NewAI Autopilot: retention that runs itself

The retention & expansion platform for B2B SaaS

Keep the customers you win. Grow the ones who stay.

Banchurn reads your product's event stream as account signals: who's slipping toward churn, and who's ready to grow. Then it acts on each: a win-back journey for the at-risk, a heads-up in Slack for the expansion-ready. One platform, from signal to action.

Track · Predict · Score · Segment · Activate · Retain

app.banchurn.com/overview
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Overview /Retention overview
Search ⌘KN
Active customers
4,182+6%
At risk
214−12 wk
Emails / wk
38.2kA− avg
Revenue saved
$52k+9%
Top at-risk · act nowchurn model
M
Meridian Health
logins ▼ 6d
82
V
Vertex Labs
key feature cold
74
O
Orbit Systems
plan downgraded
61
Active customers · 12w
Live events
signed_upAna · Growth
feature_usedexport_report
risk.risingVertex Labs · 74
campaign.sentWin-back · 214
email.openedWeekly digest
churn.savedHelio Inc
One platform

Replaces the patchwork most teams watch churn through

Product analytics
Customer data platform
Email / ESP
Churn spreadsheet
Journey builder
Segment tool
Retention reports
BI queries
Data-team asks
Guesswork
Who it's for

For any product with customers worth keeping

If your customers show up, use features, and can leave, Banchurn works. It learns your product's own baseline, so churn risk means something for how your customers actually behave.

B2B SaaS

Seat-based and usage products where an account cools long before it cancels.

A workspace's logins thin out: the save email goes out that week.

Subscription apps

Recurring consumer and prosumer plans living and dying on renewal.

Usage drops before renewal; a win-back offer lands in time.

E-commerce & DTC

Repeat-purchase brands measuring retention in reorders, not logins.

A customer's cadence slips: a replenishment nudge fires.

Marketplaces

Two-sided platforms retaining supply and demand at the same time.

A seller's listings go quiet: an at-risk journey re-engages them.

Fintech

Accounts and wallets where dormancy is the first step out the door.

Deposits stall; the account drops into a reactivation segment.

Media & content

Subscriptions where a few quiet weeks predict the cancel.

Watch time falls off: a recommendation campaign wins it back.

The engine

Raw events in. Retention out.

Banchurn runs on a single event stream. Everything you send is turned into understanding, signals, and scores, and every module reads the same stream, so a prediction can become an action with nothing in between.

What you send, made useful
An eventA feature, a flow, a trendgrouped and mapped automatically, not just counted.
A trait changeA churn signala downgrade or a drop, emitted as its own event.
BehaviorA live risk scoreper customer, measured against your product's baseline.
An accountAn expand or save signalthe whole relationship scored, PQA or PQR, not one user.
A crossingA journey or an alertenroll the account, or ping the right channel in real time.
A segmentA campaign or a destinationone live audience, targeted on-platform or pushed to your stack.
Platform

Everything you need to keep and grow customers, in one system

Thirteen modules on one event stream and one API, from the first event a customer fires to the signal that saves the account or grows it.

Track

Event tracking & identity

Drop in one snippet to capture every event, identify every customer, and turn trait changes into signals.

Learn more
Predict

Churn prediction

A per-customer risk score that learns your product's baseline, so at-risk customers surface before they leave.

Learn more
Score

Account signals

Score every account on fit, usage, intent, and risk, so you see who's ready to expand and who's slipping toward churn.

Learn more
Understand

Product intelligence

Point your events at Banchurn and it names your features, maps your flows, and flags anomalies: heuristics first.

Learn more
Target

Segments & lists

Dynamic, rule-based segments that keep themselves up to date, plus static lists you control by hand.

Learn more
Send

Campaigns

Batch or triggered, to a segment or a list: one campaign model covers every send you need.

Learn more
Automate

Automations

A visual journey builder: trigger on any event or when a customer enters or leaves a lifecycle state, then send email or in-app, wait, branch, and act.

