A deterministic decision model, trained on your outcomes.

Deterministic

The same model version and the same inputs give the same answer. Every answer names its model version, so you can check it later.

Trained on your outcomes

It learns from your own past records and what happened to them. Reasons are in your own fields, like "company size 51 to 200". Your data trains only your models.

Tested before it decides

First a backtest on records it did not train on. Then shadow mode beside your current method. It goes live only when you say so.
What it decides

One yes-or-no decision, answered the same way every time.

Each answer is a yes or no with a probability and a priority level. Datagoat needs a few hundred past records with known outcomes. It does not read free text: pair it with a general decision model for that.

How it works

Tested on your own records before it decides anything.

Run a backtest on past records, run the model in shadow mode beside your current method, and go live where it does better.

Step 1 of 3

Start with a backtest

Upload a CSV of past records or connect HubSpot. No code. Or ask your coding agent to connect: it finds the decision in your code and imports past records for you. Datagoat reads the outcome window from your file's dates and asks two questions. In a few minutes you see what Datagoat's picks captured, beside your current method's picks.

Run a free backtest
Step 2 of 3
Run it in shadow mode

Run it in shadow mode

Wrap the step with the SDK, or call the API. The model answers every new record beside your current method, and nothing acts on its answers. Every night both are compared with what happened. You get a plain result: ahead, behind, or not statistically significant yet.

Step 3 of 3

Go live when it does better

One click per decision. The model decides from the next record on. A control group stays with your current method, so the comparison keeps running. Every answer has a signed audit record. Pause at any time.

Pricing

Free to test. Pay only
when the model decides.

Free

Backtest and shadow mode

$0
Find out if it does better
No cost
No commitment
Backtests on your past records, from a CSV or HubSpot
Shadow mode beside your current method
A nightly comparison on the same records
What the model learned, in plain words on your own fields
Nothing changes until you go live
After you go live

Pay for what the model decides

On requestfor design partners
Only records the model decided
Control group and fallbacks are free
Learns from your outcomes
Compared every night
Watched for problems
Pause at any time
The model decides when it has an answer; otherwise your fallback applies
A signed audit record on every answer
Records can go to a review queue, and your current method takes over if Datagoat is down
Retrained every night on new outcomes
A new model version takes over only after it is tested against the current one
Nothing changes until you go live.
Pause at any time; your current method decides again from the next record.
Why trust it

A decision you cannot check is a guess.

Every Datagoat answer can be shown, checked and repeated. These are the rules it holds itself to.

Loading...

Deterministic

Same model version, same inputs, same answer. Ask again and you get the same answer back. A retrained model is a new version.

Honest

When your past records show no usable pattern, Datagoat trains no model and says why. Your fallback applies. It does not guess.

Bounded

Datagoat gets 800 ms to answer, with the probability, priority level and reasons. Past that, your fallback applies, so your step does not wait on it.

Yours

Every audit record is signed and names the model version that decided, so you can show what was decided, and why, later.

Private

Your data trains only your models. Nothing is pooled across customers.

Readable

Reasons are conditions on your own fields: "company size 51 to 200", "no pricing page views". They went together with the outcome. They are not proven causes.