The questions evaluation teams actually ask

Accuracy depends on your data and your sales motion, so we will not quote a single headline figure that was measured on somebody else's business. Instead we measure it on yours. During evaluation we backtest the model against your closed history and show you how it would have called past quarters, broken down by segment and by how far ahead the call was made.

After go-live the platform keeps scoring itself against every closed period. The accuracy number you work with is always your own, always current, and always visible next to the forecast it describes.

Most teams connect a CRM and see a first modeled forecast within a few days, because the historical backfill and the initial model fit are automated.

Getting to the point where the forecast drives your cadence takes longer, usually a few weeks. That time goes on reviewing field mappings, agreeing segment definitions and deciding which signals your team actually trusts. We would rather spend it up front than hand you a number nobody believes.

The common path is a CRM, a data warehouse, a billing system, a product analytics source and a support tool. You do not need all of them to start.

A CRM on its own is enough to produce a forecast. Each additional source sharpens a specific part of it: billing and usage mainly improve renewals and expansion, support data mainly improves risk scoring, and a warehouse connection covers most gaps where a direct integration does not exist.

By default, nothing at all. The platform reads from your CRM and writes nothing back, which is how nearly every team starts.

If you later want scores, forecast categories or risk flags to appear on the opportunity record, writeback can be enabled field by field, with a clear record of what is written and by whom. It is never switched on without you switching it on.

Data is encrypted in transit and at rest, access is governed by single sign-on and role-based permissions, and every access and change is written to an audit log you can inspect yourself.

Connections are read-only unless you explicitly enable writeback, and we hold only the fields your mappings actually use rather than a full copy of your CRM. Enterprise plans add data residency options. Our current security documentation is available on request during evaluation.

You need enough closed history for the patterns to be real. For most businesses that means a few hundred closed opportunities across a couple of years.

Below that, the platform is honest about it. Confidence ranges widen, the model leans on signals that generalize well such as engagement and stage velocity, and it tells you plainly which segments it does not yet have enough evidence to model. That is far more useful than a confident number built on forty deals.

Reps carry on working in the CRM. The platform reads what is already there rather than asking for a second set of updates, and where it needs something it does not have, such as an agreed next step, it flags the gap instead of adding another form.

Managers see the bigger change. One-to-ones and pipeline reviews start from scored deals with reasons attached, rather than from a list somebody sorted by close date the night before.

CRM forecasting mostly adds up what people typed in. It reflects the categories reps selected and the close dates they set, which is why it tends to move in steps at the end of a quarter rather than tracking reality week to week.

This models how deals in your business actually behave, using engagement, velocity, buying-group coverage, product usage and billing signals, and it produces a range rather than a single figure. The two are complementary: we deliberately hold the submitted forecast next to the modeled one, because the gap between them is usually the most useful number on the page.

Pricing is based on the number of revenue seats and the systems you connect. We quote after a short scoping call rather than publishing a list price, because the same headcount can mean very different scope depending on how many entities, currencies and sources are involved.

Agreements are annual, plans can move mid-term as headcount changes, and the assumptions behind your quote are put in writing alongside it. The pricing page sets out what each plan covers.

Yes. The usual shape is a paid pilot on one segment or region for a quarter, with success criteria agreed in advance, most often forecast accuracy against the submitted number and how far out the model holds.

Before that, during evaluation, we backtest against your own closed history so you can see how the model would have called quarters you already know the answer to. The backtest costs you nothing and is the fastest way to find out whether this fits your business.