Learn more
Compose

Email & deliverability

A template engine with variables, open & click tracking, unsubscribe, and A–F deliverability grading built in.

Learn more
Engage

In-app messages

Meet customers inside your product: banners, modals, slideouts, and tooltips, targeted by segment and lifecycle.

Learn more
Activate

Integrations & destinations

Push segments and account signals to Slack, HubSpot, ad audiences, and webhooks: activate your data wherever your team already works.

Learn more
Measure

Analytics

Funnels, retention curves, revenue, and insight dashboards: all over the event stream you're already sending.

Learn more
Autopilot

AI Autopilot

An AI operator that observes, plans, executes, and reports: running your engagement on a loop, within your guardrails.

Learn more
Connect

API & SDK

A JS SDK, an event-ingest API, signed webhooks, and an MCP server so AI agents can run engagement directly.

Learn more

See all features

Deep-dive every module, how it works, and how it connects.

Browse
How it works

From first event to won-back customer

01

Track

Drop in one snippet. Events, identity, and trait changes flow onto a single customer timeline.

02

Understand

Banchurn names your features, maps your flows, flags anomalies, and scores every customer's churn risk.

03

Segment

Turn who they are and what they do into live audiences that keep themselves up to date.

04

Engage

Reach them with batch or triggered campaigns and automated journeys, or let Autopilot run it.

Predict

See who's leaving, weeks early

Banchurn learns your product's baseline and scores every customer against it. Falling engagement, cold key features, and adverse trait changes roll into an adaptive risk score, so a save motion runs in the window before the cancellation, not the exit interview after.

  • Adaptive per-customer churn-risk score
  • Learned from your events: no rules to write
  • A rising score triggers a save journey on its own
Churn prediction
app.banchurn.com/churn
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Churn /Churn risk
Search ⌘KN
At riskSignalsCohorts
At risk
214
Critical
38
Saved / mo
96
Save rate
41%
CustomerTop signalRisk
MHMeridian Healthlogins ▼ 6 days82
VLVertex Labskey feature unused 9d74
OSOrbit Systemsdowngraded to Starter61
LCLumen Cosupport ticket open58
CBCobaltusage ▼ 20% MoM47
NWNorthwindhealthy · expanding22
38 critical enrolled in the win-back journey automatically
Score

Score the account, not just the user

Revenue renews at the account, so Banchurn scores the whole account across four vectors: fit, usage, intent, and risk. That one score becomes a plain signal: PQA when an account is ready to expand, PQR when it's slipping toward churn, each with the evidence behind it, so your team knows exactly why it lit up.

  • One four-vector score → PQA expand + PQR save
  • The same product signals, read two ways
  • Every crossing carries its reasons in plain language
Account signals
app.banchurn.com/accounts
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Customers /Account signals
Search ⌘KN
AccountsSignalsTimeline
Accounts
320
Expansion · PQA
24+5
At-risk · PQR
17+3
Avg fit
72
Accounts · by signal4-vector score
AC
Acme Corp
usage ▲ 30% · seats 8/10
PQA78
MH
Meridian Health
logins ▼ · owner idle 12d
PQR64
VL
Vertex Labs
invited 4 teammates
PQA71
OS
Orbit Systems
downgraded to Starter
PQR59
LC
Lumen Co
steady · no change
·41
Acme CorpPQA
Fit82
Usage88
Intent74
Risk18
Why
Seat utilization 80%, climbing
Viewed upgrade page 3× this week
Understand

Understand your product without a data team

Point your events at Banchurn and it names your features by shared vocabulary, maps how customers move with a Markov flow, and flags spikes, drops, and brand-new behaviors with an EWMA baseline. AI is used only on the margin, to name a group or explain an anomaly.

