UX/UI Design for SMBs: the method that converts at low volume
Almost all UX literature is written for sites receiving hundreds of thousands of visitors. A/B tests, cohorts, segmented funnels: all of it assumes traffic your SMB doesn’t have. At 2,000 visitors a month, half the methods in the field will teach you nothing at all. Here are the ones that still work, and why they’re enough.
testers: a qualitative starting point, adapted to the audience
form fields on average in a checkout flow (Baymard Institute)
conversions per variant in our low-traffic example
Starter design, excl. VAT/month, 12-month commitment
The problem nobody raises: your site doesn’t have enough traffic
A/B testing, the field’s flagship method, rests on statistics. That has consequences most articles skip.
An A/B test’s required sample depends on the baseline conversion rate, the difference to detect, and the chosen confidence and power levels. At 2,000 monthly visitors and a 2% rate, the expected volume is around 40 conversions a month, or 20 per variant with an equal split. That does not automatically establish a conclusion: size the test before launching it. With low traffic, qualitative observations can help prioritise changes while data accumulates.
Nielsen Norman Group recommends starting with small groups of users and iterating. The often-cited five-person, 85% figure relies on a discovery model and comparable users, not a universal guarantee. Different audiences or journeys may require separate groups and further tests.
What’s left when A/B testing is out of reach
Qualitative methods don’t require volume. They require attention.
For an SMB, five-person testing is the best cost-to-result method available. Half a day, five people who don’t know your site, one clear instruction, and you get a list of real problems rather than hypotheses.
Behavioural analysis, the other method with no volume threshold
Session recordings and heatmaps don’t measure a difference between two versions, they show what people do. Twenty sessions watched attentively teach you more than a dashboard.
The visitor clicking three times on something that isn’t clickable. The one scrolling back for information they should have found further down. The one abandoning the form at the company registration number field because they don’t have it to hand.
What UX design covers, and what it isn’t
Two confusions come up systematically in conversations with company owners.
UX and UI aren’t synonyms
UX design answers "does this work". It covers the full journey, information architecture, task flow, the mental load placed on the user. It’s a discipline of analysis and structure.
UI design answers "is this clear and pleasant". It covers the visible layer: typography, colour, components, hierarchy, motion.
Both are needed and neither replaces the other. A perfectly considered journey, badly dressed, loses credibility. A beautiful interface on confused architecture sends people round in circles attractively.
UX isn’t webdesign
Webdesign produces the visual result and manages its technical performance: what the visitor sees, how fast the page loads, how it behaves on mobile. UX design produces the method leading to that result: what goes on the page, in what order, and how you verify it works.
The two are usually handled together, but they aren’t the same skills or the same deliverables. If your interest is the visual result and load speed, our dedicated webdesign page covers exactly that dimension.
UX isn’t a layer you add at the end
Many SMBs call a UX provider once the site is built, hoping to recover a few conversion points. It sometimes works, marginally. But substantial gains almost always come from structural decisions: which page exists, what appears on it, in what order the visitor meets them. Those decisions happen before the first mockup.
The five frictions that cost SMBs the most
The same problems recur on French SMB sites. Here are five, all fixable without a rebuild.
The frictions we find everywhere
None of these five requires a rebuild. Each takes a few hours to address and can then be measured on your own indicators.
- 01Forms that are too long. Baymard Institute finds the average checkout flow carries 11.3 form fields by default, where most transactions could close with six or seven. Every extra field is an opportunity to abandon.
- 02Missing information at the moment of decision. The visitor reaches the point of contacting you and wonders what it costs. The information is nowhere. They leave for a competitor who publishes a range. You don’t have to publish an exact price: a range is enough to hold the person who was hesitating.
- 03A journey that assumes the visitor knows you. Your menu says "Our solutions". Your pages carry your internal product names. A visitor arrives with no context and wants to know in fifteen seconds whether you solve their problem. Internal vocabulary, however precise, demands a translation effort they won’t make.
- 04The primary action drowned out. A well-built page has one obvious primary action and discreet secondary ones. Many SMB sites show six buttons of identical size, colour and visual weight. Without hierarchy, the visitor picks none.
- 05Mobile as an afterthought. On most French SMB sites, mobile traffic exceeds desktop. Yet journeys are often designed on a large screen then adapted. The result shows immediately: buttons too small for a thumb, forms impossible to fill one-handed, overflowing tables.
What the UX statistics you’re quoted are actually worth
An unusual passage for an agency page, but one that will save you poor decisions.
The figure everyone repeats
You’ll meet it everywhere: every euro invested in UX returns a hundred, a 9,900% return, attributed to Forrester Research.
That figure has three problems. It rests on older work, regularly republished without revision. It presents as an average what was a ceiling observed in the best cases. And it aggregates contexts that have nothing in common: business software used eight hours a day and an SMB brochure site don’t behave alike.
