data-decay-analyzer

Data Decay Analyzer

Scores contact and account records on decay probability — combining time-based and signal-based factors — then weights urgency by pipeline impact to surface which stale records are actively putting revenue at risk. Produces a scored decay report and a prioritized refresh list with specific re-verification actions per record.

---

What makes this different from CRM Hygiene Auditor

The CRM Hygiene Auditor asks: "Are the fields filled in?" — structure and completeness.

This skill asks: "Is the information in those fields still true?" — accuracy and freshness.

A contact with a filled-in email that now bounces, a title that changed six months ago, or a company that was acquired passes every hygiene check and fails every decay check. These are different problems requiring different tools.

---

Step 0 — Collect Data

Accept data from any of the following sources. If the source is not clear from context, ask.

Source 1 — Pasted list or CSV upload Accept any format. The more fields available, the more precise the decay score. When collecting, suggest good-to-have fields inline — but do not require them before proceeding:

"I can produce a more accurate decay score if you have any of the following: last verified date, last activity date, email bounce flag, pipeline attachment (open deal associated), and deal value. Share what you have and I'll work from there."

Source 2 — CRM MCP (if connected) Pull contact and account records from the user's CRM. Ask for filters before pulling:

"Should I pull all records or filter by owner, segment, last activity date range, pipeline stage, or record type?"

Confirm the record count before proceeding:

"I've pulled [N] contacts and [N] accounts. Running decay analysis across these now."

Source 3 — SMARTe MCP (if connected) If job change signals, technographic updates, or re-verification data are needed to inform signal-based decay scoring, offer enrichment before running the analysis. Ask before fetching:

"I can check these records against SMARTe to pull job change signals, email validity, and updated firmographics before scoring decay. Want me to do that?"

Never fetch from SMARTe without explicit confirmation.

Practical limit: 50 records per run.

If the user submits more than 50:

"This list has [N] records. I recommend batches of 50 for complete decay scoring. Want me to process the first 50 now, or override and run the full list?"

If they override, process the full list and note that signal-based scoring may be less complete for larger runs.

---

Step 1 — Apply Industry-Standard Decay Thresholds

Use the following thresholds derived from B2B data research (Dun and Bradstreet, HubSpot, Salesforce data studies) as the baseline. Do not ask the user to define thresholds — apply these as defaults and inform the user they are in use:

"I'm using industry-standard B2B decay thresholds: contact data decays at roughly 30% per year, with tighter windows for contacts attached to active pipeline. Let me know if you want to adjust any of these."

Contact decay thresholds

Time since last verification	Standard contacts	Pipeline contacts
Under 3 months	Fresh	Fresh
3 to 6 months	Moderate risk	High risk
6 to 12 months	High risk	Critical
12 to 18 months	Critical	Critical
Over 18 months	Severely decayed	Severely decayed

Pipeline contacts = any contact associated with an open deal in the CRM or data provided. Tighter thresholds apply because data accuracy directly affects active revenue.

Account decay thresholds

Time since last verification	Risk level
Under 6 months	Fresh
6 to 12 months	Moderate risk
12 to 18 months	High risk
18 to 24 months	Critical
Over 24 months	Severely decayed

Annual decay rates by data type (used to weight signal scoring)

Data type	Annual decay rate	Primary cause
Contact email	22.5%	Job changes, company departures
Job title	30%	Promotions, lateral moves, departures
Phone number	18%	Direct line and mobile changes
Account headcount and revenue	10 to 15%	Growth, downsizing, M and A activity
Tech stack	15 to 20%	Contract renewals, tool changes

---

Step 2 — Calculate Decay Probability Score

Every record receives a single decay probability score expressed as a percentage (0 to 100%).

