Cortex: AI that knows a real lead from noise.
Octrackit Cortex is the AI lead quality scoring engine inside the platform. A dedicated model trained per brand that decides which marketing signals are real and worth acting on, filters out the junk, and tells you exactly why.
Every analytics tool counts every click as equal
A bot, a ghost click, and a serious buyer all land in the same report, weighted the same. So teams chase junk leads, sales time leaks, and budget flows to whatever produced the most noise. Nothing in a standard dashboard separates real intent from the rest.
Score, decide, and get smarter every time
Cortex turns each raw signal into a confident, explainable decision.
Ingest the signal
Every marketing signal and lead arrives in real time, tagged with its platform and value context.
Score the quality
The model scores each signal across multiple dimensions, from ghost identities to bot velocity to poisoned sources.
Dispatch or suppress
It decides to act on the signal or filter it as junk, each decision carrying a confidence score and a risk level.
Learn from outcomes
When a deal is won or lost, that final business truth becomes a label that retrains the model, so accuracy compounds.
The quality signals it reads
Each lead is scored across several dimensions, in plain terms:
Ghost score
Detects fake or ghost identities and paid clicks with no real behaviour behind them.
Velocity / bot detection
Flags bot and suspicious activity by comparing behaviour against normal human patterns.
Source poisoning & junk rate
Detects polluted, low-quality lead sources before they flood your pipeline.
DNA risk score
Builds an overall risk profile for each lead from the full set of quality signals.
Platform & deal value
Weighs the channel (Meta, Google, organic, CRM, direct) and the value context of the lead.
Every brand gets its own Cortex
A dedicated model is trained per client, not a shared one, so scoring reflects your audience, your channels, and what a good lead looks like for you. The longer it runs on your data, the sharper it gets.
The Cortex Strength Score shows how mature your Cortex is, moving from Initializing to Learning to Stabilizing to Strong as training data, accuracy, and balance improve.
It tells you why, not just yes or no
Cortex surfaces the top risk factors and the main reasons signals were suppressed, so a filtered lead is never a black box. You see the reasoning behind every call.
A cleaner pipeline you can actually trust
Stop chasing junk
Your team works real intent instead of burning hours on fake and low-quality leads.
Trust your pipeline
What reaches your CRM has been scored and filtered, so the numbers mean something.
Fund what sends real intent
Channel and decision data shows which sources send signal, not noise, so budget follows quality.
What is Cortex?
Octrackit Cortex is the per-client AI lead quality scoring engine in Octrackit, built by Octmark Technologies. For each incoming marketing signal or lead it makes a decision, DISPATCH (act on it) or SUPPRESS (filter it as low-quality or fake), each with a confidence score and a risk level. A dedicated machine-learning model (LightGBM) is trained per client, so every brand gets its own Cortex rather than a shared model. It scores signals across multiple quality dimensions, including a ghost score for fake identities, a velocity score for bot or suspicious activity, source poisoning and junk-rate detection, an overall DNA risk score, and platform and deal-value context. It is fail-open, so it always dispatches when it has no model yet or is not confident, and it learns continuously by retraining on real business outcomes (deals won or lost). A Cortex Strength Score from 0 to 100 (Initializing, Learning, Stabilizing, Strong) shows how mature the model is, and it surfaces insights explaining why each signal was trusted or filtered.
Cortex, answered
What is Cortex?
Cortex is Octrackit's per-client AI engine for lead quality scoring. It is a machine-learning model that decides which incoming marketing signals are real and worth acting on and which to filter out as junk or fake, then learns continuously from real business outcomes.
How does Cortex decide which leads to keep?
For each signal it makes a clear decision, DISPATCH (act on it) or SUPPRESS (filter it), each with a confidence score and a risk level of low, medium, or high. The decision is based on quality dimensions including ghost score, bot or suspicious velocity, source poisoning, an overall DNA risk score, and the platform and deal value.
Does each client get its own AI model?
Yes. A dedicated machine-learning model is trained per client, so every brand gets its own Cortex rather than a shared model. That means scoring reflects your specific audience, channels, and what a good lead looks like for you.
Will Cortex suppress good leads?
No. Cortex is fail-open. For a brand-new client with no model yet, or whenever the model is not confident, it always dispatches the signal. It never suppresses good leads when it is unsure.
What is the Cortex Strength Score?
The Cortex Strength Score is a 0 to 100 measure of how mature and trustworthy your Cortex is, with labels Initializing, Learning, Stabilizing, and Strong. It reflects training data volume, model accuracy (AUC), feature richness, and class balance, so you can see how far the model has come.
How does Cortex improve over time?
It learns from real business outcomes. When a deal is won or lost, that final business truth becomes a label that retrains the per-client model, so accuracy improves the longer it runs on your data.
See what your Cortex is filtering.
Spin up a Cortex for your brand and watch it separate real intent from noise, with the reasoning behind every call.