How to automate ServiceMinder routing

Technician in a red work suit stepping out of a white service van in a driveway
Photo: Tima Miroshnichenko on Pexels (Pexels License)

The optimizer in ServiceMinder is the easy part of routing. It’s fast, and it orders a route well. The hours go on deciding which tech gets which stop, reading the rules typed into the notes, and checking the result. To automate routing, you have to automate those three things and still leave your office in charge. This guide covers how the job works by hand, what makes it hard, how to test a plan, and what to check every day, whether or not you ever use Alvin.

On this page
  1. How ServiceMinder routing works by hand
  2. What the optimizer does and doesn’t do
  3. What makes it hard to automate: the notes
  4. What makes it hard to automate: tech areas
  5. What makes it hard to automate: keeping the office in charge
  6. How to test a plan before you trust it
  7. What to check every day
  8. Your options for automating it
  9. Where Alvin fits
  10. Questions
  11. Sources
The short answerFirst put your notes in one format, so anything can read them. Then write down each tech’s area and a normal day’s work. Then test any tool on days you’ve already run, before it touches a live day. Start with draft routes that a person commits, and let the commits go automatic only once the checks stay clean.

How ServiceMinder routing works by hand

Labels here are from the Route Builder as we’ve seen it. Your account’s setup may differ. For a recurring-service office, the job usually goes like this:

  1. Visits land in the queue. Recurring visits are generated on their schedule into the day’s queue, with no tech and no time. Overdue ones sit in the Queued Backlog.
  2. Open the Route Builder for tomorrow. The queue is on the left, filterable by service and GeoZone. The settings and the techs to route are in the middle, and the route solution is on the right.
  3. Pick a group of stops for one tech. Use the filters, or Lasso on the map, and keep in mind the tech’s side of town and how much they can do.
  4. Optimize. ServiceMinder orders the stops and shows each one’s arrival time, plus the route’s driving, miles and return time. This saves a route solution, not a change to the calendar.
  5. Check it against the notes. Open what needs opening and drag stops that landed wrong. The Existing Schedule below shows what each tech already has that day.
  6. Press Commit! That writes the route to the calendar. Then go back to step 3 for the next tech.
  7. Deal with what’s left. Leftovers, rain-outs and holidays go through the day view’s Re-Queue, Bump and Move Queued Appointments To.

An example: with 8 techs and 140 stops, that loop runs 8 times, every working day, against notes nobody else has read.

Alvin task 'Bill this week's quarterly services and sync payments' with its work log and receipts, waiting on the owner's approval of an example card charge.
How a job looks in the Alvin app: a work log and a receipt for each step. This example is billing, since routing isn't in the demo (demo data).

What the optimizer does and doesn’t do

The optimizer is good at order. It takes the stops you give it and finds a sensible sequence, and if your office commits what it produces, you’re already getting most of what ordering offers.

It doesn’t know that the customer on Elm Street wants one particular tech, or that a house must be first because a second tech is meeting there. It works from the structured service window, not from the notes. That’s why the check in step 5 exists, and why pressing Optimize more often doesn’t remove it.

The Route Builder also has settings such as Load %, Duration Scale, Disable Constraints, Optimize Vehicles and Balance Return Times. Try changes on a saved solution and compare, since nothing reaches the calendar until someone commits.

What makes it hard to automate: the notes

The real rules of your day are free text, typed by different people over several years. Anything that automates routing has to read them, and misreading one is worse than not reading it. Here are the kinds of note you’ll likely find, with made-up examples:

Kind of noteThe trap
Tech rule
SEND PJ
Notes use nicknames, so you need a list of who’s who. An unknown name must not be dropped silently.
Soft wish
SEND PAT IF POSSIBLE
Read as a hard rule, it costs driving. Decide what a wish is worth.
Time
TIME: 9AM-12
The 12 means noon, not midnight. Decide what “morning” ends at, and which windows are firm.
Days
DAY: NO FRIDAYS OR SATURDAYS
That’s two days. A simple reader catches only one.
Order
FIRST ON ROUTE
The optimizer can move the stop afterward. Check it after every re-order.
Date pin
KEEP ON 11/17
A rule that says “don’t move this”, so a tool that moves stops between days must see it.
Paired houses
SCHEDULE WITH 14 ELM ST
The other house may not be in the queue. Say so instead of guessing.
Job size
12 GALLONS
May mean two techs. Set the amount yourself. Price is a poor stand-in.

The most useful habit here is cheap, and you can start today: pick one format and have everyone use it. For example, TIME: and DAY: labels, full tech names, and one instruction per line. Then pull a few weeks of notes and read the ones that don’t fit. That list shows you what your office really asks for, and it helps your next new hire too.

The rule for any reader, a person’s or a program’s: if it can’t read a note, it should quote the note to someone. It should never guess, and never skip it.

