ferent.aiPortfolio
Customer success · Signals + Renewals

See the risk. Price the renewal. Grow the revenue.

Portfolio is one product with two halves. Signals owns the four inputs to a health score (product usage, support history, survey feedback and relationship quality) and collects all four itself, where most platforms buy three of them in. Renewals takes it from there: the quote, the contract, and the four outcomes that add up to your net revenue retention.

Request a demo Follow one morning, screen by screen
579
API paths shipped
4
Signal sources, native
0
Black-box scores
Customer health components, with each weight and threshold shown
Live capture · health components and weights
The problem

Nobody trusts a health score they cannot check.

Usage comes in through an analytics vendor. Support comes through a connector. Feedback lives in a separate survey tool. By the time a CSM sees the score, the working behind it is spread across four systems they cannot open. So they rebuild it by hand, and their VP asks for an update.

Consequence one
The CSM rebuilds the case

Five tabs open to justify one number, every time a renewal comes up.

Consequence two
The VP waits for a status update

Leadership only sees the portfolio after someone writes it up.

Consequence three
Nobody trusts the prediction

A churn percentage arrives with no model named and no confidence figure, so nobody can weigh it.

How it works

Five stages, start to finish.

Data arrives, gets standardised, scores against your settings, ranks against the rest of the book, and reaches whoever is allowed to act on it. No step waits for someone to write a report.

01 Ingest
We read the sources

Your warehouse, your support desk, your meetings and email, and our survey engine.

02 Normalise
One comparable reading

Every source becomes one reading per account, per product, per day. That gives you a trend.

03 Score
Against your configuration

You set the components, weights and thresholds. Four are computed, one is a human judgement, and all five are visible next to the score.

04 Predict
Ranked, with the working

Probability and tier, with the confidence, the contributing factors, the ARR at stake and the model version.

05 Act
Governed by role

Work goes to the person allowed to do it. Anything above their permission level goes up a level.

What never happens
Nobody files a status update

A CSM works the account and their VP watches the whole book. Neither had to ask the other for anything.

Where the AI sits
Recommending, not sending

Mira drafts and ranks. She never acts on her own.

One morning in Signals

Two red accounts, four days apart, one CSM.

A real morning in a demo workspace, followed at two altitudes: Jordan owns both accounts, Sam runs the function and owns none. Every screen below is a capture from the running product, in the order they were opened.

Jordan Lee In the accounts
08:40 The queue names both problems before he searches for anything
08:52 Pioneer, renewing tomorrow. One critical risk, with money attached
08:55 Thirteen tabs, assembled from what this customer licenses
09:15 Why the score is 42: components, weights, and a human assessment shown as its own row
09:25 Adoption read from the customer’s own warehouse. Acme down twelve
09:35 Churn ranked and dated, with a confidence figure per row
09:50 The sponsor scored them five out of ten. Same person named in the risk
10:15 He writes the save plan. The plan reports itself as at risk
11:30 Two renewals, four days apart. He has time to do one properly
Sam Rivera Above them
08:30 The whole book at once: retained, at risk, renewing
09:10 The same two accounts, already ranked, with the arithmetic shown
09:30 Moves the account. Jordan could not have, and sees why
Neither of them built a report. Sam saw Jordan’s problem at ten past nine. Not because Jordan escalated it, and not because anyone asked for an update. It was already in the record.
A queue, not a status board
08:40 Jordan
A queue, not a status board

$590K at risk across two open risks, and both accounts already named. He searched for nothing.

The account argues its own case
08:52 Jordan
The account argues its own case

Health, revenue and renewal date on one card. One critical risk, with $210K attached to it.

Thirteen tabs, assembled on load
08:55 Jordan
Thirteen tabs, assembled on load

Built from what this customer licenses and what his role allows. A customer without a module never sees its tabs.

Why the score is 42
09:15 Jordan
Why the score is 42

Components, weights and thresholds are the customer’s own. Four are computed; the CSM’s own read of 70 carries fifteen per cent as its own row.

Adoption, from their warehouse
09:25 Jordan
Adoption, from their warehouse

Read directly, with no analytics vendor in between. Acme down twelve points while everything else climbs.

