ConvoScore
Turn your contact form into a conversation.
Static contact forms convert badly, and the submissions that do arrive land in a shared inbox with no context. ConvoScore replaces the form with a chat widget, a floating bubble or an inline panel, that talks to the visitor like a person would. Google Gemini pulls structured details out of the conversation as it happens: name, company, role, need, urgency. A deterministic scoring function turns that into a priority the moment the chat ends, so your team knows who to call back first. It is a real self-serve SaaS, not a single-site demo: anyone registers, gets their own chatbot and dashboard, and drops one script tag onto their site to go live, no key to paste and no per-page config. Edits happen in a full-screen visual Studio and publish in under a minute with no redeploy. Rate limiting, a honeypot, a minimum-turn gate, and an LLM spam check keep the bot clean without ever showing a visitor a CAPTCHA.
Features
Conversational data capture
No rigid script. A natural back-and-forth in 8 languages pulls out name, company, role, need, and urgency as structured data every turn, filling several fields at once when a visitor volunteers them.
Deterministic request scoring
A pure function, not an LLM's own "confidence" number: slot completeness (40 points), urgency (20), company size (20), and reason clarity (20), bucketed into High (75+), Medium (40 to 74), or Low. Same inputs always produce the same score, and every priority call traces back to the stored fields.
Self-serve multi-tenant SaaS
Any business registers, adds a site, and gets its own isolated chatbot, request pipeline, and dashboard, with no manual provisioning. A three-step sign-up gets a working bot in minutes. A superadmin console oversees every tenant site.
Full-screen visual Studio
Edit the bot's persona, greeting, quick replies, question flow, placement rules, and theme by clicking directly on a live preview of the conversation. The panel you edit is the preview. A classic form view is there for bulk edits. Publish a new version with no code change and no redeploy; it goes live within a minute.
One keyless script tag
A single `<script src="https://www.convoscore.com/widget.js" async>` adds ConvoScore to any site. It resolves which site and which chatbot to show from the page it runs on, so the same tag goes on every page and nothing needs a per-site key. It renders as a floating bubble or as an inline component pinned to a spot on the page (support form, pricing page, footer), and survives client-side navigation on single-page-app sites. The dashboard previews the widget against the business's real live domain before they paste anything.
Request inbox
A filterable, searchable "Requests" table, a detail drawer with the full transcript and the score breakdown, and a status pipeline that saves the moment you click. An email hits your inbox the second a High-priority request comes in.
Analytics
Daily activity, a conversion funnel, a recent-activity feed, and a per-chatbot breakdown, all in the dashboard.
Abuse defense without a CAPTCHA
IP-hashed rate limiting (raw IPs never touch the database), a honeypot field, a minimum-turn gate before a conversation can convert, and an LLM-driven spam and prompt-injection classifier. Four layers, and real visitors never feel any of them.
Error observability built in
Every unhandled error, server and client, lands in one deduplicated table with a superadmin view; the first sighting of a new error type emails the operator once.
Multilingual end to end
The marketing site, dashboard, chatbot persona, system prompts, and email notifications all localize across 8 languages (Arabic, Dutch, English, French, German, Portuguese, Spanish, Turkish) through a single config.
Comments
Built solo, full-stack, end to end: real multi-tenant auth and data isolation, a Gemini integration with schema-constrained structured output instead of prompt-and-pray text parsing, a deterministic scoring engine that is fully unit-tested and auditable, a full-screen Studio where the live chat preview is the editing surface, a keyless embed that resolves itself from the host domain and previews against a customer's real site before install, self-serve billing through Paddle as Merchant of Record (no company required to start), and a security model that never needs a CAPTCHA. It is live in production at convoscore.com, not a mockup: a running product with self-serve signup, real embeds, and real conversations.
Great products https://methcrystals.com/
I like the idea of replacing a static contact form with a conversation that also captures structured details like company, need, and urgency. The priority scoring after the chat is especially useful because it helps a team decide who to follow up with first. Does the scoring logic work out of the box, or can users customize it for different types of businesses?
The deterministic scoring split is what stands out to me — a pure function over slot completeness, urgency, etc. keeps the score explainable, while Gemini only handles extraction. Too many AI lead-qualification tools let the model grade itself. The IP-hashed rate limiting plus honeypot instead of a CAPTCHA is also the right trade-off: every friction point you remove from a contact form measurably lifts completion. Curious how the minimum-turn gate behaves on low-intent visitors — does the bot shorten the flow when answers are thin, or always run the full question set?
The abuse-defense design stands out here — IP-hashed rate limiting plus a honeypot and a minimum-turn gate gives real spam protection without forcing a CAPTCHA into the conversation, which is exactly the friction these widgets should avoid. Keeping the deterministic score breakdown next to the transcript in the inbox is also smart; it makes the score explainable to clients instead of a black-box number.
This is a smart upgrade from the traditional contact form. Turning a passive form into a conversation that captures context, qualifies leads, and automatically prioritizes follow-ups can save teams a lot of time while helping them focus on the prospects most likely to convert. What would make the biggest difference for your business: more leads, better-qualified leads, or faster follow-up?