Nielsen Norman Group’s warning
Nielsen Norman Group, who have no interest in downplaying UX value, explicitly warn against three habits in these calculations: treating return purely as money, demanding perfect precision, and trying to account for everything.
Their position is that demonstrating a credible contribution to business goals beats a figure accurate to the cent.
What you can rely on
The relationship between speed and conversion is well documented: on published datasets, conversion runs around 3% at one second of load time and under 1% at four seconds. The mechanism is simple and verifiable on your own site.
Small-group tests help discover difficulties and iterate. They do not prove that no other problems exist or guarantee a percentage of coverage.
The McKinsey Design Index
It shows design-led companies growing revenue roughly a third faster than competitors. It’s a correlation, not a demonstrated causation, and should be read as such.
Track four indicators on your own site: conversion by landing page, form completion, scroll depth and time to first action. They can guide hypotheses, but interpretation depends on volume, measurement quality and context. Compare similar periods and audiences, then cross-check trends against journey observations; an isolated change does not prove the effect of a fix.
How we work on UX at Synerium
Four positions, matched to what an SMB can realistically afford.
- 01Observe
We look before we propose
Every project starts with your existing data and twenty to thirty session recordings. No three-day workshop, no fictional personas built in a meeting room. When that data is missing because nothing is instrumented, we install what’s needed and observe for two weeks.
- 02Test
Five testers rather than a dashboard
Test sessions with five people matching your target, on precise tasks drawn from your commercial objectives. Find the price. Request a quote. Understand what you do in under a minute. These sessions produce directly actionable observations.
- 03Structure
Structure gets validated in grey
Before any visual decision, mockups with no colour or imagery: blocks and real copy only. It’s the only moment where the order of your arguments and the architecture can be discussed without the distraction of colour choices.
- 04Systematise
A light design system, not a factory
A type scale, a palette with its usage rules, ten or so documented components, error and loading states. Fifteen pages, and your site stops degrading with every addition.
What UX work costs an SMB in 2026
UX services are sold at widely varying prices, often without a clear scope. Here are the orders of magnitude on the French market.
By type of engagement
- Heuristic UX audit (expert review, no user testing): 800 to 3,000 euros
- Audit with user testing (five sessions, debrief): 2,500 to 6,000 euros
- Journey redesign on one specific funnel: 3,000 to 8,000 euros
- Full UX design for an SMB site: 5,000 to 15,000 euros
- Light design system: 3,000 to 8,000 euros
These ranges line up with daily rates on the French market, where an experienced UX profile sits in the same bracket as a senior webdesigner, between 400 and 700 euros a day. To compare proposals, our guide to deciphering an agency quote details what to check.
The question to ask before signing
Ask what you actually receive. A UX audit producing forty pages with no prioritisation or costing leaves you exactly where you were.
What has value is a list of problems ranked by estimated impact and correction cost, documented with screenshots. You should be able to decide what to fix first by reading the deliverable, not after three more meetings.
The Synerium approach
UX requests sit within the design scope of the selected Infinity plan. Work can be prioritised in stages: examine a journey, design a fix, then measure its use. Integration requiring development needs Ultimate or a higher plan; timing depends on available capacity and approvals.
That works better on an SMB site, because gains come from accumulated small corrections rather than one large project.
The UX mistakes we see most often
Five reflexes that cost time and money without producing results.
Building personas instead of talking to people
A fictional persona built in an internal meeting reflects the team’s beliefs, not the market. Five conversations with real customers beat three illustrated personas.
Running an A/B test without sufficient volume
The most expensive trap, because it gives the illusion of rigour. A test concluded on a hundred visitors per variant will tell you anything with confidence, and you’ll decide on noise.
Fixing what the team finds annoying
You see your site daily, your visitors once. What tires you looks new to them. Prioritise on behavioural data, not internal feeling.
Adding instead of removing
Faced with a conversion problem, the reflex is to add: an explanatory banner, a chat, a video, a testimonial. Most often the right answer is to remove. Fewer fields, fewer choices, less text before the action.
Treating accessibility as optional
Poor contrast, missing form labels, keyboard navigation that fails: these penalise real users first, and they degrade your search performance too. Fixing them is normal work, not a separate engagement.
Synerium packages
Infinity includes Essentiel (a first free request), Starter, Premium and Ultimate; Titan is quoted individually. Scope and production capacity depend on the plan: development starts with Ultimate. Studio combines a scoped build with monthly credits over 36 months; it is not an Infinity production queue.
Infinity Essentiel and Starter
Essentiel lets you try design with one free request and two revisions. Starter covers subscription design requests, with one active task at a time. See Infinity for current prices and commitment periods.
Infinity Premium
Design, marketing and SEO, with two active tasks at a time. Development is not included, and the plan is not a full-time dedicated team.
