Skip to main content
Beton
posthog_contact_record posthog_contact_record_mvp_v1

Contact / record-based SaaS

CRM- or marketing-style SaaS priced by record count. Imports and enrichment drive positive signals; segmentation without growth is a negative pattern.

Value metric

Contacts, leads, subscribers, accounts managed

Success event

contact_created

entity_type: account

Scale

  • 300 accounts
  • 4 users/account (mean)
  • 7 sessions/user (mean)
  • 30 days of history

Research metrics proxied

  • — Contact growth rate
  • — % of contact limit used
  • — Import frequency

Signal paths

Positive paths end in contact_created. Negative paths do not. Every generated event_id belonging to a path is recorded in ground_truth.json.

Positive signals (2)

import_to_contact_growth ×85
list_imported → contact_created
cohorts: high_intent, medium_intent, power_user
enriched_contact_creation ×45
contact_created → enrichment_completed → contact_created
cohorts: high_intent, medium_intent

Negative signals (2)

import_segment_stall ×60
list_imported → segment_created
cohorts: medium_intent, low_intent, lurker
contact_segment_without_growth ×45
contact_created → segment_created
cohorts: low_intent, lurker, noisy_bot_like

Generate this dataset

Config file: configs/posthog_contact_record_mvp.yaml

Quickstart 
# Dockerized Postgres (recommended for inspection)
docker compose up -d

uv run dryfit \
  -c configs/posthog_contact_record_mvp.yaml \
  --dsn postgresql://dryfit_writer:dryfit_writer@127.0.0.1:54329/dryfit \
  --print-summary

# Or local Postgres
./scripts/generate-local -c configs/posthog_contact_record_mvp.yaml --print-summary

Full setup instructions are in the repo's README — including local Postgres, Grafana inspection, and dataset restore.

Noise parameters

DryFit injects realistic noise on top of the generated signal paths. These probabilities are per-event. Noise never touches rows referenced by ground_truth.json — your scoring logic can trust the truth file is exact.

missing event probability
4.0%
duplicate event probability
2.0%
out of order probability
3.0%
null property probability
3.0%
anonymous actor probability
1.0%
weird property probability
1.0%

Other scenarios

Combined coverage (all models)

→ upgrade_clicked

Union of event types across all business models

Credits / token-based

→ credits_purchased

Credits consumed, tokens used, compute units

Event-volume SaaS

→ custom_event_tracked

Events tracked, data points ingested, log lines

Feature-gated (tiered) SaaS

→ upgrade_clicked

Plan tier / feature access level
Browse all scenarios →

Benchmark your detector against Contact / record-based SaaS

Clone the repo, run the config, check your agent's output against ground_truth.json.

View on GitHub