Computes weighted renewal probability and generates AI retention plans for at-risk accounts.

How it works 22 steps

Inspect the data fetches, transforms, gates, and output this recipe runs.

Data fetch

▼ account account-reader

Fetch from reader

▼

Process (21 steps)

■ days_to_renewal date_diff

Calculate days between now and account contract end date

⚠ check_window guard

Stop if days to renewal is greater than

■ usage_w math

Calculate account product usage trend 90d times 1

■ nps_w math

Calculate account latest nps score times 10

■ ticket_penalty math

Calculate account support ticket count 90d times 10

■ ticket_w math

Calculate 100 minus ticket penalty

■ engagement_w math

Calculate 100 minus account days since last positive engagement

■ w_usage math

Calculate usage w times 0.35

■ w_nps math

Calculate nps w times 0.25

■ w_ticket math

Calculate ticket w times 0.2

■ w_engage math

Calculate engagement w times 0.2

■ prob_1 math

Calculate w usage plus w nps

■ prob_2 math

Calculate w ticket plus w engage

■ renewal_probability math

Calculate prob 1 plus prob 2

■ prob_low compare

Check if renewal probability is less than setting ai threshold

■ ai_ready all

Check if all conditions are true

■ hashed_account hash_replace

Anonymize sensitive fields for safe AI processing

✦ ai_retention_plan ai-prompt

AI generates a response from a custom prompt

skip: step.ai_ready equal false

■ restored_plan hash_restore

Restore anonymized fields back to real values

skip: step.ai_ready equal false

■ health_checklist to_checklist

Format results as an actionable checklist

■ metrics to_summary

Format results as a summary card

▼

Output

summary metrics

checklist health_checklist

ai_analysis restored_plan

Settings 2 configurable

Configurable at install. Defaults shown — change them anytime in Recued.

enable ai toggle = on

ai threshold number = 70