Glossary: how every score and count is calculated

Nothing in this app is a black box. Each entry below names the data it reads and the exact rule it applies.

Ranking and queues

Priority call score

Points from four inputs: renewal window, experience mod, days since your last touch, and assigned risk status.

Every account gets a score by adding up four buckets. Nothing else feeds it, and no part of it is AI generated.
Renewal window: 31 to 60 days out scores 50, 30 days or fewer scores 30, 61 to 90 days scores 20, further out scores 8, no renewal date scores 0.
Call timing rule: accounts with a mod above 1.00 or a seat in the Assigned Risk pool are scored on a tighter clock. They score full points only at 30 to 60 days out, less inside 30 days, a little at 61 to 75 days, and nothing past 75 days. Those accounts get the drip emails first and the call when the renewal is close enough to act on.
Experience mod: 1.35 or higher scores 30, 1.15 to 1.34 scores 20, above 1.00 to 1.14 scores 12, 1.00 or below scores 0. No mod on file scores 0.
Days since last touch: never contacted scores 25, 90 days or more scores 20, 45 to 89 scores 12, 21 to 44 scores 6, more recent than that scores 0.
Assigned risk: 25 points when the account is flagged as being in the NJ Assigned Risk pool.
The maximum is 130. Ties break in favor of the sooner renewal date.

Today call queue

The top 5 accounts by priority call score, plus anything you added to today's list yourself.

The queue starts as the 5 highest priority call scores among live accounts. Won, lost, and parked accounts are excluded.
Anything you add through 'Add to today's calls' or 'Send to today's list' joins the same queue, so the count grows past 5. Manual entries carry an 'Added by you' tag.
Membership is stored per date, so a manual add applies only to the day you added it.
Done logs a touch against the account. Skip hides it for the rest of the day. When an entry leaves the queue, the next ranked account backfills so you never drop below 5 while untouched candidates remain.

Accounts list order

A separate score used to order the Accounts tab, weighted toward follow-ups you promised and appointments today.

Overdue follow-up adds 120, a follow-up due today adds 100, an appointment scheduled today adds 80.
Renewal timing adds 60 within 30 days, 40 within 60, 25 within 90, 10 within 120.
An engaged, active, or closing relationship adds 15. Priority 1 adds 10, priority 2 adds 5. An experience mod above 1.00 adds 10.
Recent activity pulls the account down so you do not call it twice: called today subtracts 30, and a message left within the last two days subtracts 15.
Won, lost, and parked accounts are left out of the list entirely.

About to slip

Accounts renewing in 31 to 38 days with no call logged and no contact date on file.

The 31 to 60 day band is the window where an incumbent has not locked the renewal yet. An account with 38 days or fewer left in that band and zero recorded contact is about to fall out of it untouched.
The account leaves this list as soon as any call is logged against it.

Account state

Relationship temperature

Derived from your logged calls, sent emails, and the pipeline phase. It never cools on its own.

Cold: nothing logged yet.
Warming: at least one call logged or one email sent.
Engaged: a call logged as a warm lead, or two or more calls with at least one live conversation.
Active: the pipeline phase is info gathering, policy received, ACORD in progress, handed to the account manager, or quote received.
Closing: the pipeline phase is proposal presented.
Parked: the phase is lost or parked, or a call was logged as not interested. Won accounts show as won.
The temperature only moves up automatically. It comes back down when you park the account or set it yourself.

Psychology stage

Where the buyer is in their own thinking, derived from pipeline phase, mod, OSHA history, and renewal timing.

Oblivious: no phase movement yet, no trigger on the account.
Curious: early outreach or interest building with no strong trigger.
Triggered: early outreach where the mod is above 1.00, there are OSHA citations on file, or the renewal is 60 days or fewer away.
Shopping: info gathering, policy received, ACORD in progress, or handed to the account manager.
Comparing: quote received, or proposal presented while the relationship is not yet closing.
Switching: proposal presented with a closing relationship, or the account is won.
The stage drives the suggested ask and tone on the call sheet.

