Conversation & Lead Intelligence

Lead Intelligence 2.0

An AI model trained for inbound and outbound sales, powered by AntVenture.

Available through the Interact and Influence products of AntVenture, so your teams can close more deals and bring lost leads back to life.

Learn more at antventure.ai →

Traditional scoring vs Lead Intelligence 2.0

How a static, CRM-style lead_score / MQL / lifecycle_stage compares to a conversation‑native signal graph once all of the Lead Intelligence 2.0 keywords are applied to the same conversations.

Traditional lead score / MQL

Lead Intelligence 2.0 intent curve

lead_score, MQL, lifecycle_stage, opp_stagevsCGS, PIW, CCI, ODM, GPS, RLLS, PEA, CLI, RES, IAG, CCR…

Vertical: signal strength · Horizontal: conversation turns

First replyDiscoveryObjectionsPeak intentHandoverOutcome

Traditional: one flat score for the whole journey

Legacy stacks compress website behaviour, form fills, email events, and CRM fields into a single lead_score / MQL / lifecycle_stage. It nudges upward over time, but almost never reacts to what was actually said in chat.

On the graph, that’s the grey line: a slow, almost linear climb that can’t show missed shots, late replies, or whether the “hot lead” is already gone.

Lead Intelligence 2.0: a live intent & control curve

Once CGS, PIW, CCI, ODM, GPS, RLLS and the outbound terms (PEA, CLI, RES, IAG, CCR…) are applied, the same conversation becomes a high‑resolution signal graph.

You can see exactly where intent spikes, when objections stack up, where control is lost, and which “lost” leads are still statistically recoverable.

Deep lead measurement language

The signals most teams never measure

Below are the lead and outbound algorithms this model tracks — each code is a distinct way it scores intent, control, objections, recovery and more inside conversations.

InboundLead Intelligence

CGS(Commitment Gradient Score)
PIW(Peak Intent Window)
CCI(Conversational Control Index)
ODM(Objection Density Map)
GPS(Ghost Probability Score)
HSD(Human Shot Detection)
BCCS(Bad Commitment Credit Score)
RLLS(Recoverable Lost Lead Score)

OutboundOutbound Intelligence

PEA(Psychological Entry Angle)
CLI(Cognitive Load Index)
SRT(Social Risk Temperature)
TRS(Temporal Relevance Score)
IAG(Intent Alignment Gap)
LFC(Language Familiarity Coefficient)
SRM(Synonym Resonance Map)
CCR(Curiosity-to-Commitment Ratio)
AFB(Authority–Friendliness Balance)
RES(Reply Ease Score)

Advanced lead benchmark for your team

Have a look at the advanced lead benchmark our AI model is trained to assist. It helps support teams, sales staff, and business developers to close more clients and get old lost business back to fully active.

Available for integration with

WhatsAppLinkedInInstagramWeb chatCSV importn8nMCPCRMCustom APIs

The framework behind the model

Below are the concepts the model is built on. They focus on how intent forms, peaks, stalls, or gets lost in real conversations—and how your outbound messages are perceived. Two sets work together:

Lead Intelligence — What happens inside the conversation (commitment, objections, missed shots, recoverable leads).
Outbound Intelligence — How messages land (cognitive load, reply ease, psychological angle, authority vs friendliness).

Browse example conversations →

1. Big picture: Lead × Outbound

Lead performance vs Outbound performance

High Outbound · High Lead

Clear, low-friction messages and strong handling of objections & peaks.

High Outbound · Low Lead

Good replies coming in, but team misses shots or loses control.

Low Outbound · High Lead

Team is strong in live chats, but many great leads never start the conversation.

Low Outbound · Low Lead

Hard-to-read messages and weak in-chat handling — both need work.

Conversation journey (simplified)

1
Outbound message sent
PEA, CLI, SRT, TRS, IAG, LFC, SRM, CCR, AFB, RES describe how this first message feels.
2
First reply & discovery
CGS begins to rise, CCI shows who is steering, ODM and GPS watch for objections and fading intent.
3
Peak intent & close
PIW locates the short window to act; HSD checks if the human took the shot or missed it.
4
Outcome: won, stalled, ghosted, or recoverable
GPS, BCCS, and RLLS separate lost-because-of-us vs lost-because-of-them — and which leads can still be recovered.

