You read everywhere that building from scratch — pure custom development — is the ultimate weapon for your website. Maximum performance, perfect SEO, rock-solid security, clean code, total freedom. All of that is true. And yet, in 80% of the small business projects I encounter in the Gironde region, classic from-scratch development is the wrong choice.
Not because it's inherently bad. But because it is generally over-sold, overpriced, and unsuited to what the majority of craftsmen, retailers, and independent professionals actually do with their website. I'm going to explain the precise cases where it becomes relevant — and why, since AI entered the development process, the question itself is changing in nature.
"From-scratch development is not a product. It's a response to a specific business need. If you don't have that need, you're paying for something that will never serve you.
Why 80% of small businesses that pay for from-scratch didn't need it
The from-scratch market as it developed throughout the 2010s–2020s rested on a simple economic argument: to have a fast, secure, SEO-optimised site, you needed proprietary code — which meant expensive custom work. Agencies were selling a unique product, justified by a need for unique technical quality.
Except that in practice, most small business websites don't need that unique quality. A five-page showcase site presenting your services, a contact page, a contact form — that's a standard need. Standard doesn't mean mediocre: it means thousands of businesses have exactly the same need as you, and there are proven, high-performance, secure solutions that address that need without rewriting everything by hand.
The classic mistake is confusing quality of outcome with uniqueness of code. A properly built WordPress site with a custom theme can reach 95% of the performance of pure from-scratch development, for 30% of the budget. And you'll gain those remaining 5% of performance only if you're willing to pay three or four times more to get them. For an e-commerce business turning over €500,000 a year, that can be justified. For a craftsman who wants a showcase site, it's a hard sell.
My recommendation: before considering from-scratch development, ask yourself one simple question. "What does my site need to do that standard solutions can't?". If you're stuck for more than thirty seconds on that answer, from-scratch is probably not for you — at least not classic from-scratch.
When your site needs to do something nobody else does
The first case where from-scratch becomes relevant is when your business has a functional logic that existing plugins and themes can't properly replicate. Not a presentation, not a form — a genuine business tool integrated into your site.
Some concrete examples I come across: a dynamic quote configurator for a joiner (wood selection, dimensions, options, real-time price calculation), a multi-slot booking system with team management for a medical practice, a warranty comparator for an insurance broker, a pricing module that connects to your business software (Sage, EBP, Pennylane). In these cases, you're not looking for a website — you're looking for a work tool accessible via the web.
That's where standard solutions quickly hit their limits. A WordPress calculator plugin will do 70% of the job, but the remaining 30% — the fine-tuned integration with your process, the tailored user experience, visual consistency, robustness under heavy load — can only be achieved with specific code. And there, paying €8,000 or €15,000 for that functionality makes sense, because that tool will save you time or generate revenue with every single use.
My recommendation: if you identify a genuine business logic unique to your activity, don't try to "hack" it together with plugins. Invest in custom development focused on that specific logic — not the entire site, just the part that truly differentiates you. Everything else can remain standard.
When every millisecond and every ranking position genuinely costs money
The second case where from-scratch is justified is when your website is your primary acquisition channel and your market is competitive enough that every detail matters. Not just "I'd like to rank on Google" — but "my position in the top 3 determines 60% of my revenue."
In that context, Core Web Vitals (the performance metrics Google uses to rank sites) become a critical lever. A 0.3-second improvement in LCP can shift your position by several ranks on a commercial query. A well-built WordPress site often plateaus at 75–85 on mobile Lighthouse because of accumulated plugins. A from-scratch build regularly hits 95–100. That gap, in a competitive niche, shows up in monthly revenue.
But be careful: this case applies to very few businesses in reality. For it to truly apply, (a) your site must be your real commercial channel, (b) your niche must be competitive, (c) you must have already maxed out the other levers (content, backlinks, Google profile). If you're a plumber in a small suburb, unlikely. If you're an e-merchant in a contested niche — gift boxes, luggage, food supplements — possible. If you're a franchise competing with other franchisees on local queries — probable.
My recommendation: before investing in from-scratch for SEO reasons, objectively measure your current acquisition channel. If more than 50% of your clients come from Google, and your niche is saturated, the maths can work. Otherwise, better to invest that money in content and backlinks — you'll gain more positions, at lower cost.
A client was selling premium gourmet gift boxes on a standard Shopify store. Mobile Lighthouse score of 62, average position 7–9 on her main queries. Rebuilt with AI-assisted custom development, optimised for Core Web Vitals: 96 on mobile Lighthouse, moved into the top 3 on three main keywords within eight months. Her organic sales tripled. Without the precision of from-scratch, that climb would have been impossible.
When your brand identity simply can't fit inside a theme
The third case, rarer and often misunderstood, is that of strategic visual differentiation. Not "I'd like my site to look nice" — everyone would. But "my brand positioning rests on a visual experience that nothing existing can replicate."
This is typically the case for a high-end interior architect whose site is his first calling card to demanding clients, a luxury artisan brand where every visual detail reinforces the premium positioning, a creative agency whose site itself must be a manifesto, a product or service sold on emotion rather than functionality. In these cases, starting from a theme — even a premium one, even heavily customised — locks you into visual conventions that your brand is precisely trying to transcend.
But let's be honest: this case applies to fewer than 5% of small businesses in Gironde. The majority of craftsmen, retailers, and independent professionals need a professional, credible, coherent site — not a visual manifesto. Wanting a "different" site without having built a brand that justifies that difference is buying a prestige object that will bring you nothing.
My recommendation: ask yourself this frank question: "Do my clients choose me because of my image, or because of my skills?". If it's the image, visual from-scratch is justified. If it's the skills, better to invest that money in proof of your skills (case studies, testimonials, content) than in exceptional design.
The 3 cases where from-scratch changes everything, and the rest where it's oversold.
Bad reflex
80% · Of small businesses paying for from-scratch don't need it.
Business case
1 function · A genuine functional logic unique to your activity.
SEO case
Contested niche · Every millisecond genuinely matters.
Brand case
5% · Strategic visual differentiation (rare).
What I've written above was true until 2023, when from-scratch cost 10 to 20 times more than a classic CMS. Since AI entered the development process, the entry cost has dropped dramatically: AI-assisted custom development is now delivered at the price of a tamed CMS.
The consequence: the question is no longer "do you need it?" but "why not take advantage of it, since it's now accessible?". That's the subject of my article on the CMS vs custom debate in 2026.