Pipeline phase

Set by you, not calculated. It records where the opportunity actually stands.

Phase is a manual field on the account. Nothing changes it automatically.
It does feed other numbers: temperature, psychology stage, and whether the account appears in ranked lists at all. Won, lost, and parked drop out of every queue.

Signals

Rule-based observations pulled from data already on the account. None of these are AI generated.

Assigned risk pool status, taken from the NJCRIB plan file match.
Experience mod level and the direction it moved from the prior year.
OSHA citations recorded on the account.
Renewal proximity: within 30 days is flagged as late, within 60 as the open window.
Going stale: no contact in 14 days on an account that is not cold.
A standing risk note for the trade category.

Data sources

Trade category

Assigned by keyword rules run against the imported trade text and company name.

The trade text and company name are combined and matched against keyword rules in order. The first rule that matches wins, so 'BUILDING MATERIALS/LUMBER (RETAIL DEALERSHIP)' becomes Materials supplier.
Anything that matches no rule falls into Other. Roughly 49 accounts in the current book sit there because their source text is too generic to classify.
The category drives trade labels, the risk note on the call sheet, service matching, and the Blindspot map.

Experience mod

Imported from the NJCRIB experience rating file or entered by hand. Change is current minus prior year.

The weekly NJCRIB job parses the experience rating modification file and loads every employer record, including the governing class code and the rating effective date.
Values on an account can also be typed in manually, which overrides the imported number.
Mod change is simply current minus prior. A positive number means the mod went up.
A mod is a published rating factor, not a quote. It says what the carrier will multiply premium by, not what the account currently pays.

Assigned risk

Matched against the NJCRIB 'Plan Risks by Expiration Date' file on normalized employer name plus city.

The weekly plan risks job loads every employer in the NJ Assigned Risk pool with its servicing carrier and policy expiration date.
Each plan row is matched to a mod record by normalized employer name, preferring a row where the city also matches. Normalization uppercases the name, strips punctuation, and removes suffixes like INC, LLC, and CORP.
A match is a name match, not a confirmed identity. Two similarly named businesses in the same city can collide, so confirm the pool status on the call before leaning on it.
Accounts in the pool are paying pool rates, which is the strongest single reason to call them.

Renewal date and days to renewal

Comes from the CRM or a verified bureau date. Past dates are projected forward one year.

A date carried over from the CRM is treated as unverified until it is confirmed against the bureau file.
Days to renewal counts from today to that date. If the stored date has already passed, it is projected forward a year so the account keeps cycling instead of dropping out.
No date on file means the account scores zero on the renewal portion of every ranking.

Blindspot map counts

Counts accounts whose trade category appears in that service's target trade list. It is a fit signal, not a confirmed finding.

Each service in your catalog carries a list of trade categories it applies to, plus a flag for whether most incumbent agents provide it.
A row counts every live account whose trade category is on that service's list. Class code review, for example, targets nine trades where misclassification is common, which is why it matches a large share of the book.
Nothing here has been checked against an actual policy. It says 'this trade is where the problem is usually found', not 'this account has the problem'.
Accounts inside a row are ordered by soonest renewal.

Mod leads import stats

Reported by the last run of each weekly NJCRIB job.

Parsed is how many employer records the parser read out of the source file. A run that reads fewer than expected fails on purpose rather than loading a partial book.
Inserted and updated split those rows into new employers and employers already on file.
Matched is how many assigned risk plan rows were tied to a mod record by name and city.
Snapshot date is the day the file was downloaded, not the date the bureau published it.

Activity and coverage

Day so far

Counts what you logged today across the whole book.

Calls counts every call logged with today's date, whatever the outcome.
Conversations counts calls logged as talked to them or warm lead.
Warm leads counts calls logged as a warm lead.
Emails counts drafts marked sent today. Nothing is counted automatically, so the tally reflects only what you logged.

