Product
Seven stages between the file and the windshield.
Nothing here is a module you buy separately, and nothing here is a model you have to trust. Every stage is arithmetic a rep can argue with, which is the only kind of plan a rep will actually follow.
STAGE 01
The file you already have
A vendor territory workbook, as it arrives. The header row is detected rather than assumed, columns are mapped by suggestion and then shown to you for correction, and every row is validated before anything is committed. Nothing is written until you have seen exactly what a commit will create.
- Column mapping is suggested from the headers, including the vendor vocabulary a target file uses rather than a generic CRM’s.
- Duplicate detection returns a score, not a boolean. A name-only match caps at “review”; only corroboration from an address, phone or email pushes it to auto-skip.
- Existing accounts are recognized. These files mark one by putting the practice name in the NPI column and leaving the surname and street blank — miss that and the most valuable records in the territory import as unnamed providers.
- Geocoding is cache-first and works offline. No map key is required to plan a territory.
| NPI | Last | Specialty | City | ZIP | Wtd Decile | Volume |
|---|---|---|---|---|---|---|
| 1447‑•••••• | — | Endocrinology | Houston | 77024 | 10 | 5,562 |
| 1902‑•••••• | — | Family Practice | Katy | 77494 | 8 | 1,204 |
| Cypress Metabolic | blank | blank | blank | 77429 | — | 812 |
| 1336‑•••••• | — | Internal Medicine | Pearland | 77584 | 7 | 341 |
| 1558‑•••••• | — | Nurse Practitioner | Sugar Land | 77479 | 6 | 97 |
STAGE 02
Scored, then tiered — and the vendor gets the biggest vote
The vendor already did the hard part. Their weighted decile ranks each prescriber’s weight inside this territory, and it is 45 of the 100 points — dominant on purpose. We rank inside a decile, we do not second-guess it.
Script volume is log-scaled, and that is the load-bearing choice on this page. 5,562 prescriptions is not sixteen times more interesting than 341, and on a linear scale a handful of mega-prescribers flatten everyone else to zero — which turns a 539-name target list into a five-name one.
- Four tiers with the frequency each one promises: weekly, twice-monthly, monthly, and a Coverage tier that promises nothing and is worked when the route already passes the door.
- Two qualitative signals promote a record outright: an existing sample, and a prescription of the on-label drug. Both mean the conversation has already started.
- Tier sizes are yours. “Top 50” is a metro-sized number, not a law — a 90-name territory wants a Top 25, and the display name follows the size.
- A specialty nobody has characterized is slightly penalized, never zeroed. One unfamiliar label should not silently bury a cohort.
- Weighted decilethe vendor's own ranking45
- Primary volumelog-scaled30
- Secondary volumelog-scaled10
- Specialty liftconfigurable per vertical10
- Territory fitcost to serve5
DEFAULT_SCORE_WEIGHTS. The two volume slots take whichever columns your file carries — prescriptions, tests ordered, specimens received — and the specialty table is configurable, so the weights hold whatever the list counts. Two flat bonuses sit outside the hundred: five points for an existing sample, five for an on-label prescription.Stage 03
A weekday always means the same place.
This is the structural decision the whole plan rests on. Monday is one patch, Tuesday is another, and scheduling happens inside a zone rather than across the metro.
Why zones at all
A weekly provider can only be seen weekly without the rep crossing the city twice if their patch comes round on a fixed day. Pin the zone to the weekday and the cadence becomes drivable instead of aspirational.
Geography decides, capacity breaks ties
Balancing on workload alone produced a Friday running two towns twenty-four miles apart. A day the rep cannot drive is not balanced, it is evenly bad. Any move that pushes a zone past its spread limit is rejected outright.
When nothing has room
The overflow goes to the zone that stays most drivable, not to the emptiest one. Sending it to whichever zone was lightest is what produced a twenty-six-mile Tuesday the first time round.
Both of those rules exist because their naive versions shipped and failed. Pure clustering, meanwhile, produced a Tuesday carrying twenty cadence commitments — eleven field hours — against a Friday carrying six. Tidy zones, absurd workload. Every weekday gets the same number of hours, so the zones have to hold roughly the same amount of work.
STAGE 04
The cadence is honored exactly, or the plan says it could not be
Weekly means weeks one, two, three and four. Twice a month means weeks one and three, or two and four — never one and two, because two touches in consecutive weeks and then three weeks of silence is not “twice a month” to the person being visited.
- Week-sets are not alternated down the list. Doing that looks balanced on paper and put nineteen commitments on one Tuesday against eight on another, because the tiers do not interleave evenly. Each record takes whichever legal combination is currently least loaded for its weekday.