Churn ranked, dated, and named
09:35 Jordan
Churn ranked, dated, and named

Acme 74% critical, Pioneer 66% high, a confidence figure per row, and the model named with its run time.

The sponsor scored them 5 of 10
09:50 Jordan
The sponsor scored them 5 of 10

The same person named in the account’s open risk. It arrived through a governed survey, not a forwarded email.

The plan reports itself at risk
10:15 Jordan
The plan reports itself at risk

Objectives with owners, tasks with dates, due the day before the renewal it protects. The plan reports its own status as at risk.

Two renewals, four days apart
11:30 Jordan
Two renewals, four days apart

Pioneer today, Acme in four days. Both at risk, both improving, both under 60. He has time to do one properly.

The book, not an account
08:30 Sam
The book, not an account

Retained, at risk, renewing, and whether the work underneath is happening. Nothing was requested from Jordan.

The same two, already ranked
09:10 Sam
The same two, already ranked

Pioneer 82, Acme 79. Acme is the worse risk and the bigger account; Pioneer still comes first, because it renews today.

The decision came up a level
09:30 Sam
The decision came up a level

Same queue, same numbers, one difference. Jordan can read the priorities and act on them. He cannot recompute them.

How health scoring works

Five components, weighted the way you decide.

You own the components, the weights and the thresholds. Four are calculated from data. The fifth is the CSM's own assessment, kept as its own row, so their read stays visible instead of being blended into another component.

Computed
Adoption

Read straight from your data warehouse, at one reading per account, per product, per day.

Computed
Engagement

Meetings and email we capture ourselves, with influence and role tracked per contact.

Computed
Support

Ticket volume, severity and SLA performance, taken directly from your support desk.

Computed
Sentiment

NPS, CSAT and CES from our survey engine, on a rolling ninety-day window.

Human
CSM assessment

What the person who knows the account thinks. Weighted like the rest, but kept as its own row rather than folded into another component.

A real account, scored
Composite health · 42 of 100
0 50 100 COMPONENT CONTRIBUTION Adoption 38 30% WEIGHT Engagement 35 20% WEIGHT Support 40 20% WEIGHT Sentiment 35 15% WEIGHT CSM assessment 70 15% WEIGHT · HUMAN

Four computed components, then the CSM’s own read at seventy. It carries fifteen per cent like the rest: the difference is that it is a person’s judgement, shown as its own row rather than folded into one of the others. Check it: 38×.30 + 35×.20 + 40×.20 + 35×.15 + 70×.15 = 42.

Churn predictions, each with tier, confidence, factors and the named model
Live capture · churn predictions

Every prediction comes with its working.

Probability and tierRanked across the whole book
ConfidenceA confidence figure on every row
Contributing factorsNamed, with the evidence behind each
ARR exposedAttached to the risk when it is raised
Model version and runNamed and timestamped on the page

The page says what the prediction is for: deciding what to work on first, not declaring a customer lost.

Integrations and the data model

From signal to decision.

Your CRM stays your system of record. Signals adds the path from the event to the person who needs to know about it.

Three loops, each one traceable from source to outcome
Survey answered Feedback recorded Sentiment component Health score moves
Ticket breaches SLA Support component Risk opened, ARR attached Renewal forecast changes
Usage drops in your warehouse Adoption component Churn model reruns Success plan objective

None of these wait for someone to notice. Building the same paths from separate tools means three integrations to maintain. One competitor acquired a company just to get the survey-to-health link.

And one on the roadmap
Onboarding completes Project CSAT fires Crediting rule CS bonus is calculated

Portfolio shares its survey and compensation engines with the platform, so one event can move a health score and feed a bonus calculation. Onboarding as a delivery module is still being built; the NRR and renewal-outcome bonus calculation runs today.

Your warehouse

Direct database connectors, including PostgreSQL. No analytics vendor in the path.

Your CRM

We read accounts, contacts and opportunities. Writes go back only for fields you enable, and only after someone approves the change.

Conversations

Teams, Zoom and Slack contribute metadata and transcripts. Recording comes from our own recorder, named in the room, and only after someone attests to consent.

Support and documents

Freshdesk, Zendesk, Jira Service Management and ServiceNow CSM feed health directly. SharePoint documents stay where they are. We hold a pointer and your permissions still decide who opens them.