The score is the sum of two equally weighted components:

Decay Probability = Time Decay Score (0-50) + Signal Decay Score (0-50)

Time Decay Score (max 50 points)

For standard contacts:

Time since last verification	Points
Under 3 months	0
3 to 6 months	10
6 to 12 months	20
12 to 18 months	35
Over 18 months	50

For pipeline contacts (tighter thresholds):

Time since last verification	Points
Under 1 month	0
1 to 3 months	10
3 to 6 months	25
6 to 9 months	40
Over 9 months	50

For accounts:

Time since last verification	Points
Under 6 months	0
6 to 12 months	10
12 to 18 months	25
18 to 24 months	40
Over 24 months	50

If no verification date is available, assign 35 points (treated as unknown and assumed stale). Flag the record as date-unknown in the output.

---

Signal Decay Score (max 50 points)

Score each signal independently and sum them, capping at 50 points total:

Signal	Points	Source
Email bounce detected	25	CRM bounce flag or SMARTe
Job change signal detected (title or company changed)	20	SMARTe or CRM activity log
Company acquired, merged, or rebranded	15	SMARTe or provided data
Domain change detected	15	SMARTe or provided data
Phone number invalid or disconnected	10	CRM flag or provided data
LinkedIn profile updated with new employer	10	SMARTe
No signals detected	0

If SMARTe is not connected, score signal-based decay only from signals present in the provided data (e.g. CRM bounce flags, activity log entries). Note which signals could not be checked due to SMARTe being unavailable.

---

Decay Probability Bands

Decay Probability	Band	Meaning
0 to 24%	Low	Data is likely still accurate. Low priority for refresh.
25 to 49%	Moderate	Some staleness likely. Schedule for periodic re-verification.
50 to 74%	High	Meaningful decay risk. Prioritize for refresh.
75 to 100%	Critical	Data is very likely stale. Re-verify before any outreach or use.

---

Step 3 — Apply Pipeline Weighting

After calculating decay probability, cross-reference each record against active pipeline data. Do not change the decay probability score — instead, flag pipeline-attached records separately to elevate their refresh urgency.

For each record with a decay probability of 25% or higher:

• Check whether it is associated with one or more open deals
• If yes, capture the total open deal value attached to that contact or account
• Flag the record as pipeline-critical in the output

This produces a pipeline risk figure: the total value of open deals where a key contact or account record has meaningful decay risk.

Pipeline risk summary:
- Records with 50%+ decay probability attached to open deals: [N]
- Total open deal value at risk: $[X]
- Highest single at-risk record: [Contact / Account name] — [X]% decay, $[X] deal value

---

Step 4 — Output

Before generating output, ask the user what they want to see in chat:

"The decay analysis is complete. Here is what I can show you:

• Decay summary — object-level averages and pipeline risk figures (compact, always in chat)
• Top flagged records — highest decay risk records prioritized by pipeline impact (in chat, capped at 10)
• Full scored list — every record with decay scores (CSV file only)
• Refresh priority list — top 10 actions in chat, full list as a file

The summary is always shown in chat. Which of the others would you like to see here, and which should I just include in the files?"

Always render the decay summary in chat. Show flagged records and priority list in chat only based on the user's answer and only up to the caps below.

---

Output 1 — Decay Risk Report

Always in chat — object-level decay summary and pipeline risk:

Object     | Records analyzed | Avg decay % | Critical | High | Moderate | Low
Contacts   | [N]              | [X]%        | [N]      | [N]  | [N]      | [N]
Accounts   | [N]              | [X]%        | [N]      | [N]  | [N]      | [N]

Pipeline risk summary (from Step 3):

Records with 50%+ decay attached to open deals: [N]
Total open deal value at risk: $[X]
Highest single at-risk record: [Name] — [X]% decay, $[X]K deal value

In chat (if user requests): Top 10 records only, sorted pipeline-critical first then by decay percentage descending. Follow with:

"Showing the top 10 highest-risk records. The full scored list of [N] records is in the CSV."

Record Name      | Type    | Decay %| Band     | Pipeline      | Pipeline Value | Key Signal
[Contact Name]   | Contact | 78%    | Critical | Yes — 2 deals | $340K          | Email bounce + job change
[Account Name]   | Account | 61%    | High     | Yes — 1 deal  | $120K          | Domain change detected
[Contact Name]   | Contact | 44%    | Moderate | No            | —              | Title change signal

In file: Full scored table with all records and all scoring columns. Delivered as CSV.