What makes it hard to automate: tech areas

Who gets which stop is mostly knowledge in one person’s head. Each tech keeps a territory, often near home. Each has a normal number of stops. Your history already holds most of this: your committed routes show where each tech works and how much they do. Things to get right:

  • Learn from recent routes. A few weeks is usually enough to see a pattern, and old routes include techs who have left.
  • Start points. Check whether techs really start from the address on file, and whether the drive home should count.
  • Who’s on. Call-outs and small Saturday crews change the day. A plan that assumes everyone is working will be wrong before it starts.
  • Things the map misses. A river with few bridges, or a tech who has to finish near the interstate. Write these down, because no history will explain them.

What makes it hard to automate: keeping the office in charge

A routing tool nobody trusts gets redone by hand, which is worse than having none. The office needs to stay in control of these:

  • A cutoff. Changes land all morning. Pick a time the plan runs from, and know that a stop removed after that stays on its route until someone takes it off.
  • Draft or commit. Drafts are safest while you’re learning. Later, some offices prefer routes committed to the calendar so they can fine-tune there. Either way, make sure you know how to uncommit.
  • What committing sets off. Find out whether a commit triggers customer confirmations in your account before anything commits for you. Try it first on one small route.
  • Leaving people’s work alone. A rerun should replace only its own uncommitted drafts. A tech your office has already routed keeps that route.
  • Login. Give the tool its own access if you can, and keep it out of chats and shared documents.

How to test a plan before you trust it

Use days you’ve already run. For each one, plan the day’s appointments as if they were still queued, using only the routes from before that day and only the techs who worked it. Then compare:

  • Notes broken, by the plan and by the routes you ran.
  • Stops late for their window.
  • Stops left over.
  • Latest finish and total crew hours.
  • Driving.

Be careful with the driving number. We got this wrong ourselves. Our first comparison scored the office’s routes in the order they ran and our plan in an order tuned to the same estimate that scored it, and that made our plan look far better than it was. When we compared both in the same order, most of the gain went away. What was left was mainly the routing time saved. So compare both plans in the same order, using one source of drive times, ideally ServiceMinder’s own. Be wary of any vendor’s savings figure that doesn’t say how it was measured, including ours.

After past days, run a week of live days as drafts nobody has to use, and have the person who routes today mark them up. Separate cosmetic preferences from real errors. Fix the errors first, and only then decide about committing.

What to check every day

Whether a person or a tool built the routes, this list takes minutes and catches most of the trouble:

  • Every queued stop is on a route, or listed as left over.
  • Every note that couldn’t be read has been seen by a person.
  • Timed stops arrive inside their window, by ServiceMinder’s own times.
  • First-on-route stops are still first.
  • “Send” and “do not send” stops are with the right tech.
  • Anyone off today has no route, and every stop that names them is handled.
  • Large jobs have their second tech.
  • No route has an odd finish time or a long drive. Look on the map for routes that double back.
  • Stops canceled since the cutoff are off their routes.

Your options for automating it

  • Clean notes, a written area list and the optimizer. A faster routing job at no extra cost, and it makes every later option work better. A person still does the loop every day.
  • Your own scripts. Full control, if you have someone technical who stays. Expect upkeep whenever ServiceMinder changes, and ask what your plan allows first.
  • An assistant that does the job in your account. It takes the daily loop, with a person checking the results. Day-of judgment stays with you, and it needs setting up with your office.
  • Other routing software. You get whatever that product does, plus a migration when your calendar, billing and history all live in ServiceMinder.

Where Alvin fits

Alvin is the third option. It signs in with your office’s login, reads the next day’s queue and notes, turns the notes into rules, learns each tech’s area from your past routes, and plans one route per tech by code. It builds each route in the Route Builder, checks it against the notes and ServiceMinder’s drive times, then leaves drafts or commits, as your company chooses. Then it emails the office what to look into, with a link to each appointment. It doesn’t send customer messages, and a note it can’t read goes to a person.

It’s turned on one company at a time today, and only for ServiceMinder. The details are on the AI for routing page.

Frequently asked questions

Can ServiceMinder route automatically?

ServiceMinder’s Route Builder can order stops for you with its optimizer. It doesn’t decide from your notes which tech should get which stop, so someone still has to do that part or hand it to a tool that can.

Does ServiceMinder’s optimizer read appointment notes?

In the Route Builder we’ve worked with, no. It works from the structured service window. A time, a tech or a first-on-route instruction typed into the notes has to be checked by a person or by another tool.

Does saving a route change the calendar?

No. A route solution is saved on its own. The calendar changes when someone presses Commit, and a committed route can be uncommitted.

Will automated routing cut our driving?

Don’t assume so. If your office already uses the optimizer, the bigger gain is usually the office’s time. Test on your own past days, and compare both plans in the same order with the same drive times.

Can Alvin route for any ServiceMinder company?

Not yet. Alvin’s routing is turned on one company at a time and set up with the office. It works with ServiceMinder only today.

Sources

Hand the routing job to Alvin.

Routing is turned on one company at a time, for offices on ServiceMinder. Join the waitlist, or tell us about your routes.