2. Lead Intelligence

How intent forms, evolves, peaks, stalls, or gets lost within conversations.

CGS

Commitment Gradient Score

Tracks how a lead progresses from curiosity to concrete buying intent through micro-commitment signals in conversation.

How it’s measured or calculated

Calculated by counting and weighting micro-commitments (e.g. sharing details, confirming preferences, accepting next steps). Each commitment type gets a score; the gradient is the trend over turns. Rising score = lead moving toward conversion; flat or falling = stalling.

CGS rising with each micro‑commitment

Lead: “Tomorrow evening, Dubai Marina. Budget around 6k.”

Agent: “Great. I’ll share 3 options and book a viewing if you like.”Gradient rising

The lead moves from vague interest to concrete details (time, location, budget) and accepts a next step. Each of those adds to the commitment gradient.

PIW

Peak Intent Window

Detects the short time window where a lead is most ready to convert based on conversational signals.

How it’s measured or calculated

Identified by signals such as asking for price, requesting location, or saying they’re ready to act. The window is the span of turns where these signals cluster. Measured by intent-strength indicators and response latency; a missed or delayed reply inside this window is tracked as a lost opportunity.

Peak intent handled in time

Lead: “Send location, I’ll come now.”

Agent: “Here’s the pin. Can you reach in 20 minutes?”Peak window

The message “Send location, I’ll come now.” marks a very short window where closing is easiest. The agent responds quickly with a concrete next step instead of delaying.

CCI

Conversational Control Index

Measures whether the business or the customer is actively driving the conversation toward a decision.

How it’s measured or calculated

Scored by who asks closing questions, proposes next steps, and narrows options. Agent-driven moves (e.g. 'Tomorrow or Saturday?') add to business control; prospect-only questions with no agent direction reduce it. Often expressed as a ratio or segment (e.g. control vs no control).

Agent keeps the conversation moving

Lead: “I’m looking for a 2BHK in Business Bay.”

Agent: “Got it. Tomorrow or Saturday works better for viewing?”High control

The agent acknowledges the request and immediately narrows to a next step with a choice question, keeping control and momentum toward a decision.

ODM

Objection Density Map

Analyzes the number, type, and resolution quality of buying objections raised during a conversation.

How it’s measured or calculated

Objections are tagged by type (price, trust, timing, etc.) and by whether they were resolved, partially addressed, or ignored. Density = count of objections per conversation or per turn. Resolution quality is scored from the follow-up messages (e.g. direct answer vs deflection).

Multiple objections, properly handled

Lead: “Price seems high, and I’m not sure about quality.”

Agent: “Fair point. Our price includes warranty and same‑day support. I’ll send 3 reviews so you can see results. If that feels safe, we can book.”Objections resolved

Two objections (price, trust) are raised and directly addressed with value explanation and proof, lowering objection density instead of letting them pile up.

GPS

Ghost Probability Score

Predicts the likelihood that a lead is about to disengage before conversion based on fading conversational signals.

How it’s measured or calculated

Uses signals like longer reply delays, shorter messages, and repeated acknowledgements without progress. A model or rule set scores these patterns; higher score = higher risk of ghosting. Can be updated turn-by-turn as the conversation continues.

Saving a lead before ghosting

Lead: “Ok.”

Agent: “Quick yes/no: should I reserve the slot for you, or close this chat so we don’t bother you?”Ghost risk handled

Short, non‑committal replies (“Ok.”) raise ghost probability. The agent reacts with a tiny yes/no decision instead of another big ask, giving the lead a low‑pressure way to stay or exit.

HSD

Human Shot Detection

Identifies moments where a human agent failed to ask for or guide the next decisive action.

How it’s measured or calculated

After clear buying signals (e.g. 'I want to book', 'Send the link'), the next agent message is checked for a closing move (e.g. two-option close, ask for details). If the agent instead sent something non-decisive (e.g. brochure only), that turn is flagged as a missed shot.