The second half · Renewals

Knowing which account is only half the job.

Everything above tells Jordan which account to work. What follows is how he acts on it: the quote, the contract, and the four outcomes that make up your net revenue retention. Same product, same record, no handoff in the middle.

A renewal starts from the contract you already have.

It is a loop, not a pipeline. The subscription on the books is the starting position, health tells you which way it is heading, the quote is seeded from the lines already there, and signature closes one term and opens the next.

1 · The position you hold

Contracts and subscription lines, with what each one is worth and when it comes up. Nothing starts from a blank quote.

2 · Price the renewal

The quote is seeded from the lines already on the subscription, then priced against your price books and rules. Discounts route for approval in the same flow.

3 · Close it

Eight stages to signature, with the proposal built from versioned templates so the sent and signed versions share a history. At-risk status is computed from health, not read off a calendar.

4 · Account for the movement

One of four outcomes lands: renewed, expanded, contracted or cancelled. Each writes an audited ARR movement, and the NRR and GRR bridges trace back to the records behind them.

Where the two halves meet.

A renewal is where the two halves of Portfolio meet. Renewals prices and forecasts it, Signals scores the health behind it, and the shared compensation engine pays out on it.

Health drops in Signals Renewal posture turns at risk Forecast category changes The CRO sees it this week

Nobody rekeyed anything or wrote a report. The forecast changed because the record did.

Governance
Mira drafts. A person approves. Nothing reaches your customer otherwise.

That boundary, the row-level isolation under it, and who else touches your data are all on one page rather than scattered through the product pages.

Read Trust and security
Capabilities

Everything in Portfolio.

Two halves of one product. Signals ends where a decision is needed; Renewals begins there.

Signals Know which account, and why
Understand
Customer 360
One record: contacts, contracts, products, history
Product adoption
Licensed versus active, read from your warehouse
Support signal
Ticket volume, severity and SLA behaviour, read directly
Surveys and NPS
NPS, CSAT and CES on a governed question engine
Conversation intelligence
Sentiment, key topics and contact influence
Meeting capture
Native recording with consent and retention rules
Judge
Health scoring
Five components, your weights, all of them visible
Churn prediction
Probability, tier, confidence, factors, ARR exposed
Risk register
Severity, category and money attached to each risk
Portfolio intelligence
The whole book ranked at once
Segmentation
Cohorts that follow the data
Stakeholder roles
Influence and relationship role per contact
Act
Success plans
Objectives, owners, metrics and mutual commitments
Playbooks
Templates with trigger types and ordered steps
Onboarding
From the closed-won handoff to first value
Alerts and digests
Timed email digests, including pre-renewal windows
Mira, the assistant
Briefings, Q and A, and skills you write yourself. She drafts and stops. Sending is a separate step a person takes
Prove
Expansion signals
Scored opportunities through a real pipeline
Dashboards
The portfolio at whichever altitude you work at
Feedback record
Every response kept and attributable to an account
Lifecycle handoffs
Sales to success, recorded
Survey delivery
Governed links, with the privacy mode set when the survey was built
Audit trail
Who changed what, enforced below the application
Renewals Then price it, renew it, and account for it
Price
Quotes
Configured, priced and versioned
Proposals
Generated from templates you control
Clause library
Reusable, versioned contract language
Agreements and SOWs
Templates that carry through to delivery
Renew
Renewal stages
Eight of them, with probability per record
Forecast categories
Five, computed from the record
At-risk posture
Derived from account health
Subscriptions
Terms, amendments and the contract behind them
Account
ARR snapshots
A defensible point-in-time position
Movement ledger
Expansion, contraction and churn, audited
NRR and GRR bridges
Traceable to the records that moved them
Incentives
Renewal-outcome incentives and CS bonus calculation, with versioned payout rules

Two hundred and forty-two API endpoints across the two halves · every row above ships today

Request a demo

Bring your hardest account.

We will pull up its health score and go through the components with you. Forty-five minutes on your data.

01 A working session rather than a pitch
02 Every claim on this page, shown running
03 Security review material sent ahead if you need it

We only use this to set up the call. No sequence, no newsletter.