  • Automatic feature grouping and flow mapping
  • EWMA / z-score anomaly and novelty detection
  • Heuristics in the hot path; AI only on the exceptions
Product intelligence
app.banchurn.com/intelligence
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Intelligence /Product intelligence
Search ⌘KN
FeaturesFlowsAnomalies
Feature groups auto-named
Checkout41 events
Reports28 events
Onboarding22 events
Billing14 events
Integrations9 events
Flow · top path
Home·72%→Reports·64%→Export
drop-off: 38% exit after Reports
Anomaly · flagged
email_open▼ 0 · z −6.2 · critical
Likely cause: sending domain DKIM failure. Suggested: verify SES identity.
Target

Audiences that keep themselves current

Group customers by who they are and what they do. Dynamic segments are live rule sets that re-evaluate on a schedule; static lists are fixed cohorts you control. Every segment carries an engagement score, and any segment can target a campaign or enroll a journey.

  • Dynamic, self-updating rule-based segments
  • Static lists for the audiences you curate
  • One audience model, used everywhere on the platform
Segments & lists
app.banchurn.com/segments
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Segments /Segments
Search ⌘KN
SegmentsListsRules
+ New segmentFilter6 · auto-refreshing
SegmentTypeMembers
At-risk this weekrisk ≥ 60dynamic214
Power users≥ 5 sessions / wkdynamic1,240
Trial ending · 3dtrial_ends ≤ 3ddynamic88
Onboarded, not activatedno key eventdynamic342
Churned, win backuploadedlist512
Expansion readyusage ▲ 30%dynamic176
Send

Every campaign, one model, and every send graded

A campaign is two choices: an audience (a segment, a list, or everyone) and a type: batch or triggered. A welcome email is triggered on a signup; a weekly digest is batch and recurring. Each send is validated, suppressed, delivered through SES, and graded A–F.

  • Batch or triggered, to any audience
  • Suppression, entry controls, and A/B built in
  • SES + DKIM delivery with an A–F grade on every send
Campaigns
app.banchurn.com/campaigns
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Campaigns /Campaigns
Search ⌘KN
SendsScheduledTriggered
Sent · 7d
38.2k
Open rate
47%+3
Click rate
12%+1
Avg grade
A−
CampaignTypeAudienceGrade
WelcometriggerSigned UpA
Win-backtriggerAt-risk ≥ 60A−
Weekly digestbatchAll activeB+
Trial endingtriggerTrial · 3dA
Feature launchbatchPower usersB
Dormant nudgetriggerUnused 14dC+
every send validated · suppressed · delivered via SES · graded A–F
Automate

Turn every signal into the right action

Banchurn runs on an event bus, and Automations is how you program it. Trigger on any event or trait change, or the moment an account crosses into at-risk or expansion-ready, then branch on conditions, wait on timers, and fan out: send an email or in-app message, update a trait, move a customer between segments, enroll a whole account, or fire a webhook. Actions emit events, so journeys compose.

  • Trigger on events, lifecycle states, or account signals
  • Conditions, delays, and rich multi-channel actions
  • Actions emit events, so journeys chain together
Automations
app.banchurn.com/automations
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Automations /Automation engine
Search ⌘KN
RunsBuilderTriggers
TriggerConditionWait 2dActionprocessor · 15s
Event · busacts
risk.risingVertex · 74→ start win-back
signed_upAna · Growth→ send welcome
feature.unused_7dexport_report→ nudge email
email.clickedWeekly digest→ move to engaged
trial.endingOrbit · 3d→ discount offer
churn.savedHelio Inc→ set trait: saved
Activate

Push every signal to where your team works

A signal is only worth what you do with it. Banchurn's destinations framework sends segment membership and account signals straight to your GTM stack (a Slack ping when an account goes at-risk, a HubSpot property when it's expansion-ready, an ad audience kept in sync, a signed webhook to anywhere else), in real time, off the same membership changes that drive everything else.