Data coverage counts

Splits accounts with no date at all from accounts carrying an unconfirmed CRM date.

No date on file: nothing to work with, so the account cannot be ranked on renewal timing.
Unverified CRM date: a date exists but has not been confirmed against the bureau file. It is used for ranking, but treat it as approximate.
The two together are what the coverage total reports, which is why that number is larger than the no-date count on the pipeline.

AI content

AI generated content

Call scripts and email drafts are model output with a rules-based fallback. Review before sending.

Scripts and email drafts are generated from data already on the account: trade, mod, renewal timing, pool status, and your service catalog.
When the model is unavailable, a deterministic rules fallback writes the same content from the same inputs, so the feature never blocks a call.
Nothing is sent automatically. Every draft is editable and saved back to the account before you copy it.

AI researched contacts

Contact candidates found by research, saved as unverified until you confirm them on a call.

Enrichment looks for likely owners and decision makers from public sources, and returns a confidence level and source links for each candidate.
It never invents an email address or phone number. A field it cannot source comes back empty rather than guessed.
Nothing is saved without you accepting it. Accepted candidates are stored with an 'AI, unverified' badge until you mark them verified after confirming on a call.

Territory and compliance

Territory

Distance in miles between the centre of the account's ZIP code and the centre of your home base ZIP.

You set a home base ZIP and a radius in Settings. Every account is then tagged with its county and its straight line distance from that ZIP, using published ZIP code centre points.
It is a straight line distance, not drive time, so an account listed at 28 miles may be a 40 minute drive.
Accounts with a missing or unrecognised ZIP are never filtered out. They show without a distance so you can judge them yourself.
The lead filter converts your radius into a short list of ZIP prefixes, which is why a lead just outside the line can still appear.

Contractors only

Keeps leads whose governing class code is a construction or trade code, and drops clerical and office codes.

The filter reads the governing class code published in the rating bureau file. Codes on the contractor list are kept: concrete, electrical, drywall, carpentry, roofing, plumbing, HVAC, masonry, excavation and the rest of the building trades.
Clerical and office codes are always dropped, including 8810 clerical office employees, 8742 outside sales, and the executive and professional codes.
When a lead has no usable class code, the class description is read instead and matched against the same trade words.
Premium is not published in the file, so small accounts cannot be filtered by premium size. Clerical exclusion is the closest available proxy, and the list still needs your eye.

Automatic lead push

Each morning the job adds qualifying rating bureau leads straight onto today's call list.

A lead qualifies when all of these are true: mod above 1.00 or assigned risk, renewal 30 to 60 days out, a contractor class code, and inside your territory radius.
Leads already claimed as accounts in your book are skipped, so nothing is added twice.
The number added per morning is capped by the limit in Settings, highest mod first.
Every pushed account is stamped with the time it was pushed and appears in the briefing under new from the rating bureau file.

Conflict scrub

Home office clearance status for each account: not scrubbed, sent, cleared, or conflict.

Exporting the scrub list downloads a CSV of every account still marked not scrubbed and marks them as sent.
When the home office replies, mark the sent accounts cleared. Cleared accounts are protected and are never included in a later export.
Accounts the home office flags can be set to conflict so they stay out of every future list.
The export carries account name, address, county, class description, mod, current carrier, assigned risk flag, expiration date, and your producer and agency name.

Assigned risk view

The pool accounts, shown with the angles that separate your call from the other agents working the same list.

The assigned risk flag comes from matching the plan risks file to the mod file on normalised employer name plus city.
Pool accounts are the most vulnerable book in the state, which also means they are the most heavily called. The differentiators shown above the list are fixed talking points, not model output.
Combined with the call timing rule, these accounts are worked at 30 to 60 days out after the drip emails have run, not at 90 to 120 days.