.
