The Shortlist Is Written Before Anyone Visits Your Website
Enterprise buyers used to open ten tabs and compare. Now more of them open one question: who are the best partners for an SAP S/4HANA migration, a Salesforce rollout or a ServiceNow clean-up? An AI assistant replies with three or four names, a sentence on each and a few links.
If your firm is not in that reply, you are not in the first round. You will also never see the lost deal in your analytics, because the visit never happens.
This guide covers what an assistant needs from a technology firm, and the specific changes that earn a place in the answer.
What an Assistant Needs From a Technology Firm
An assistant builds its answer from pages it can read, facts it can cross-check and sources that agree with each other. For a services firm, that comes down to four things:
- A clear identity. One exact description of who you are, what you do and who you do it for, worded the same way on your site, LinkedIn, partner listings and press.
- Direct answers. Pages that state the answer in the first two sentences, then support it.
- Proof it can verify. Named engagements, named platforms and named industries, with the outcome described plainly.
- Access. Pages that load without scripts hiding the content, and crawlers that are allowed in.
Miss one and the assistant has less to quote, so it quotes someone else.
Write the Answer First, Then the Argument
Take the question: how long does an SAP ECC to S/4HANA conversion take? A page that opens with a two sentence answer and the factors that move it can be quoted. A page that opens with a company story cannot.
Use one rule across every service page and article: one question per heading, the answer in the first two sentences, then the detail. Headings should say something. "Our Approach" tells a machine nothing. "A system conversion keeps your data and processes, a new implementation starts clean" tells it exactly what the section answers.
It is also better for people. A CIO skimming on a phone gets the point in ten seconds. Our guide to choosing an SAP implementation partner follows the same pattern.
Make the Facts Machine Readable
Structured data tells a machine the same facts a person reads on the page: who the organization is, which services it offers, which questions a page answers, when an article was published. The types worth having on a services site are Organization, Service, Article, FAQPage and BreadcrumbList.
The rule that matters: markup must match the visible text. Marking up content that is not on the page is ignored at best and penalised at worst.
Then the technical basics that decide whether a page can be used at all:
- One H1 per page, a unique title and description, and one canonical address per page.
- A sitemap with real modification dates.
- A robots.txt that does not block the crawlers you want. OAI-SearchBot (ChatGPT search), PerplexityBot and Googlebot (which feeds Google AI Overviews) are the ones to check. Read your server logs rather than assuming.
- Content that is in the HTML, not assembled later by scripts.
An llms.txt file listing your key pages is cheap to add. No major assistant has confirmed it relies on one, so treat it as housekeeping, not strategy.
Earn Mentions Where Assistants Look for Agreement
An assistant trusts a claim more when independent sources repeat it. For an enterprise technology firm those sources are partner directories (where you are genuinely listed), review platforms, industry publications, event pages and case studies that clients agree to publish.
One accurate third party description is worth more than ten pages of self praise. Keep it accurate: name the platforms you really work on, the industries you really serve and the scale you really deliver.
Do not buy mentions or invent them. Search engines and assistants discount manufactured ones, and a wrong statement about your firm is harder to correct than a missing one.
Measure the Answer, Not Only the Click
Assistant traffic is small and often arrives looking like direct traffic, so counting visits hides the effect. Measure the answer itself.
Write 20 questions your buyers really ask, across your practices, regions and buying stages. Run them each month in ChatGPT, Perplexity, Gemini and Google. For each one, log whether you are named, what is said about you, and which pages are cited.
When a statement is wrong, fix it at the source page, not in the assistant. Add "AI assistant" as an option on your enquiry form and ask new leads how they found you. The first honest answers are usually surprising.
Where an Outside Specialist Helps
Most of this is work your own team can start this week: tighten pages, add schema, audit crawler access and fix inconsistent profiles. A specialist helps when you need a baseline across many pages and competitors, or when content and technical changes have to move together.
That is the job of AI search optimization: finding out how assistants describe your firm today, then fixing the pages and sources they draw on. For Google in particular, this step by step guide to ranking in Google AI Overviews walks through the method in detail.
For the technology side, such as making sure your SAP, Salesforce and ServiceNow pages describe real, verifiable work, see how our AI practice and delivery teams document it.
A One Week Starting Plan
- Day 1: write the 20 buyer questions.
- Day 2: run them in four assistants and record the results.
- Day 3: rewrite your three most important pages answer first.
- Day 4: validate your structured data and check robots.txt against the crawlers above.
- Day 5: make your firm description identical across your site and profiles.
Then repeat monthly and compare. If you want a second pair of eyes on how your SAP, Salesforce, ServiceNow or AI work is described, talk to our team.