  • Slack, HubSpot, Meta audiences, and webhooks
  • Real-time on every segment or signal change
  • One connector framework: new surfaces drop right in
Integrations & destinations
app.banchurn.com/destinations
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Settings /Destinations
Search ⌘KN
DestinationsDeliveriesRecommended
Connected
4
Sent · 7d
1.2k
Delivered
99.7%
Failed
3
Destinations+ connect
S
Slack
PQA + PQR → #gtm-signals
H
HubSpot
Account signals → properties
M
Meta Ads
At-risk → Custom Audience
W
Webhook
Segment enter/exit → endpoint
G
Google Ads
Customer Match
soon
in
LinkedIn
Matched Audiences
soon
Live deliveriesreal-time
PQR → SlackMeridian Health
enter → MetaAt-risk +1 · synced
PQA → HubSpotAcme · signal=pqa
PQR → WebhookOrbit · 200
exit → MetaRecovered −1 · synced
secrets encrypted at rest · masked on read
Measure

Analytics that link straight to action

Build conversion funnels over any event sequence, watch retention curves by cohort, track revenue and its movements, and read auto-generated insights: all over the same stream everything else reads, so a leaking step is one click from an audience you can engage.

  • Funnels, retention cohorts, revenue, and insights
  • No separate analytics pipeline to maintain
  • Every chart links to a targetable segment
Analytics
app.banchurn.com/analytics
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Analytics /Analytics
Search ⌘KN
FunnelsRetentionRevenue
MRR
$214k+8%
Active
4,182+6%
30d retention
88%+2
Churn
2.1%−0.4
Onboarding funneldrop 32% @ activate
Signed up4,182
Activated2,844
Habit formed1,715
Paid962
Retention · by cohort
Revenue · 12w
expansion+$4,200
new+$2,100
churn−$980
Connect

API-first, and agent-ready

Send events with the JavaScript SDK or the ingest API, backfill in bulk from CSV or Stripe, subscribe to webhooks, and point an AI agent at the 20-tool MCP server to operate the platform directly. Everything Banchurn does is available over the API.

  • JS SDK, ingest API, bulk import, and webhooks
  • A 20-tool MCP server for AI agent integration
  • Org-scoped keys: everything the product does, over the API
Explore the API
docs.banchurn.com/api
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Settings /API reference
Search ⌘KN
ReferenceSDKWebhooks
POST/v1/events
key
POST/v1/identify
key
POST/v1/import
key
GET/v1/segments
key
POST/v1/webhooks
key
MCPbanchurn-mcp
agent
POST/v1/eventsingest key
curl -X POST https://api.banchurn.com/v1/events \
  -H "authorization: Bearer bnc_live_…" \
  -H "content-type: application/json" \
  -d '{
    "email": "ana@naridon.example",
    "event": "feature_used",
    "traits": { "feature": "export_report" }
  }'

# → 200  event recorded → segments, scoring & journeys update
AI Autopilot

Put your retention on autopilot

Autopilot closes the loop. On a cadence it observes your customers, plans the next-best actions against your goals, executes them through the same audited engine you'd use by hand, analyzes the result, and reports back, inside guardrails you set.

How Autopilot works
app.banchurn.com/autopilot
Banchurn
Naridon
Workspace
Overview
Customers
Segments
Campaigns
Engage
Automations
Templates
Churn
Intelligence
Measure
Analytics
Settings
Mmo@banchurn.com
Overview /AI Autopilot
Search ⌘KN
CycleGoalsGuardrails
Goal · reduce 30-day churn running · every 30m
01
Observe214 customers crossed risk ≥ 60 this cycle
02
PlanEnroll critical (38) in win-back; offer 62 a nudge
03
Execute38 enrolled · 62 nudged · 12 flagged for review
04
AnalyzeWin-back re-engaged 41% of last cohort
reported to #retention · every action logged & reversible
13
modules, one platform
1
event stream behind every module
2
account signals: expand + save
A–F
grade on every send
Trust

Built to be trusted with retention

The load-bearing path is deterministic, your data stays scoped to you, and every automated action is logged and reversible. The safe path is the default path.

API & SDK

Deterministic where it counts

Heuristics run the high-volume path: feature grouping, flows, scoring, anomalies. AI only names a new group or explains a fresh anomaly, so behavior is predictable and testable.

