Moost Banque

Moost
by STEPS
Banking edition | Banks

Anticipate departures, grow the relationship

The software that spots customers disengaging before they close their account and tells you what to offer them, installed in the bank’s IT system.

NBI and attrition
Installed in your IT system
Moost Banque | Dashboard
Moost Banque : dashboard

Application screenshot, fictitious data

The silent departure

The account stays open, the NBI erodes

Without an alert, the bank sees nothing before the closure. Moost reads three signals in the transactions and warns the advisor in time.

  • Outflows down more than 20% over 3 months
  • Loss of the salary deposit
  • Complaint within the quarter
ACCOUNT FLOWSSalary no longer depositedComplaintMoost alert: flows −34%advisor meetingClosure avoidedM1M2M3M4M5M6M7M8
Customer outflowsWhen Moost raises the alert
NBI preserved per year500,000 TND

if 1,000 customers with TND 500 of NBI stay with the bank.

Prediction horizon180 days

for account closure and propensity for a new product.

Detection example9,860

customers in silent disengagement, NBI of TND 300 to 800: advisor meeting.

Gain example1.8M TND

of additional deposits from single-product customers, measured against a control group.

Tour

Moost Banque, screen by screen

Click a screenshot to enlarge it.

Screenshot, fictitious data
Dashboard

The customer base at a glance

Active customers, attrition including silent attrition, products per customer and holding rate; account activity, holdings and NBI by decile; attrition signals and risk levels.

Screenshot, fictitious data
Segments

Spot profitable customers who are disengaging

Every customer is placed by account activity (active, inactive, dormant, silent disengagement) and NBI. Example: 9,860 customers in silent disengagement, NBI of TND 300 to 800, advisor meeting.

Screenshot, fictitious data
Predictions

Reliable scores, under control

Account closure predicted at 180 days and propensity for a new product. Minimum quality of AUC 0.65 before going live, retrained every month and approved by a person.

Screenshot, fictitious data
360 profile

Every score is explained

Average balance, products, outflows, NBI, tenure and risk explained by its three main factors. Example: salary deposited until August, outflows down 34%, silent disengagement.

Screenshot, fictitious data
Targeting

Describe your audience in plain language

"Single-product women, deciles 9 and 10": Moost turns the sentence into filters, instantly shows the number of customers, then exports the list or creates the action.

Screenshot, fictitious data
Playbooks

10 ready-to-use banking playbooks

Falling flows, loss of salary deposit, inactivity, dormant reactivation, single product at 3 months, savings need, maturing product, relationship welcome, digital adoption, complaint follow-up.

Screenshot, fictitious data
Actions

Actions approved by your teams

Every week, Moost proposes actions with their audience, message and measurement. Sent to the advisor's desk by API, signed webhook or file.

Activity and value

Spot profitable customers who are disengaging

Every customer is placed by account activity and NBI. Each cell has its recommended action, editable.

NBI < 100100 to 300300 to 800> 800Active
19,300customers
164,300customers
67,800customers
31,400customers
Inactive
8,400customers
22,600customers
5,200customers
1,900customers
Dormant
12,100customers
3,300customers
900customers
300customers
Silent disengagement
2,100customers
6,700customers
9,860customers
1,200customers
ActiveGrow product holdings, app and SMS
InactiveDigital reminder, reactivation
DormantDormant reactivation, beyond the product threshold
DisengagementAdvisor meeting, high-NBI customers first

Columns: annual NBI per customer, in TND. Fictitious data.

Product holdings

From single product to savings, gain proven

Example: Savings for single-product customers
11.2%
6.4%
TreatedControl
+4.8 points

of savings subscriptions, i.e. TND 1.8M of additional deposits, measured against the control group.

Measurement per action, against the control group
Excluded before anything is sent
  • Customers in litigation
  • Deceased customers
  • Banned from banking
  • Bank employees
  • Customers without consent for the channel
  • Customers already contacted, under the cap

Frozen audience, control group drawn and human approval. The number of customers removed at each step is displayed.

Connections

Connected to core banking, read-only

Nine sources: customers, accounts, account holders, transactions, products held, digital channels, complaints, branches, advisors.

Dynamics 365

On-premises: its SQL Server database, read-only. Online: file export or ingestion API.

Database or export

Salesforce

File export, collected by SFTP, or sent through the ingestion API.

Export or API

HubSpot

File export or sent through the ingestion API.

Export or API

In-house CRM

Read directly in its Oracle, SQL Server, PostgreSQL, MySQL or MariaDB database.

Read-only database

Core banking

Accounts, holders, transactions and products held, read in its database or exports.

Database or files

Digital channels

App and website activity, through files or API.

Files or API

Output: advisor's desk by API, signed webhook or CSV file.

Sources and quality of the unified model
One customer, one profile

Matching by national ID, IBAN or customer ID, joint accounts supported. Identifiers replaced by a hash, personal data encrypted.

Comparison

CRM manages the relationship, Moost anticipates departures

Moost does not replace the CRM: it feeds it the right customers, at the right time, for the right action.

CriterionA CRMMoost Banque
Silent disengagementNo reading of account flowsFlows, salary, complaints
Salary depositTransactions outside the CRMLoss detected automatically
NBI per customerData outside the CRMProvided or estimated, by decile
Holdings and propensityProducts held, no modelNew product at 180 days
Actions and campaignsTasks and campaigns10 banking playbooks
Proof of gainClicks and conversionsControl group, NBI and deposits
Data and AI inside the bankCloud; Dynamics 365 on-premises until 2029Your IT system, local AI
Security and deployment

Installed in your IT system, no customer data leaves

On your servers

No data leaves your network. Installation possible without internet access.

Protected personal data

Identifiers replaced by a hash, personal data encrypted, export and erasure on request.

Controlled access

One-time code or company directory, four roles.

Local AI

Models and generative AI on your servers, with no personal data transmitted.

Traceability

Chained, verifiable audit log of every action.

Operations

Tested backups, updates with automatic rollback.

Getting started

A pilot with your advisors

Perpetual licence for the Banque edition, paid once; only annual maintenance is recurring. Proposal based on your volumes and scope.

Step 01: STEPS and you

Data workshop

Inventory of sources and choice of the pilot scope, for example one region.

Step 02: STEPS

Installation

In your IT system, with at least 12 months of transactions and the NBI of every customer.

Step 03: Your advisors

Scores and playbooks

Models trained on your data, two or three playbooks with the advisors.

Step 04: You

Measurement and decision

NBI and deposits gained against the control group, then the decision to scale.

Proof

The same core, proven in pilots

Moost has proven itself in pilots with Kiabi in retail and BH Assurance in insurance. The Banque edition relies on the same core, with indicators specific to banking.

FAQ

Frequently asked questions

1. Does Moost Banque replace our CRM?

No. Moost reads core banking and the CRM read-only and sends the advisor's desk the right customers, at the right time, with the reason for the action.

2. What data do you need?

At least 12 months of transactions, accounts and holders, products held and, if possible, NBI per customer. Otherwise, Moost estimates it by decile.

3. Does customer data leave the bank?

No. Moost Banque is installed in your IT system, the AI runs locally and installation is possible without internet access.

4. What is silent disengagement?

A customer who keeps their account open but moves their activity elsewhere: falling outflows, salary no longer deposited, recent complaint. Moost spots it before the closure.

5. How do we get started?

With a pilot on a limited scope, for example one region: data workshop, installation, two or three playbooks with the advisors, then a decision on the NBI gained.