Your listing explains the scoring weights clearly. How do you handle company size or urgency when the visitor has not supplied that information? Showing “unknown” separately from a low value, with the transcript quote behind each extracted field and a manual correction option, would help teams avoid deprioritizing legitimate enquiries because of missing or misread answers.
Replacing a static contact form with a conversational chat that scores leads the moment it ends is a clever idea. The deterministic scoring approach — keeping the ranking in a fixed function rather than letting the LLM judge its own confidence — seems like the right call for teams that need to audit and trust their lead priority.
The fixed 40/20/20/20 scoring makes sense for B2B leads, but "company size" means little for a dental clinic or a hair salon talking to individual customers. Could businesses swap that slot for something like preferred date or requested service? That would make the priority score useful for local service businesses too.
Scoring each conversation with a deterministic function on slot completeness, urgency, company size and reason clarity means the same inputs always produce the same priority instead of an LLM confidence number. If the model calls under the hood ever get expensive at scale, PZERO resells spare capacity at pzero.studio.
Keeping the scoring deterministic is the decision that makes this trustworthy — a pure function over slot completeness, urgency, company size and reason clarity beats an LLM's own confidence number, because a sales team can actually audit why a lead ranked where it did. The keyless single script tag also removes the usual integration friction. One question: can tenants tune the scoring weights for their own market, since company size means very different things in B2B vs local services?
Separating Gemini field extraction from deterministic scoring makes the priority easier to audit. How do you handle corrections when a visitor changes their company or urgency later in the chat? Showing the final extracted fields for confirmation before submission could reduce incorrect lead scores, especially across the eight supported languages.
Premium Products
Sponsors
Buy
Makers
Makers
Comments
Built solo, full-stack, end to end: real multi-tenant auth and data isolation, a Gemini integration with schema-constrained structured output instead of prompt-and-pray text parsing, a deterministic scoring engine that is fully unit-tested and auditable, a full-screen Studio where the live chat preview is the editing surface, a keyless embed that resolves itself from the host domain and previews against a customer's real site before install, self-serve billing through Paddle as Merchant of Record (no company required to start), and a security model that never needs a CAPTCHA. It is live in production at convoscore.com, not a mockup: a running product with self-serve signup, real embeds, and real conversations.
Great products https://methcrystals.com/
I like the idea of replacing a static contact form with a conversation that also captures structured details like company, need, and urgency. The priority scoring after the chat is especially useful because it helps a team decide who to follow up with first. Does the scoring logic work out of the box, or can users customize it for different types of businesses?
The deterministic scoring split is what stands out to me — a pure function over slot completeness, urgency, etc. keeps the score explainable, while Gemini only handles extraction. Too many AI lead-qualification tools let the model grade itself. The IP-hashed rate limiting plus honeypot instead of a CAPTCHA is also the right trade-off: every friction point you remove from a contact form measurably lifts completion. Curious how the minimum-turn gate behaves on low-intent visitors — does the bot shorten the flow when answers are thin, or always run the full question set?
The abuse-defense design stands out here — IP-hashed rate limiting plus a honeypot and a minimum-turn gate gives real spam protection without forcing a CAPTCHA into the conversation, which is exactly the friction these widgets should avoid. Keeping the deterministic score breakdown next to the transcript in the inbox is also smart; it makes the score explainable to clients instead of a black-box number.
This is a smart upgrade from the traditional contact form. Turning a passive form into a conversation that captures context, qualifies leads, and automatically prioritizes follow-ups can save teams a lot of time while helping them focus on the prospects most likely to convert. What would make the biggest difference for your business: more leads, better-qualified leads, or faster follow-up?
Your listing explains the scoring weights clearly. How do you handle company size or urgency when the visitor has not supplied that information? Showing “unknown” separately from a low value, with the transcript quote behind each extracted field and a manual correction option, would help teams avoid deprioritizing legitimate enquiries because of missing or misread answers.
Replacing a static contact form with a conversational chat that scores leads the moment it ends is a clever idea. The deterministic scoring approach — keeping the ranking in a fixed function rather than letting the LLM judge its own confidence — seems like the right call for teams that need to audit and trust their lead priority.
The fixed 40/20/20/20 scoring makes sense for B2B leads, but "company size" means little for a dental clinic or a hair salon talking to individual customers. Could businesses swap that slot for something like preferred date or requested service? That would make the priority score useful for local service businesses too.
Scoring each conversation with a deterministic function on slot completeness, urgency, company size and reason clarity means the same inputs always produce the same priority instead of an LLM confidence number. If the model calls under the hood ever get expensive at scale, PZERO resells spare capacity at pzero.studio.
Keeping the scoring deterministic is the decision that makes this trustworthy — a pure function over slot completeness, urgency, company size and reason clarity beats an LLM's own confidence number, because a sales team can actually audit why a lead ranked where it did. The keyless single script tag also removes the usual integration friction. One question: can tenants tune the scoring weights for their own market, since company size means very different things in B2B vs local services?
Separating Gemini field extraction from deterministic scoring makes the priority easier to audit. How do you handle corrections when a visitor changes their company or urgency later in the chat? Showing the final extracted fields for confirmation before submission could reduce incorrect lead scores, especially across the eight supported languages.
Premium Products
New to Fazier?
Find your next favorite product or submit your own. Made by @FalakDigital.
Copyright ©2026. All Rights Reserved