Missed shot after buying signal

Lead: “Ok, I want to book.”

Agent: “Here is our brochure.”Missed shot

Instead of using the buying signal to close (e.g. offering two time options), the agent goes backwards to generic info. HSD flags this as a missed shot.

BCCS

Bad Commitment Credit Score

Flags leads that showed strong intent but stalled due to missing or mishandled follow-up actions.

How it’s measured or calculated

Intent strength is scored from earlier turns; then follow-up quality is assessed (response time, relevance, handoff). Leads with high intent but slow reply, wrong handoff, or over-push get a worse score. Used to separate 'we failed them' from 'they weren’t ready'.

High intent, weak follow‑up

Lead: “I’ll come tomorrow, send me the location.”

Agent: “Sure, our manager will contact you tomorrow.”Bad follow‑up

The lead commits to coming, but the agent defers to a manager instead of sending the location and confirming a time. Strong intent but poor follow‑up hurts the BCCS.

RLLS

Recoverable Lost Lead Score

Identifies non-converted leads that were highly likely to convert but were lost due to internal response failures.

How it’s measured or calculated

Combines intent strength, peak-intent timing, and failure type (delay, missed close, objection left open). Leads that had high intent and a clear internal failure get a high recoverability score. Often used to prioritize re-engagement or process fixes.

Lead lost, but still recoverable

Lead: “Final, send the payment link please.”

Agent: “Sorry for late reply, here is our brochure.”Recoverable lost lead

The lead clearly wanted to pay, but response was slow and off‑topic. RLLS flags this as recoverable if you follow up fast with a direct link and apology.

Conversation timeline: CGS, PIW, objections, and ghost risk

CGS gradient

Peak Intent Window (PIW)

Turns 4–5: asks price & location→ act fast here

ODM: 2× price objectionsGPS: ghost risk ↑ after slow reply

Missed shot example (HSD)

Lead: “Ok, I want to book.”

Agent: “Here is our brochure.”Missed shot

After a clear buying signal, the agent should ask a closing question (e.g. “Tomorrow 6pm or Saturday 11am?”) instead of going back to generic info.

Recoverable vs non‑recoverable lost leads (RLLS)

Recoverable lost lead

CGS was high and PIW clearly appeared.
Internal delay or weak follow‑up caused the drop.
RLLS says: this is worth a targeted re‑engagement.

Not really recoverable

CGS stayed low; lead was just browsing.
No strong buying signals before they went quiet.
RLLS says: accept as natural drop‑off, not a process bug.

3. Outbound Intelligence

How messages are perceived, processed, and responded to—especially in outbound or broadcast communication.

PEA

Psychological Entry Angle

Identifies the psychological approach used to initiate a conversation, such as curiosity, empathy, authority, or efficiency.

How it’s measured or calculated

The opening message is classified into one or more angles (curiosity, empathy, authority, efficiency, fear, etc.) using keywords, structure, and tone. Can be rule-based or model-based. Often reported as the primary angle and whether it fits the audience.

Curiosity‑based entry

Message: “Quick one: are WhatsApp leads a big channel for you right now?”

Recipient: “Yes, almost all of them.”Curiosity entry

The opener is light, specific, and easy to answer. It enters through curiosity instead of a heavy pitch, making it safer to reply.

CLI

Cognitive Load Index

Measures how mentally demanding a message is to read, understand, and respond to.

How it’s measured or calculated

Based on message length, sentence complexity, jargon, number of distinct asks, and formatting. Shorter, simple, one-ask messages score lower (easier); long, dense, multi-ask messages score higher. Can use readability metrics and custom weights per channel.

High vs low cognitive load

High‑load message: “We provide multi‑agent orchestration, embeddings, and omnichannel funnels. Please share CRM, ads, budgets, and chat logs so we can run an audit.”

Low‑load message: “Yes/no: do you get many WhatsApp leads that never turn into bookings? If yes, I can send one screenshot.”CLI comparison

The first version is long, technical, and asks for many things at once. The second is short, concrete, and one‑ask — much lower cognitive load.

SRT

Social Risk Temperature

Estimates how socially risky or awkward it feels for a recipient to reply to a message.