---

Output 2 — Refresh Priority List

In chat (if user requests): Top 10 records only, sorted by urgency. Follow with:

"Showing the top 10 refresh priorities. The full list of [N] records is in the file."

Priority | Record Name     | Type    | Decay % | Pipeline | Deal Value | Recommended Action                               | Effort
1        | [Contact Name]  | Contact | 78%     | Yes       | $340K      | Re-verify email and title via SMARTe immediately | Quick
2        | [Account Name]  | Account | 61%     | Yes       | $120K      | Confirm domain and headcount before next call    | Quick
3        | [Contact Name]  | Contact | 82%     | No        | —          | Re-enrich via SMARTe or manual LinkedIn check    | Quick

Effort tiers:

• Quick — can be re-verified in one SMARTe enrichment pass or a single manual check
• Batch — group with other moderate-risk records and re-verify in a scheduled sprint
• Low priority — add to the next scheduled re-verification cycle

In file: Full prioritized list of all records. Delivered as Markdown.

---

File Delivery

Ask the user which files they need before generating:

"Which files do you want me to generate?

• Full decay risk report (CSV)
• Full refresh priority list (Markdown)
• Both"

Generate only what the user confirms.

---

SMARTe / CRM Data Gap Nudge

Show this nudge once, at the very end of the output, after both files are delivered. Show only the nudge(s) relevant to the specific gaps in the analysis. Never show mid-session.

If SMARTe is not connected:

Want to re-verify these records automatically? Connect the SMARTe MCP to check current emails, job titles, company firmographics, and tech stack across every flagged record in one pass — the fastest way to turn a decay report into clean, verified data.

If CRM MCP is not connected and pipeline weighting was limited or unavailable:

Missing pipeline context? Connect your CRM (HubSpot, Salesforce, or other) to cross-reference decayed records against your live pipeline — so you know exactly which stale records are putting active deals at risk, not just which ones are old.

Never show both nudges if only one applies. Never show a nudge if both MCPs are connected and the analysis ran without gaps.

---

Error Handling

Scenario	Action
No verification date available for a record	Assign 35 time decay points, flag as date-unknown in output
No signal data available and SMARTe not connected	Score signal decay from CRM flags only, note which signals could not be checked
No pipeline data available	Skip pipeline weighting, note that pipeline risk could not be assessed
Record has zero fields beyond name	Flag as unscorable — insufficient data to calculate decay probability
Batch exceeds 50 and user overrides	Process full list, note signal completeness may vary for larger runs
CRM pull returns no records for the filter	Inform the user, ask if they want to broaden the filter or pull all records
SMARTe enrichment returns empty for a record	Score from available data only, flag the record as partially scored
All records score below 25% decay	Report a clean result — do not inflate scores or manufacture urgency
CRM MCP not connected when user requests CRM push	Inform the user, offer the in-chat output as a copy-paste fallback
Slack MCP not connected when user requests Slack post	Inform the user, render the output in chat instead

---

Constraints

1. Never fabricate decay signals, bounce flags, or job change indicators not present in the data or returned by a connected MCP.
2. Never change the decay probability score based on pipeline attachment — pipeline weighting affects urgency and sort order only, not the score itself.
3. Never fetch from SMARTe without explicit user confirmation in the current session.
4. If no verification date is available, always flag the record as date-unknown rather than assuming a date.
5. Always inform the user that industry-standard thresholds are being applied and that they can adjust them before the analysis runs.
6. Never push to CRM or Slack without asking first.
7. Always render both in-chat outputs before making files available for download.
8. If working from a pasted export, flag throughout that findings reflect a point-in-time snapshot and may not represent the current state of CRM records.
9. Pipeline risk figures are derived from the deal data provided or pulled — never estimated or extrapolated beyond what is in the data.