The Website Is the Easy Part: TSD Makes Commerce’s Messy Plumbing Pay

The most revealing word on TSD’s website is not “AI,” “enterprise” or even “e-commerce.” It is “just.” Just connect the ERP. Just migrate the customers. Just let trade buyers see their own price, while consumers see another, then reconcile both with stock held somewhere else. In a modest brief, “just” is where the budget, calendar and optimism go to have an argument.

TSD has built a 23-year business around that argument. Founded by Tim Gaunt in 2003 as The Site Doctor, the privately held company sits near Ross-on-Wye in the spectacularly un-tech-hub-sounding hamlet of Hole-in-the-Wall. Its office is a converted farmhouse. Its work is the digital machinery used by cider makers, jewellers, marketers, charities, manufacturers and financial-services startups. The company’s legal entity arrived in 2008; the accumulated portfolio reaches back to a 2004 Adriatic holiday portal written in classic ASP.

Calling it a web agency is accurate in the way that calling a restaurant a room with an oven is accurate. TSD designs storefronts, but its defensible work happens behind them: custom pricing, payment flows, stock synchronisation, membership rules, single sign-on, reporting, data validation and the careful replacement of old systems that everybody dislikes and nobody can switch off.

The expensive sentence begins with “just”

TSD occupies the stretch of market between a templated shop and a giant systems integrator. Its typical buyer is not launching a first candle store. It is an established organisation with an ERP that must remain the source of truth, a CRM full of history, peculiar rules that exist for good commercial reasons and staff quietly copying data between systems. Shopify and off-the-shelf apps may still appear in the solution. So do Umbraco, Ucommerce, Sitefinity, .NET, Azure and a crowded cast of APIs. The point is not allegiance to a stack; it is making the stack behave like the business.

The company is unusually direct about price. A focused integration may begin around £15,000. Many bespoke applications land between £75,000 and £150,000. A full platform starts around £150,000 and can keep going. For work that cannot honestly be scoped at first glance, TSD starts with estimate bands, uses discovery to expose the unknowns, then commits to what Gaunt calls a watermark figure with a visible contingency ceiling. That does not make uncertainty disappear. It stops uncertainty dressing up as certainty in a proposal.

“Nobody’s really buying a website. They’re buying operational improvement.”Tim Gaunt, founder and CEO

Cider, diamonds and domestic-abuse data

The portfolio makes the positioning concrete. Westons Cider had an ageing WordPress brochure site separated from its store. Peak-period outages during the pandemic made that split more than untidy. TSD brought content and commerce together, connected Microsoft NAV and the CRM, imported order history and added product upsells and automated re-engagement. Westons reports uptime above 99.99 percent. The more important change was internal: its team could edit the site while one joined-up platform carried the customer journey.

At Bond Jewellery, the visible problem was a slow, awkward store that the team disliked managing. The hidden problem was jewellery itself: metal variants, diamond grades, live availability, certification and prices that move. TSD migrated the operation to Umbraco, built a taxonomy and faceted search, linked a live Nivoda diamond feed, handled currency conversion and presented more than 6,500 products. Launched before Black Friday and Christmas, it increased sales and engagement, though TSD publishes no percentage. The site won the 2025 Umbraco Jury’s Choice award after more than 200 projects were submitted across the awards.

SafeLives had higher stakes. The domestic-abuse charity’s Marac data platform ran on Drupal 7 as that architecture approached end-of-life. Validation was cumbersome, users were frustrated and the deadline was real. TSD won the tender with a competitive price plus project-management support, then built multiple routes for data entry, inline checks and graphing in an interface familiar to users. There is no grand numerical victory published. There is a more credible one: better data quality, higher user satisfaction and fewer manual processes in a system that supports frontline work.

A coffee, six months and a lot of APIs

QuoteHub shows what the same integration habit looks like with AI. The idea began over coffee: let an electrician describe a job and quickly obtain suitable wholesale quotes. In under six months, TSD helped turn it into iOS and Android software built on .NET MAUI. OpenAI and Azure interpret requirements; Auth0 handles identity; Twilio sends messages; Stripe and Two handle different forms of payment; Loqate checks addresses; Propello supplies rewards; Shopify places orders. Offline transcription lets a user capture a job where reception is poor. An admin application gives QuoteHub’s staff control over orders, customers, retailers, refunds and support.

That distinction matters. Generative AI is one component, not the business model. Elsewhere, TSD supplies engineering support and fractional CTO input to Alphastream, a private-credit intelligence company. It also builds conventional internal applications, performs platform migrations and supports existing systems. Its customers buy projects and ongoing care. Product income from tools such as analytics platform TrendSeam and Umbraco builder Igloo supplements agency revenue; TSD says it does not exceed it.

The failure arrived before the pivot

The best evidence for TSD’s operating philosophy is a product that stumbled. TrendSeam launched on Shopify in 2017 and initially picked up multiple installs a day with little marketing. Then reviews noted a breaking API integration. TSD had not kept pace with Shopify’s changes. Marketplace ranking fell and installs did not taper gently. They stopped.

That changed how the company thinks about products built on somebody else’s land. Its budget now assumes annual maintenance equal to 20 to 30 percent of the original development time. A quarterly scorecard watches platform releases, marketplace growth, competing features, reviews, installs and churn. Core value is kept platform-agnostic where possible. TrendSeam’s analysis can travel even if its Shopify connector cannot.

TSD also learned that agency products die when treated as hobbies. At one point it tried four simultaneously; none received enough attention. The current cap is two launches a year. Product time is booked like client time, with named developers and regular stand-ups. A blunt 30 to 40 percent of that time is reserved for documentation, support and marketing. The building, it discovered, is perhaps 60 percent of the work.

The bits worth stealing

TSD’s playbook is practical because it is built from things that went wrong. A buyer or another agency can borrow it without borrowing the company’s technology choices.

1. Begin with the operation. Ask what happens after an order, where pricing lives and which manual task hurts. A feature list will not reveal any of that.
2. Keep one source of truth. For mid-market commerce, TSD generally leaves the ERP in charge and uses hybrid synchronisation rather than pretending every fact belongs in the storefront.
3. Price uncertainty visibly. Start in broad bands, pay for discovery, then show the contingency. False precision merely postpones an uncomfortable meeting.
4. Budget for dependence. If a product lives in a marketplace, reserve time every year for versions, APIs and security before calculating the return.
5. Protect product work. Cap the pipeline, give it real capacity and account for the manuals, support and marketing before celebrating the build.

The approach is not universal. A standard catalogue, simple pricing and modest integration needs are often better served by a template and a competent implementer. A cash-poor startup that needs to test demand may not be ready for a £75,000 application. Nor does TSD’s platform-product strategy suit a company whose survival depends on one marketplace and lacks engineers to follow every release. Bespoke software earns its keep when the underlying business is genuinely bespoke, the operational gains can repay the build and somebody is prepared to own it after launch.

For the right customer, TSD’s difference is not a secret framework. It is senior engineering applied to unfashionable details, followed by support after the launch photographs have been taken. The company competes with specialist Umbraco shops, custom-software agencies, in-house teams and large consultancies. Its useful middle ground is a buyer too complicated for plug-and-play but too pragmatic to commission a transformation opera.

In Hole-in-the-Wall, that has become a coherent proposition: sell the operational improvement, admit what it costs, and treat the website as the visible end of a much longer pipe. The work rarely fits in a clever slogan. It does, however, fit inside the word “just.”