How industrial buyers build a shortlist before they ever call you

Picture a maintenance engineer at a cement plant outside Pleven. On Monday the Falk gearbox on the main conveyor starts running hot. By Thursday she has three repair shops on a list, has read two of their websites twice and has asked a colleague at another plant who he used last time. None of the three shops knows she exists yet.

That list of three is where the order will go. If your company is not on it, you do not lose the job in any way you can see. You simply never hear about it.

The choice is made before the first call

The three studies point the same way. Most of the time your sales team is not competing to win a buyer over in a meeting. It is competing to be one of three names written down before anyone picks up the phone, and that list is built from what the buyer can find and check without you.

By the time a quote request reaches your inbox, with a drawing, a quantity and a deadline, the research behind it is finished. It happened on Google, in an AI assistant and on a few websites, and none of it shows up in your CRM unless you track it.

The path an engineer actually takes

Back to the engineer with the hot gearbox. Her research is specific, quick and mostly silent. It usually runs in this order, and a buyer looking for a CNC machine dealer follows the same steps.

Step one depends on knowing which searches your buyers type, and they are rarely the words in your brochure. The buyer keywords playbook shows how to pull them out of Search Console and your own sales emails.

Step two is the newest. In the 2026 State of Marketing to Engineers research, 69% of technical buyers said they use generative AI during the purchasing process, yet they rate their trust in its answers at 4.7 out of 10, so they click through and check the cited page. The assistant decides who gets looked at; your website still decides who gets the request. I cover how to get named in those answers in the AI search playbook.

Where Bulgarian suppliers fall off the path

Each step is a filter. Eurostat's 2025 figures show how many Bulgarian companies fail one before a buyer has formed any opinion of them. Bulgaria has the lowest share of companies with a website in the whole EU.

Read the rows against the path. A manufacturer with no website, which is almost half of them, is out at step one and has nothing to show at step four. A site that lists no products or prices fails step three, because the engineer cannot confirm fit and opens the next supplier instead. A site in one language loses the buyer in Vienna or Bucharest who searches in German or Romanian.

There is a useful side to this. In many Bulgarian industrial niches, page one of Google is held by a handful of thin directory listings. You do not need to beat fifty strong competitors, only be clearer than the few who show up.

What engineers count as credible

Asked what makes a supplier credible, engineers give a well-kept website an average of 4.0 out of 5. Trade show sponsorship scores 2.8. The stand at a fair starts the conversation; the website is where it gets checked afterwards.

Notice what is missing from that list: slogans, mission statements, photos of handshakes. Engineers want the facts that decide whether your part or service fits, written on the page as text, where Google and AI assistants can read them too. And four in ten want price or stock information first, so a “from” price, a typical lead time or a stock line is worth more than a paragraph about quality.

For a repair shop, that means one page per brand and type of job, with the sizes you handle and your turnaround. For a CNC dealer, one page per machine with travels, spindle, control, what is in stock in Bulgaria and how fast it can be installed. I go through the structure in capability pages that win RFQs.

What it looks like when you are on the path

Two of my projects show both ends of the path: the classic Google search and the newer AI question.

An industrial repair company in Texas, specialising in gearboxes and pumps, had decades of workshop experience and a website that did not show it. I built brand and model pages for the gearboxes its customers run: Falk, Western, Philadelphia Gear and Lufkin. A plain guide, “What are centrifugal pumps used for”, became the site's #2 landing page, after the homepage.

In 2022 against 2021, leads went from 234 to 800 and clicks from Google from 260 to 1,800. The site had 29 keywords in Google's top 10, with “gearbox overhaul”, “industrial gearbox repair”, “gear hobbing services” and “gearbox repair services” at #1. The honest read: bounce rate rose slightly, from 59.7% to 63.3%, as more first-time readers came in through the guides. Leads still more than tripled.

ITCE, a B2B IT training and ServiceNow consulting company in Sofia Tech Park, shows the AI side. From 19 June to 16 September 2026, against the 90 days before, visits from AI assistants went from 8 to 77, and key events from AI from 0 to 28. By mid-September, AI engines were citing 48 itce.com pages. The volume is still small, but the quality is not: AI visits converted at 10.4%, against 10.6% for Google organic.

The shortlist audit

You can run this yourself in one afternoon. Start with your best-margin product line and write the answers down, because you will want to repeat this in three months.

Where to start

Count the gaps the audit found: each one is a step where an engineer can drop you before calling. Then fix one product line before you touch the rest: the page, the specs as text, a price or lead time, a second language. Count what comes in afterwards, so you know which searches turn into quotes; what counts as a qualified lead explains the yardstick I use.

If you want a second opinion on your results, book a call and I will go through them with you.