Most businesses treat their contact form as an afterthought. It's the last widget added before a site goes live — a few fields, a submit button, and hope that something useful comes through.
We treated ours the same way. Then we replaced it with an AI-powered intake system, and the results were surprising enough that we think it's worth writing up honestly.
The problem with our old contact form
Our old form had five fields: Name, Email, Phone, Service Required (dropdown), and Message. It was clean, functional, and generated a steady stream of submissions.
The problem was what those submissions actually contained.
A significant portion were completely off-topic — people enquiring about services we don't offer, or about businesses we have no relation to. Others were so vague ("interested in your services") that following up required another round of back-and-forth to understand what the person actually needed. And a smaller number were clearly not genuine business enquiries at all.
The result was that every morning, going through the contact form inbox felt like sorting mail. Most of it wasn't actionable.
What we switched to
We replaced the form with IntentForm. Instead of fields, visitors see a single prompt: "Describe what you're looking for." They can type or speak their request in plain language.
We added a business context — a two-paragraph description of what our business does, who it serves, and what we don't do.
That's it. Setup took about 15 minutes including embedding on the site.
What changed immediately
The off-topic submissions dropped sharply. The AI compares each submission against our business context. Enquiries about services we don't offer get flagged as low-relevance. They still land in our system — we can review them — but they don't mix with the genuine leads. Our main inbox became genuinely useful.
The data quality went up. When someone describes their need in natural language, the AI extracts the key details — budget, timeline, location, what they actually want. By the time a submission reaches us, we already have structured information without asking for it explicitly. Visitors mentioned things in their description that we'd never have thought to put in a dropdown.
The first response time improved. With AI auto-reply enabled, visitors get an immediate response that references their specific request and explains what happens next. This alone reduced the "did you receive my enquiry?" follow-up messages to almost zero.
What surprised us
The voice input got used more than we expected. We assumed most visitors would type. A meaningful number tapped "Speak" and dictated their request — particularly on mobile. The AI processed voice the same as text, which meant no change to our workflow.
The urgency scoring turned out to be genuinely useful. Submissions flagged as High Urgency tended to convert at a higher rate when we followed up quickly. The AI isn't always right, but it was right enough that we started prioritising our callback queue by urgency score.
What we learned
Natural language input lowers the barrier to enquiry. Some visitors who wouldn't bother filling out five separate fields will happily write a sentence about what they need. The format feels less like filling out a government form and more like sending a message.
Business context is the most important configuration. The difference between a useful and a noisy AI system comes down to how well you've described your business. Vague context produces vague filtering. Specific context — what you do, for whom, and what you don't do — produces genuinely useful pre-qualification.
The AI doesn't replace your judgment — it saves time before you apply it. The highest-value submissions still need a human to read and respond thoughtfully. What changed is that we spend far less time on the ones that shouldn't have been there in the first place.
Worth it?
For our use case — B2B lead intake — yes, clearly. The improvement in inbox quality alone made it worth the change. Everything else was a bonus.
If your contact form is generating a high volume of off-topic or vague submissions, an AI intake system addresses that at the source rather than on your end.