Anne is Head of Marketing at AI Findr. She has led marketing for both multinational corporations and high-growth startups, including Run:ai (acquired by NVIDIA), Sage, Arthur Cox, Fields the Jeweller, and Boxever (now Sitecore CDP).
What's inside
Churn starts long before the customer closes the tab.
The insurance industry suffers quote abandonment rates of 84%,¹ neck and neck with application abandonment in banking for the highest abandonment rates across all sectors. Not because people don’t want coverage or credit cards, but because they can’t find the answers they need to confidently understand and compare these products. And yet, many digital teams don’t see the connection between search and customer drop-offs.
Customers are frustrated by dead-end searches, irrelevant results, and clunky navigation. In fact, 49% of customers who left a brand to which they’d been loyal in the past 12 months say it’s due to poor CX.²
As expectations for clarity and speed grow, digital teams urgently need to fix search and product discovery or risk continued loss of revenue from missed opportunities.
Keyword search fails in banking and insurance
Keyword search wasn’t designed for financial decision-making. In banking and insurance, where many terms are unfamiliar to the average consumer, relying on keywords alone is a major problem.
Why it fails:
- Boost and bury logic surfaces the wrong products. A mortgage applicant searching “low-risk home loan” may be shown high-margin HELOCs instead of fixed-rate options.
- Metadata gaps hide key offers. A user searching “car insurance for teens” might never see bundled family plans if tagging is inconsistent.
- No synonym handling. Users typing “home loan,” “retirement plan,” or “income coverage” may miss results if your system only understands “mortgage,” “annuity,” or “disability insurance.”
- No intent modeling. A freelancer looking for a checking account shouldn’t get the same results as a high school student.
- No real-time logic. Keyword search doesn’t adapt to changes in eligibility, rates, or underwriting unless updated manually—a serious risk in regulated environments.
- Personalization software and search are separate systems. Even logged-in users are treated like strangers. Most search systems return the same results for everyone.
This kind of rigid interaction pushes users away, often toward public AI chatbots that promise simpler answers.
Public AIs increase the risks of inaction
Given the many customer pain points with keyword search, it’s no surprise that less than 1% of banking website visitors use on-site search. Instead of fixing the problem, up to 25% of major banks have removed search entirely, leaving users with no way to navigate complex offerings.³
When they can’t find answers, customers turn to Large Language Models (LLMs) like ChatGPT, Perplexity, or Claude. But those tools come with real risks:
- No regulatory guardrails. A hallucination of insurance exclusions might not be your legal responsibility, but it’s still your customer’s poor outcome—and your brand they remember.
- No live data. Public LLMs don’t reflect current rates, policies, or eligibility rules.
- No verified product information. These models don’t understand your differentiation and may favor competitors with more LLM-optimized content.
Worse, your competitors can shape the narrative. If someone asks ChatGPT, “Why is Bank X better than Bank Y?” and your competitor has published a comparison page, you’ve already lost the conversation.
LLM optimization (LLMO) is important (more on that below), but it’s no substitute for keeping customers on your site with helpful, compliant, up-to-date answers.
AI search and product discovery for regulated industries
AI Findr replaces keyword search with conversational discovery. Built on a large language model, AI Findr understands natural language, intent, and context. That means:
- A user types: “apartment insurance for WFH”
- AI Findr clarifies: “Are you working from home for your own business or remotely for an employer?”
- Business rules apply: coverage options that match your guidelines, promotions, or region appear first
- Content adjusts in real time, surfacing location-specific offerings and timely options
The system adapts not to what was typed, but to what was meant.
It starts with ingestion. AI Findr pulls in your content (like policy docs, disclosures, FAQs, blogs) and structures it into a machine-readable format. This becomes the brain behind both conversational search and business logic.
And because AI Findr is an LLM itself, your content is already formatted for visibility in public tools like ChatGPT.
When search works, customer drop-off improves
Financial institutions using AI Findr see:
- 15% average increase in application completions
- 52.6% more fully digital sales (no salesperson intervention)
- 200% improvement in CSAT from self-serve customer support
- 39-point average increase in NPS
Beyond KPIs, AI Findr gives digital teams more control. It makes your products more approachable, your content more useful, and your search strategy directly tied to revenue.
LLMO without the busywork
Public AI search is here to stay. But you don’t need to rewrite every product page or flood your site with keyword-stuffed FAQs to be cited.
Because AI Findr already structures your content into a centralized knowledge base, its sister product, LLM Findr, repurposes that same data and optimizes it for public LLMs.
LLM Findr surfaces your content in the right formats (like JSON-LD and llms.txt), tracks citations, and helps ensure that compliant, accurate answers are what customers see, even if they begin their journey off-site.
AI Findr helps users get answers on your domain.
LLM Findr makes sure they’re still hearing your voice when they ask AI tools for help.
Final thoughts
Customer drop-offs start when your search fails. AI Findr rethinks the discovery experience around how financial customers actually talk, research, and decide.
And with LLM Findr extending that foundation to public LLMs, you cover both sides of the buyer journey, on-site and off.
Footnotes:
1. Statista
2. Survey commissioned by Emplifi
3. In-depth study of search and chatbots on more than 200 major banking institutions conducted by AI Findr.