Ultimate and Titan
Development is added from Ultimate onwards. Capacity and additional services vary by plan; the Infinity page details the current scope and pricing.
Studio: Vitrine, E-commerce and App
A scoped build, with a setup fee and monthly payments over 36 months. Changes use the credits included in the selected plan. Ownership transfers at the end of the 36 months; source code can be supplied on request from the 12th month of the contract, without ending the remaining payment obligations, under the terms and conditions.
What’s the difference between UX design and UI design?
UX design shapes the journey and architecture: what the visitor meets, in what order, with what effort. UI design shapes the visible layer: typography, colour, components, hierarchy. UX answers "does this work", UI answers "is this clear and pleasant".
Can I run A/B tests with my SMB’s traffic?
It is possible if the test is sized for your baseline conversion rate, the difference to detect, and the chosen confidence and power levels. At 2,000 monthly visitors and 2% conversion, the expected volume is around 40 conversions a month. The required duration may not fit a quick decision. User testing and expert evaluation can help prioritise changes without replacing quantitative evidence.
How many people do you need for a user test?
Five people can form an initial qualitative group for a defined audience and journey. The 85% figure often associated with Nielsen Norman Group depends on a discovery model and comparable users: it is not guaranteed coverage. Use separate groups for different audiences, then iterate after the first changes.
How much does a UX audit cost for an SMB?
Between 800 and 6,000 euros depending on depth. A heuristic audit runs 800 to 3,000 euros. An audit with five test sessions and a debrief runs 2,500 to 6,000 euros. At Synerium, UX is built into the Infinity subscription with no separate billing.
Is the "1 euro in UX returns 100" figure reliable?
Treat it with caution. Attributed to Forrester Research, it rests on older work, presents a ceiling as an average, and aggregates very different contexts. Nielsen Norman Group themselves warn against this kind of calculation. Better to measure four simple indicators on your own site.
How long does UX work take?
A heuristic audit takes two to four days. An audit with user testing takes one to two weeks, allowing time to recruit participants. Complete UX design for an SMB website takes four to eight weeks. With a Synerium subscription, work is organised in monthly stages rather than a single engagement.
What’s a design system and do I need one?
A library of reusable visual components with their usage rules. For an SMB, a light system of around fifteen pages is plenty. It becomes useful as soon as several people touch the site.
Where do I start if I’ve never done UX?
With twenty session recordings watched attentively. It’s free or close to it, requires no particular skill, and you’ll spot problems neither you nor your agency had identified.
Does UX improve my search ranking?
UX primarily aims to make journeys useful and accessible. Some improvements to loading, stability and responsiveness overlap with the Core Web Vitals Google considers. This does not guarantee better rankings, and longer visits alone do not establish an SEO effect. Measure site usage and organic acquisition separately.
Do I need to rebuild the whole site to improve UX?
No, and it’s rarely the right answer. The fastest gains come from targeted fixes: removing form fields, publishing a price range, clarifying a page’s primary action, rewording menu labels.
How do I know whether my problem is UX or traffic?
Look at conversion by landing page. If traffic rises but conversion stays flat, the problem is downstream: visitors arrive but are not convinced. If conversion is healthy but volume is low, work on visibility first. Investing in UX on a site with no traffic improves the experience of visitors who never arrive.
Which tools do you use?
Figma for design and prototypes, behavioural analytics tools for session recordings and heatmaps, and your existing analytics tool for quantitative data. We prefer connecting what you already have to adding another subscription.
Is UX included in the Synerium subscription?
Yes, for design requests within the selected plan. Infinity includes Essentiel, Starter, Premium and Ultimate; Titan is quoted individually: Essentiel is a limited trial; subscriptions process requests through a queue according to the chosen scope and capacity. Premium does not include development, which starts with Ultimate. Studio covers a scoped build and changes within its monthly credit allowance, not an unlimited Infinity queue. See the offer pages for details.
What if user tests contradict what I believe?
It happens regularly and is exactly what makes testing useful. We show you the recordings rather than asking you to take our word for it. The final decision remains yours, but it is then based on observations rather than assumptions.
Should I repeat user tests regularly?
After a structural change, yes; after a simple visual adjustment, no. In practice, two or three test sessions a year are enough for an SMB whose site evolves incrementally. Reviewing session recordings every month matters more: that is where new problems first appear.
Going further
Rémi Lenclud
Founder of Synerium
Rémi Lenclud runs Synerium, a subscription-based digital agency for French SMBs. He has spent over ten years on web projects where design and technical execution are handled by the same team, advising owners on the methods that produce results at low traffic volume.
LinkedIn profileStart by watching what your visitors do
Journey observations and a few indicators can guide an initial diagnosis. We discuss the available data and an appropriate method in thirty minutes, with no commitment.