How it’s measured or calculated

Scored from ask size (e.g. call now vs yes/no), pressure language, and permission framing. 'Yes/no' and 'want a screenshot?' lower risk; 'book a call now' or 'reply immediately' raise it. Often combined with channel norms (e.g. LinkedIn vs WhatsApp).

Low vs high social risk

Low‑risk ask: “Is it ok if I send you a 30‑second example clip? You can just reply ‘useful’ or ‘not useful’.”

High‑risk ask: “Can you jump on a call right now and share your number?”SRT contrast

The first makes it easy and safe to say yes or no. The second asks for time and phone number immediately, which feels much riskier to reply to.

TRS

Temporal Relevance Score

Evaluates how strongly a message answers the question 'why now' for the recipient.

How it’s measured or calculated

Presence and clarity of a time-based reason (season, campaign, recent event, deadline). Messages with a clear 'why now' (e.g. 'before your next campaign') score higher; generic or absent timing scores lower. Can be binary or a strength scale.

Clear vs weak ‘why now’

Strong TRS: “Before your Ramadan campaign starts next week, do you want a quick lead‑triage setup so your team doesn’t drown in chats?”

Weak TRS: “We do AI for chats, interested?”Timing anchor

The first line ties the message to a real upcoming event, making timing obvious. The second has no reason to act now.

IAG

Intent Alignment Gap

Measures the distance between the recipient's current intent stage and the action requested in the message.

How it’s measured or calculated

Intent stage is inferred or assumed (e.g. cold, aware, considering); requested action is mapped (e.g. reply yes/no, watch video, book call). Gap = mismatch between the two. Small ask for cold audience = low gap; big ask for cold = high gap. Often scored per message or per sequence.

Ask too big for the stage

Message: “You don’t know us yet, but approve AED 15k and share full WhatsApp access so we can start.”

Recipient: “No, that’s too much for now.”Big IAG

This asks for money and access immediately from a cold contact — a huge gap between current intent and requested action.

LFC

Language Familiarity Coefficient

Assesses how closely the message language matches the recipient's industry, role, and cultural vocabulary.

How it’s measured or calculated

Term overlap with role/industry lexicon, tone (formal/casual), and avoidance of alien jargon. Can use word lists, embeddings, or manual tags per segment. Higher score = message 'speaks their language'; lower = generic or off-vocabulary.

Speaking their language

Good LFC: “Do more ‘appointment requests’ turn into no‑shows, or do people drop off after asking price on WhatsApp?”

Poor LFC: “We deliver LLM‑based multichannel orchestration for business synergies.”Familiar vs alien words

The first uses words a clinic or sales team actually says. The second is generic AI jargon that doesn’t match most operators’ daily vocabulary.

SRM

Synonym Resonance Map

Maps which word choices emotionally resonate best with a specific audience segment.

How it’s measured or calculated

A/B or multivariate tests of phrasing (e.g. 'serious vs browsing' vs 'ready vs not ready'); engagement or reply rates by variant. Builds a map of which synonyms or frames perform best per segment. Often qualitative first, then validated with response data.

Testing synonyms in a message

Version A: “We flag ‘serious vs browsing’ chats so your team focuses on the right people.”

Version B: “We flag ‘ready vs not ready’ chats so your team focuses on the right people.”SRM test

Both versions mean the same thing, but one wording may emotionally resonate more for your audience. SRM tracks which phrasing performs better over time.

CCR

Curiosity-to-Commitment Ratio

Balances how much curiosity a message creates versus how much commitment it demands too early.

How it’s measured or calculated

Curiosity is scored from hooks, questions, and teasers; commitment from ask size (e.g. share data, book call). Ratio = curiosity score / commitment score. Healthy outbound often aims for high curiosity and low commitment early; ratio shifts as the sequence progresses.

Healthy curiosity vs heavy early ask

Good CCR: “Quick question: do you lose serious WhatsApp inquiries because replies are inconsistent? If yes, I can send one 45‑second clip.”

Bad CCR: “Share your CRM, budgets, and 3 months of chats so we can run a full audit and then do a 60‑minute call.”Curiosity vs commitment

The first line builds curiosity and asks for almost nothing. The second demands a lot of work before earning interest.