Org-scoped by construction

Every query is pinned to your organization in the data layer, and API keys are scoped to exactly the integration you mint them for.

Verified deliverability

Email is delivered through AWS SES on a DKIM-verified domain, with suppression, unsubscribe, and complaint handling enforced before anything sends.

Guardrailed & reversible

Every automated and Autopilot action is logged, frequency-capped, and reversible, with approval gates you control per action class.

The thesis
“Customers tell you they're leaving weeks before they go. Banchurn is what's listening, and what acts.”

Why we built Banchurn

FAQ

Frequently asked questions

What exactly is Banchurn?+

A retention and expansion platform for B2B SaaS. It captures your product's events and identity, predicts each customer's churn risk, and rolls that up into an account score that flags who's ready to expand (PQA) and who's slipping toward churn (PQR). From there it builds live segments, reaches customers with campaigns, automated journeys, and in-app messages, pushes those signals out to Slack, your CRM, and ad tools, and measures the whole thing with built-in analytics and an AI Autopilot: all on one event stream and one API.

How does churn prediction work without me training a model?+

Banchurn learns your product's own behavioral baseline directly from the events you're already sending: what healthy engagement looks like for your customers. Each customer is scored against that baseline as their behavior and traits move, so falling engagement, cold key features, and adverse trait changes surface the accounts most likely to leave. There's no model to train and no churn rules to hand-write.

Isn't this just churn prediction? Where's the expansion side?+

The same signals, read two ways. Banchurn scores every account across four vectors (fit, usage, intent, and risk), and that one score surfaces both the accounts slipping toward churn (we call it product-qualified risk, or PQR) and the accounts leaning in and ready to grow (product-qualified accounts, PQA). One is your save motion, the other your expansion motion, and both come from the product usage you're already tracking: no second tool and no separate model. PQR wins ties, so risk is never hidden behind opportunity.

Can Banchurn push signals into Slack, our CRM, or ad tools?+

Yes, that's the destinations framework. When an account crosses a signal or a customer enters a segment, Banchurn can post to Slack, write properties back to HubSpot, keep a Meta ad audience in sync, or fire a signed webhook to anything else, in real time, driven by the same membership changes that power the rest of the platform. Your GTM team acts on the signal in the tools they already live in, not in one more dashboard.

Do I have to rip out my current analytics and email tools?+

You can, and most teams do over time, because Banchurn replaces the analytics tool, the CDP, the ESP, and the churn spreadsheet with one system that shares an audience model. But you don't have to start there: drop in the SDK, send events, and adopt modules as you go. It's one platform precisely so your data, segments, and sends stop living in separate tools.

The product intelligence uses AI. Isn't that slow and expensive?+

Not in the load-bearing path. Feature grouping, flow mapping, churn scoring, and anomaly detection are all deterministic heuristics that run over millions of events cheaply and predictably. AI is called only on the margin, to name a newly discovered feature group or explain a fresh anomaly, so cost scales with how much is surprising, not with your traffic. Ten times the events costs the same in AI.

What can the AI Autopilot actually do on its own?+

As much or as little as you allow. Autopilot runs an observe → plan → execute → analyze → report loop against goals you set, acting through the same audited engine you would: sends, segment moves, journey enrollments. You configure guardrails: allowed action types, frequency caps, and approval gates. Run it in suggest-only mode until you trust it, then let it execute. Every action is logged and reversible.

How do I get my data into Banchurn?+

Five ways, all resolving onto one customer timeline: the JavaScript SDK for browser events and identity, the ingest API for trusted server-side events, bulk CSV import for backfill, a Stripe sync for subscription and revenue events, and scheduled endpoint syncs. You can also point an AI agent at the 20-tool MCP server and have it operate the platform directly.

Stop losing customers you could have kept

Book a walkthrough and see churn prediction, account signals, segments, campaigns, destinations, and Autopilot working together, on one platform.