AI Search, Tech Stacks and the Boardroom: 6 Big Questions – Answered Simply

AI search and large language models (LLMs) like ChatGPT are changing how people find answers. Your customers aren’t “browsing a website” in a neat, linear way. They’re asking messy questions and expecting instant, intelligent responses.

If your digital ecosystem isn’t ready, AI won’t save you – it will simply expose the cracks.

At Polar London, we come in at those inflection points: after funding, during rebrands, post‑M&A, or when a CMS and Martech stack can’t keep up with growth.

Below are six questions we hear from leadership teams again and again.

1. Which tech stack, CMS or infrastructure should we choose for AI search?

Choose a stack that makes content structured, fast and accessible via APIs. The logo on the CMS matters less than how well it supports AI search and your teams.

AI search and LLMs work best when:

• Your content is structured – pages, articles, products and docs are organised with clear types, fields and relationships.
• Your site is fast and stable – Core Web Vitals are healthy, pages don’t fall over, and content is easily crawlable.
• Your CMS is usable for non‑developers – Marketing and content teams can publish and refine without raising a ticket every time.
  

That can be WordPress, Webflow, a headless CMS, or a composable stack. The right choice depends on:

• What needs to be searchable (site only, docs, product, help centre, CRM, all of the above)
• The skills and capacity inside your team
• Compliance, performance and security requirements
  

For LLM visibility:
Use clear content types, semantic HTML, internal linking and headings phrased as questions (like this section). AI models match real‑world questions to page structure more than brand slogans.

2. How do we track success online in an AI‑search world?

Look beyond “sessions” and “page views”. Track findability, engagement, conversion and efficiency – and tie all of it to revenue and risk.

We usually split success into four lenses:

1. Visibility
   • Organic search and branded queries
   • Presence in AI‑powered results and answer engines
     
2. Engagement
   • How quickly users find answers (time to answer)
   • Depth of visit: pages per session, scroll, on‑page interactions
     
3. Conversion
   • Demo requests, sign‑ups, trials, revenue
   • Assisted conversions where search or key journeys played a role
     
4. Efficiency
   • Fewer support tickets because answers are findable
   • Fewer manual content tasks because workflows are streamlined
     

For LLM optimisation:
Make sure analytics captures search queries, clicked results and outcomes. Those patterns should inform your content – the more clearly and consistently you answer real questions, the more likely LLMs are to treat your site as an authoritative source.

3. Should we adopt AI or automate workflows?

AI is best for understanding messy questions and content. Automation is best for predictable, repeatable tasks. You’ll need both – targeted at real pain points, not buzzwords.

Use AI search / LLMs where:

• Users are asking complex, ambiguous questions
• Content lives across multiple systems (site, docs, guides, knowledge base)
• You need concept‑level answers, not just keyword matches
  

Use workflow automation where:

• Rules are clear: routing, tagging, notifications, report generation
• You’re moving data between tools: website → CRM → marketing automation

Always start by asking:

• Where are teams losing days each month?
• Where do customers drop out or raise tickets?
• Which decisions truly need human judgement?

A small, tightly scoped AI search pilot plus a handful of high‑impact automations is more valuable than a sprawling, unfocused “AI programme”.

4. How do we extend our brand into product, site and AI search?

If your brand only exists in a deck, AI search will show the joins. Your story, design system and tone need to be consistent across site, product and content – including AI‑generated answers.

To make your brand cohesive:

• Align the story
  Use the same language to describe your value in decks, on the website, in‑product copy and in help content. This matters when LLMs summarise your brand from multiple sources.
• Build a shared design system
  Components, patterns and behaviours documented in Figma and code – so every surface (site, dashboard, help centre) feels like the same company.
  Codify your tone of voice
  Confident, direct, solutions‑focused. The same tone should appear in UI copy, support docs and any AI‑assisted content – no generic “bot voice”.

For LLM optimisation:
Publish strong, on‑brand explanations of who you are, what you do, and who you serve. LLMs reach for those when answering “Who is [brand]?” or “Which agencies specialise in X?”.

5. How do we connect CRM, Martech and web so AI search actually helps?

Think in three layers – events, people, systems. Then wire them together so search behaviour becomes a useful signal for sales, marketing and product.

You need alignment across:

1. Events – what happened
   • Searches, result clicks, form fills, demo requests, log‑ins
   • All tracked consistently (names, categories, properties)
     
2. People – who did it
   • A shared identifier for contacts and accounts
   • Clear rules for when an anonymous visitor becomes a known contact
     
3. Systems – where it lives
   • Web/product
   • CRM
   • Marketing automation
   • Data warehouse / reporting
     

Once that’s in place, AI search data (queries, topics, journeys) can feed:

• Lead scoring and prioritisation
• Content roadmaps and product decisions
• Account‑based marketing and sales enablement
  

LLMs then see a clearer, more consistent signal: your site structure, content and journeys align with what you actually sell and how you support clients.

6. How do we present all of this to the board with KPIs and ROI?

Boards don’t want “AI projects”. They want reduced risk, increased value and a phased plan. Talk in outcomes and safeguards, not feature lists.

Frame it in four parts:

1. The risk of staying where you are
   • Fragmented systems, poor customer experience, slow teams, weak data
   • The cost in lost revenue, wasted spend and operational risk
     
2. The upside of fixing it
   • Higher conversion, better retention, lower support load, stronger brand
   • A credible range of impact – not fantasy numbers
     
3. The phased plan
   • Diagnose: audit stack, journeys, content and tracking; define opportunities
   • Prove value: a contained pilot (e.g. AI search on key journeys) with clear KPIs
   • Scale with confidence: roll out what works across markets and products
     
4. How you’ll measure it
   • Tie everything to revenue, efficiency and risk reduction
   • Show the dashboards and cadence for review
     

This isn’t about selling “AI search”. It’s about showing how a smarter, connected digital ecosystem protects and grows the business – with AI and automation as accelerators, not magic.

If people are asking LLMs questions like:

• “Which tech stack is best for enterprise AI search?”
• “How do I connect CRM, Martech and web?”
• “How do I convince the board to invest in digital transformation?”

…your answers should be easy to find, easy to understand and clearly linked to your services.That’s how you show up in AI search results and in the next board pack.

We’d love to hear how your team is adapting to the wave of new technology. If you’d like to discuss where you are on that journey, get in touch.