Customer revenue risk intelligence

ChurnRisk: Know which customers are quietly slipping away.

ChurnRisk connects your accounting and communication data to identify declining customers, explain what changed, and help your team act before the revenue disappears.

Join the Design Partner Program See How It Works

Read-only connections. Evidence behind every warning. Nothing sends to your customers without your approval.

Sample data

Henderson Industrial
Risk$184,230HighReorder 34 days past this buyer's own cadence. Two unanswered replies since March 11.

Northlake Packaging
Watch$92,480HighCorrugated volume steady. Stretch film moved elsewhere over two quarters.

Rivet and Sons
Healthy$61,900MediumOn cadence and on terms. Nothing to do.

Acme Fabrication
Unknown$47,110Coverage 38 percentNo mailbox connected for this account owner. Not enough visibility to assess.

The problem

You already have the clues. They are just in different places.

The revenue is shrinking in your accounting system. The replies are getting slower in your sales rep's inbox. There is an unresolved complaint buried in a thread from March. The buyer you built the relationship with left in February and nobody told you.

Every one of those is visible. None of them is visible together.

So the pattern only becomes obvious after the customer is gone, when someone finally asks why the Henderson invoices stopped, and the answer turns out to have been sitting in four systems for five months.

ChurnRisk reads both sides. It knows what a normal reorder looks like for each of your customers, notices when the pattern breaks, checks whether the relationship explains it, and tells you which accounts are worth your Monday morning.

Concentration

What is your top ten worth?

In most repeat-revenue businesses, a handful of accounts carry most of the margin. Losing one is not a bad month. It is a bad year.

Most owners have never run the number. The calculator takes five inputs: customer count, annual revenue, top-ten concentration, average relationship length, and your own stated retention rate. It returns four figures: the revenue sitting in your top ten, what one top-decile loss costs you, what your stated retention rate quietly costs every year, and the lifetime revenue behind an average customer. No account required, no email gate, arithmetic you can check by hand.

$184,230Revenue at stake in one sample account

Sample data, labeled as such. From the illustrative board above.

34 daysPast this buyer's own reorder cadence

Sample data. The product measures each account against its own history.

Run the Revenue Risk Calculator

Every account

Four separate answers. Not one number nobody trusts.

On every account, ChurnRisk gives you four readings. Separately. Always.

Risk.

How likely this customer is to stop buying, buy less, or move volume elsewhere.

What this is based on

Reorder gap versus this buyer's own cadence
Unanswered threads in connected mailboxes

Revenue at stake.

What it is actually worth in dollars, so a $900 account never outranks a $180,000 one.

What this is based on

Invoice history from your books, read-only
Each account's own order values, never an average

Confidence.

How much of this customer we can actually see. When we cannot see enough, we say so instead of guessing.

What this is based on

Which mailboxes and channels are connected
Missing coverage lowers confidence, never raises it

What to do.

The specific next step, with the message drafted, ready for you to edit and approve.

What this is based on

A drafted message, waiting for a person's edit
Nothing sends without your approval

Mechanism

Three connections. One Monday morning list.

Connect your books.

QuickBooks Online, read-only. We learn each customer's normal ordering rhythm, product mix, and payment behaviour from the last two years.

Connect your sales inboxes.

Gmail or Outlook, only the mailboxes you choose, minus any labels you exclude. We look for slowing replies, unresolved complaints, competitor mentions, and contacts who have gone quiet.

Get your Monday morning list.

Ranked by revenue at stake. Every warning shows what changed and where the proof is. Every recommended action waits for your approval.

See How It Works

Trust

Every warning shows its work.

No black-box score. Every risk names what changed, with the invoice, the order gap, or the email thread behind it.

We do not train models on your data.

We store the minimum we need and link back to the original rather than copying your inbox.

We never message your customers without you.

Every email, text, and call is written, reviewed, and approved by a person on your team. You can stop everything with one switch.

When we cannot see enough, we say so.

A customer we do not have data on is marked unknown, never healthy. Missing information lowers our confidence. It never gets treated as good news.

Sample data

Henderson Industrial

Reorder 34 days past this buyer's own cadence. Two unanswered replies since March 11.

What this is based on

Invoice history: ordered every 3 weeks for 14 months
Email thread: short shipment raised March 11, no reply from us since

The evidence panel behind one warning, on synthetic distributor data.

Fit

Built for the businesses that get surprised.

Built for repeat-revenue B2B businesses where relationships are named and customers come back: wholesale distributors, packaging and print suppliers, service contractors, MSPs, agencies, staffing firms, and freight brokers.

Between 50 and 5,000 active customers. QuickBooks as the financial source of truth. Sales relationships that live in email.

Not built for consumer retail, one-off project businesses, or SaaS companies whose churn signals live in product usage rather than in the ledger.

Objections

Questions owners actually ask.

We already know our customers.

Your team knows individual relationships. This watches every account continuously and connects relationship changes to revenue, so the important exception does not depend on someone remembering to raise it.

I do not want AI reading our inbox.

Read-only, only the mailboxes you pick, minus labels you exclude. We store structured signals and short quotes, not your mail. No model is trained on it. Nothing sends without your approval.

Is this another CRM?

No. It does not manage a pipeline and does not ask anyone to enter data. It reads the systems already running the relationship.

Health scores are made up.

Agreed. There is no blended number here. Risk, revenue at stake, confidence, and the next action are separate answers, and every warning shows the invoice or the email behind it.

Half our orders come in by phone.

Then we mark those customers and lower our confidence rather than guessing. That is exactly why confidence is a separate number.

Will it email my customers automatically?

No. Nothing sends without a person on your team approving it, and you can stop everything with one switch.

What does it cost?

Plans are priced on monitored active customers, never on seats. Unlimited users and every security feature on every plan.

Details on the pricing page

Find out in March, not in September.

We are working with a small number of design partners to prove this on real customer histories before we open it up. If you run a repeat-revenue B2B business and you have been surprised by a lost customer in the last year, we would like to talk.

Apply to the Design Partner Program

We reply within one business day. No sequence, no SDR, no surprise calendar invites.