- When a zone’s cadence commitments alone overrun the clock, that is recorded as a signal about the zone — not silently trimmed to make the page look tidy.
- Everything overdue in the app is overdue against the tier’s own interval and nothing else. There is no second, invented benchmark.
| Tier | Providers | Frequency | Calls / month |
|---|---|---|---|
| High Potential | 31 | 4× / month124 calls / month | 124 |
| Top 50 | 50 | 2× / month100 calls / month | 100 |
| 2nd Top 50 | 50 | 1× / month50 calls / month | 50 |
| Coverage | 408 | opportunisticno commitment | — |
Calls scheduled per day. The goal band is 10–12; the number is whatever the clock and the geography actually allow.
STAGE 05
The day is built against a clock
Ten to twelve calls a day is the goal, and it holds where the geography allows it. The day itself is bounded by time: drive time between offices, about twenty-two minutes per call, and a lunch that eats an hour. The count falls out of that.
Some days land at twelve. Some land at eight. The plan says which, per day, before the week starts — because a plan that pretends otherwise is a plan the rep quietly abandons in week two.
- Window08:00 – 14:20
- Drive32.5 mi
- Deferred2
- Westheimer Endocrinology08:00
- Briar Forest Family Medicine08:26
- Gessner Park Internal Med08:56
- Dairy Ashford Primary Care09:26
- Eldridge Metabolic Health09:58
- Lunch — Memorial Drive Internal Med11:50
- 6 more stops through 14:20
- Kirby Endocrinology · Bissonnet Internal Med
STAGE 06
Then it leaves the software
The measure of this product is whether a rep can act on it without opening it again. So the day exports: the whole tour into Google Maps or Apple Maps, twelve events into a calendar, or a CSV for the manager who wants a spreadsheet.
- Map links are chunked at ten waypoints and overlap by one stop, so the second link starts where the first ended.
- Coordinates travel, not address strings. A practice’s postal address frequently resolves to a suite across the plaza from the door the rep needs, and the geocoded point is the one the optimizer planned around.
- Calendar events carry a fifteen-minute alarm and the stop’s own detail, so the phone is enough.
- Google MapsChunked at 10 waypoints, overlapping by one stop
- Apple MapsSame tour, native handoff on iPhone
- Calendar (.ics)One event per stop, with a 15-minute alarm
- CSVThe whole day, for the manager who wants a spreadsheet
STAGE 07
Automation, generated from the import and shipped switched off
The import knows your tiers and their intervals, so it can write the rules that keep the territory worked once the import is forgotten: a cadence-lapse rule per tier, an intake rule for new targets, an ownership rule for the ZIPs in the territory.
Every one of them arrives off. Each shows, in plain English, what it does and how many providers it would fire for right now — a live evaluation, not an estimate. Arming automation because somebody uploaded a spreadsheet is how a VP finds out about their own automation from their inbox.
- Nulls never coerce. A provider who has never been visited has not “stopped ordering”, and no condition reaches a match by accident.
- A condition can be undetermined as well as true or false — and when it is, the screen says so out loud instead of guessing in whichever direction is convenient.
- High Potential cadence lapse — 7 daysIf a High Potential provider goes 7 days without a logged call, put a task on whoever owns them.would fire for 31 providersOff
- Top 50 cadence lapse — 15 daysSame rule, at the Top 50 interval.would fire for 50 providersOff
- New target intakeA provider landing in one of this territory's ZIPs gets an owner and a first-detail task.would fire for 0 providers todayOff
And the CRM parts
It is still a CRM underneath
The planning layer is the reason to switch. The rest has to be there, or the planning layer is a toy sitting beside a system of record.
Practices, providers, opportunities
Accounts, contacts and deals in the vocabulary the field uses — practices and providers, not generic objects. Three pipelines, because a physician can start ordering next week, a group takes a quarter and a health system takes a year, and running those in one pipeline makes every forecast meaningless.
Activity that costs one tap
A call logged from the stop it belongs to, on a phone, in a parking lot. Activity capture that takes a desktop form is activity capture that does not happen.
Lists, filters and exports everywhere
Column selection, filters, sort, group-by, saved views, bulk select and CSV export on every list — built once as shared components rather than per object.
Territories and ownership
ZIP-based ownership rules generated from the import, arriving off, so nothing starts reassigning records the day after a file lands.
Adelo CRM is configured per sales motion. The examples on this page use a diagnostics vertical — deciles, script volumes, physician offices — because that is the motion it was built against. The object labels, personas, pipelines, tiers and scoring weights are all configuration. If you sell a device into the same offices, the shape holds and the vocabulary changes.
Bring a territory. We will run all seven stages on it.
You get back the scored file, the zones, the four-week plan and the routed days — the same artifacts a rep would get. It is a better proof than anything on this page.