AFB

Authority–Friendliness Balance

Measures whether a message feels appropriately authoritative or approachable for the target audience.

How it’s measured or calculated

Tone and claims are scored on authority (expertise, proof, confidence) and friendliness (warmth, permission, humility). Balance is segment-dependent; B2B might favor authority, SMB might favor friendliness. Off-balance (e.g. too pushy or too vague) is flagged.

Too aggressive vs balanced tone

Too authoritative: “You are losing leads every day without our system. You must fix this now or you’ll regret it.”

Balanced: “We found a way to spot ‘ready’ leads in chat so your team doesn’t miss them. If it’s not relevant, I’ll stop here.”AFB contrast

The first line is all pressure and no warmth. The second shows expertise but also permission and respect for the recipient’s time.

RES

Reply Ease Score

Calculates how easily a recipient can respond with a short, low-effort reply.

How it’s measured or calculated

Based on whether the message ends with a clear, low-friction ask (e.g. yes/no, choose A or B, one word). Open-ended or vague endings score lower. Can include length of expected reply and number of steps. Higher score = easier to reply; linked to reply rates.

Easy vs hard reply endings

High RES: “Yes/no: do you want to see a 1‑minute example of how we flag ready‑to‑buy chats?”

Low RES: “Let me know what you think about improving your lead intelligence.”Reply ease

The yes/no ending is trivial to answer and scores high on RES. “Let me know what you think…” is vague and harder to respond to.

Message anatomy: PEA, CLI, RES, IAG

PEA – curiosity

“Quick one: are WhatsApp leads a big channel for you today?”

CLI – load

Keep this line short, concrete, and one‑ask to keep cognitive load low.

RES & IAG – reply ease & ask size

End with an easy close: “Yes/no: want one screenshot of how we flag serious vs browsing chats?” — ultra simple to answer.

Before / after outbound improvements

Before

“We guarantee 3× leads in 7 days. Share your CRM, ad spend, and team details today and let’s book a 60‑minute call.”

High CLI
Huge IAG
High SRT

After

“Yes/no: are WhatsApp leads important for you? If yes, I can send one screenshot showing how we flag ‘ready’ chats first.”

Lower CLI
Smaller ask (IAG)
Higher RES

Curiosity‑to‑Commitment Ratio (CCR)

Healthy outbound usually starts with high curiosity and low commitment, then slowly increases commitment as intent rises.

Too vagueHealthyToo demanding

5. “Which terms matter for my problem?”

If you’re asking…	Look at these Lead terms	And these Outbound terms
We get lots of chats but few actual bookings. What’s going wrong?	CGSPIWCCIODMHSD	PEACLIIAGRES
People ask for price or location, then disappear.	PIWODMGPSRLLS	CLISRTTRSRES
Our messages feel pushy or awkward. How do we fix tone?	—	PEASRTAFBIAG
We respond, but the team misses closing moments.	HSDCGSPIWBCCS	—

5. Example metrics & dashboards (illustrative)

These aren’t real numbers from your data, but they show how each term can be treated as a measurable signal rather than a vague idea.

Lead Intelligence snapshot

• Average CGS rise over 6 turns: +3.1
• Peak-intent windows hit on time: 62%
• Conversations with strong control (CCI): 48%
• Recoverable lost leads (RLLS high): 14%

Outbound Intelligence snapshot

• Messages with low cognitive load (CLI): 71%
• High reply-ease endings (RES): 64%
• Healthy curiosity-to-commitment ratio (CCR): 0.8–1.2
• Balanced authority–friendliness (AFB): 73%

Outcome patterns (illustrative)

• Success / soft_yes: 28%
• Follow-up needed: 39%
• Polite_no or blocked: 18%
• No_response / ghost: 15%

How the two sets work together

Outbound Intelligence shapes how conversations start and progress. Lead Intelligence evaluates what actually happens inside those conversations. Together they reveal where intent formed, where it peaked, where it weakened or stalled, and whether it can be recovered—shifting focus from surface metrics to decision quality inside conversations.

See example conversations →