Stop discounting the trial users who would have paid anyway.
KaQuill predicts which trial users will convert in real time. It fires nudges, sales alerts, and webhooks only for the fence-sitters — so you keep margin on inevitable conversions and save the winnable ones.
Margin you burned discounting 23 trials who would have paid full price.
Time-based drip campaigns are flat. Your trial users aren't.
You discount the wrong users
Day-7 trial email fires at everyone equally. The users who were going to pay full price get the discount too — you train them to wait for it.
You lose the fence-sitters
The trials who would have converted with one well-timed nudge churn silently. No one knows they were close. They just don't show up on Day 14.
Your sales team chases ghosts
Reps spend hours on trials that were never going to pay. The intent signal is in the product usage data — no one is reading it.
Three steps. No model fine-tuning. No data team required.
Drop in the SDK
A server-side call — the Node SDK is four lines, or hit the REST API directly. Send trial signups, feature events, plan views — anything you already track.
KaQuill watches and learns
Behavioral fingerprinting + reinforcement learning. Per trial, per moment, it predicts conversion probability with calibrated confidence.
Intervene precisely
An in-app nudge you render, a Slack alert, or a signed webhook to anything you own. Fired only for fence-sitters — never for the users already on their way to converting.
See exactly how much margin your drip campaign burns each month.
For every discounted conversion, KaQuill computes the counterfactual: would this user have paid full price anyway? When the answer is yes, that's Ghost MRR — revenue your blanket discount strategy is destroying.
It's the number you didn't have. The one that makes the case for intent-based intervention impossible to argue with.
New to the metric? Read: What is Ghost MRR?
Not another rules engine wearing AI lipstick.
Agent, not rules
Userpilot and Appcues make you write the if/then logic by hand. KaQuill learns it — it predicts each trial's intent, adapts as it sees more outcomes, and decides whether to act at all instead of firing the same rule at everyone.
Counterfactual aware
Separates conversions it caused from conversions that would have happened anyway. Every decision logs the propensity it was sampled with, so lift can be measured offline instead of asserted.
Action-layer agnostic
KaQuill returns the decision and the copy; you own the surface. Render it in-app, or let KaQuill POST it to a signed webhook or Slack. The decisioner doesn't care what the action is.
Four calls. Ten minutes to live.
Server-side Node SDK, or POST JSON to the universal endpoint. Other languages and a browser package are on the roadmap.
import { Kaquill } from '@kaquill/node';
const kq = new Kaquill({ apiKey: process.env.KAQUILL_KEY });
// On trial signup
kq.identify(user.id, {
plan: 'trial',
trial_ends_at: '2026-06-23',
});
// On any meaningful product event
kq.track('feature_used', {
user_id: user.id,
feature: 'export_csv',
});Free to see the leak. Paid to plug it.
See your Ghost MRR. Read-only insights, no interventions.
- Up to 100 nudges per month
- Ghost MRR dashboard
- Conversion probability per trial
- Community support
Unlimited nudges. Cancel any time.
- Unlimited tracked trials
- In-app nudge decisions + signed outbound webhooks
- Slack alerts for high-intent trials
- All Free features included
- Email support
Save ~$200 vs monthly. Same unlimited nudges.
- Everything in Monthly
- Two months free vs monthly billing
- Priority support
- Locked-in pricing for the year
Need higher volume? Contact us for custom enterprise pricing.
Questions, answered.
How is this different from Userpilot, Appcues, or Pendo?+
Those are rules-based: you write if/then logic for tooltips. KaQuill is a continuously-learning agent that predicts conversion probability per user, per moment — and decides whether to intervene at all. The Ghost MRR metric also doesn't exist anywhere else.
What events do I need to send?+
At minimum: trial signup, plan change, and any 'meaningful' product action (feature use, document creation, anything that signals engagement). The richer your event schema, the better the model. We provide example schemas for common SaaS verticals.
How long until KaQuill starts being useful?+
Ghost MRR and conversion probabilities appear as soon as events start landing. The learning policy takes over after 1,000 tracked trial users on a paid plan — until then a deterministic rule engine makes the calls, and you can watch it in shadow mode before giving it any traffic. You always see the confidence and which path decided.
Where does the intervention actually fire?+
Wherever you render it. KaQuill returns the action and the copy from the API, and your app decides what to do with it. It can also deliver for you: a signed webhook POSTed to any endpoint you own, or a Slack incoming webhook for high-intent alerts. Email works by pointing that webhook at your existing transactional provider — we don't send mail on your behalf, so your sender reputation and your templates stay yours.
What's the data story? GDPR, etc.?+
You own your data. We never sell or share it. Full DPA available before signup, right to delete, right to export, hashed user identifiers. We don't currently offer region-pinned storage — if data residency is a hard requirement for you, talk to us before you sign up rather than after.
Will this work for my specific SaaS?+
If you have a free trial or freemium plan and care about conversion, yes. We've designed KaQuill to be vertical-agnostic — same engine for project management tools, developer tools, AI tools, design tools. The model learns your specific buyer's behavior.
See what your discounting actually costs.
Free forever to see Ghost MRR. Upgrade only when you're ready to act on it.
Start free