The Context: Beyond The 'Man in a Van'
The 'man in a van' ordering on a mobile phone has become the central user story for digital strategy in the building materials sector. It is a useful, if oversimplified, persona. It correctly identifies the core user, the primary device, and the challenging environment: a muddy site, patchy 4G, and a time-sensitive need for materials. We see this borne out in client data, where over 60% of trade-account traffic to builders' merchant websites now originates from mobile devices, a figure that has climbed steadily year-on-year. The focus for many technology leaders, therefore, has been on mobile-first design and performance optimisation. These are necessary, but they are not sufficient.
The real problem is not the screen size or the network latency, though both are factors. The real problem is the nature of the search query itself. A trade professional is not 'browsing for inspiration'. They are placing a precise order for a known item, but they are using industry shorthand to do so. This shorthand is dense with technical specifications. Analysis of our own project data shows that over half of all searches on a typical builders' merchant platform are for commodity products like timber, aggregates, or plasterboard, where the query contains multiple technical attributes like dimensions, grade, or treatment type. A query like '4x2 T C16 reg 4.8m' is not a search term; it is an instruction. Failure to correctly interpret this instruction as a machine-readable set of attributes is the single biggest point of failure for B2B commerce search.
Why Default Search Configurations Fail
Most modern commerce platforms, including Adobe Commerce, and dedicated search services like Algolia or Elasticsearch are powerful. However, their default configurations are almost universally tuned for B2C retail behaviour. They prioritise typo tolerance, broad synonym matching, and features that encourage product discovery and comparison. This is the correct approach if you are selling shoes or electronics, where a search for 'running trainers' should surface a wide range of options. For builders merchants, this behaviour can be counterproductive. When a user searches for a specific length of C16 graded timber, showing them different grades or 'similar' products is not helpful; it is noise that impedes a critical, time-sensitive transaction.
The core technical misunderstanding is treating a trade query as a simple text string. A standard search engine will see '4x2 T C16 reg 4.8m' and look for products whose titles or descriptions contain those characters. If your product is named 'Treated Sawn Carcassing C16 Regularised 47mm x 100mm (4x2) 4.8m', it might find a match. But if the user types '4x2 C16 treated 4800', it will likely fail. The system doesn't understand that '4x2' is a nominal size equivalent to '47mm x 100mm', that 'T' and 'treated' are synonyms, or that '4.8m' and '4800' are the same length. This results in the dreaded 'zero results' page, which from our analysis of client sites, leads to immediate session abandonment in over 90% of cases. The user does not rephrase; they call their trade representative or switch to a competitor.
"A tradesperson searching for '4x2 T C16 reg' isn't browsing. They are ordering. Your search must behave like a precision tool, not a discovery engine."
Deconstructing Queries: A Data Problem, Not a Search Problem
The solution is to treat this as a data parsing problem before it becomes a search problem. Your application needs an interpretation layer that sits between the user's input and the search engine. This layer's job is to deconstruct the query string, identify known patterns, and convert them into a structured query. For example, it should use regular expressions to identify patterns like [number]x[number] as dimensions, [number]m or [number]mm as a length, and terms like 'C16' or 'C24' as a strength grade. These extracted values are then used to build a filtered query against the search index.
This approach is more robust than relying on synonyms alone. You can add 'four by two' as a synonym for '4x2', but this does not capture the dimensional logic. A parsing engine understands that '4x2' and '2x4' refer to the same product dimensions. It can also handle unit conversions, translating inches to millimetres or metres to millimetres automatically. This logic should not live in the search index itself, which should remain a clean, fast repository of product data. The intelligence layer belongs in your application code, where it can be refined, tested, and expanded over time by analysing failed search queries.
This is a key focus in our own replatforming work, such as the B2B commerce platform for Bradfords. By building a dedicated query parser, we can handle the huge variation in regional and trade-specific terminology. The system can identify product codes from competing merchants, understand abbreviations like 'PSE' (planed square edge), and correctly interpret pack-size requirements from the query string. This turns the search bar into a fast and effective ordering tool that reflects how trade professionals actually talk and think, rather than forcing them to learn the specific naming conventions of your product catalogue.
The Commercial Value of Precision
Getting this right has a direct and measurable commercial impact. For every user who can successfully find and order a complex product via search, you reduce the reliance on phone calls to branch staff, lowering your cost-to-serve. More importantly, you build trust and dependency on the digital platform. A reliable search function that understands the user is a powerful retention tool in a competitive market. It demonstrates a genuine understanding of the customer's operational needs, which is far more valuable than any marketing campaign.
On the large-scale Adobe Commerce platforms we run for clients like Huws Gray, search is not an accessory; it is the primary interface to a catalogue of tens of thousands of products. When we pair an effective query parser with real-time stock availability from their ERP and customer-specific pricing, the search results page becomes a complete ordering environment. The user sees the exact product they searched for, confirms it is in stock at their local branch, sees their agreed price, and adds it to their basket. The entire process can take less than 30 seconds. This level of efficiency is what drives digital adoption among a trade audience that is notoriously resistant to poor technology.
