Branching on customer data, spend and sentiment
Branch a workflow on customer data, spend and sentiment.
Read the lesson
Goal: Design a workflow with IF and ELSE paths where every path ends in a safe action.
You need: Lesson 1 of this course. Core, Growth or Franchise.
Why this matters
Customers are not all the same. A first-time caller, a $5,000 replacement customer and an angry customer need different next steps. One message for everyone is either too cold for your best customers or too pushy for the rest. Branching lets one workflow treat each customer the right way. It also creates a new risk. A customer who matches no branch can fall through and get nothing. That is how a good customer ends up ignored.
How branching works
A Condition step asks one yes-or-no question. Yes goes one way. No goes the other way. You can chain conditions. The No path of the first condition can hold a second condition.
Think of it as IF, ELSE IF, ELSE.
- IF the first test passes, take path 1.
- ELSE IF the second test passes, take path 2.
- ELSE take the last path.
Order matters. The first test that passes wins. Put the most urgent test first. A frustrated customer who also spent a lot needs the callback first, not the thank-you.
The last ELSE path must have an action. If it is empty, customers who fail every test get nothing. Give it a safe default. A simple note, a task or a plain thank-you is enough.
What you can branch on
A Condition reads a field you type as text. It compares it with Equals, Not equals, Contains, Greater than or Less than.
Customer data. Tags, the contact's status and custom fields are the usual choices. Use Settings → Contacts & Pipeline → Custom fields to add the fields you need. Field types include Number, Dropdown, Yes/No and Date. Pick Number for anything you compare with Greater than.
Spend. ServiceAgent does not keep a lifetime spend total on its own. Add a Number custom field, such as Total spend. The office updates it, or a workflow updates it with Update Contact. Then a Condition can test it with Greater than.
Add a Number field for Total spend
Add Field
Sentiment and scores. Use the survey triggers. Low Score Received and High Score Received start a run on a survey score. Survey Completed starts a run on any answer. The AI Tag Conversation step can tag a conversation, and a later Condition can read the tag. The Create tab also has a Negative Sentiment Alert recipe to start from.
Not every field path you type will read a value. Always test the exact field path with Test before you publish.
Design the branches on paper first
Draw the branches before you open the builder. It takes 5 minutes and saves an hour.
- Write the trigger at the top.
- List the groups of customers you expect, from most urgent to least.
- Write one test for each group. Make each test a single field and a single compare.
- Write the action for each group.
- Write the safe default for everyone else.
Then ask: "Who could match two tests at once?" Put the more urgent test first for them. Ask: "Who could match no test?" Make sure the default suits them. Ask what happens when the field is empty. An empty Total spend is not Greater than 1000. So the customer falls to the next path. That is usually fine. Just know it will happen.
Four sample contacts across three paths
| Sample contact | Score, spend | Path taken |
|---|---|---|
| Unhappy customer | 2, $3,000 | IF: owner call today |
| Top customer | 5, $1,800 | ELSE IF: thank-you call |
| Happy small customer | 5, $400 | ELSE: tag survey-done |
| New customer | 4, empty | ELSE: tag survey-done |
Keep each workflow to 3 paths or fewer. If you need 5 paths, split it into two workflows with clear triggers. Small workflows are easier to test, easier to fix and easier to hand to the next Operator.
By business size
An owner-operator rarely needs more than 2 paths. The owner knows most customers by name. A fleet benefits most from branching. The office cannot remember 3,000 customers, but a Condition can. A multi-location client may branch on location. Test that the location field holds the value you expect, in each location. On Core, remember the 10 active automations. One workflow with 3 paths counts as 1. Three separate workflows count as 3.
Keep a record of each branch
Write each workflow's branches in your Operator notes in one line per path. For example: "Path 1: score under 3, owner task. Path 2: spend over 1000, thank-you call. Default: tag." When the owner asks why a customer got a call, you can answer in 10 seconds. When you hand the account to an assistant, they can read the logic without opening the builder. Update the note every time you change a test or an action. Date each change, so you can match it to a change in the KPIs. Check the note against the Summary tab once a quarter.
The review rule
Branching must never decide who gets a review link. Google does not allow review gating. The FTC's 2024 rule on consumer reviews bans review suppression. So every customer gets the review link through Review requests & settings. A low score only adds a step: a call from the owner. It never removes the link.
Review gating vs the right way
- Only happy customers get the link
- Unhappy customers are hidden
- Everyone gets the link
- Unhappy customers get an owner call too
Build it in ServiceAgent
This build follows up a post-service survey.
- Open
Workflows→Create→Build from scratch. Name it "Survey follow-up." - Set
Trigger TypetoSurvey Completed. Add step→Condition: the score is Less than 3. This is the IF path.- On the Yes path,
Add step→Create CRM Task.Title"Call unhappy customer today."Ownerthe business owner.PriorityHIGH. - On the Yes path, add
Send Notificationto the owner as well. - On the No path, add a second
Condition: "Total spend" Greater than 1000. This is the ELSE IF path. - On its Yes path, add
Create CRM Task: "Thank-you call to a top customer."PriorityMEDIUM. - On the last No path, add the safe default:
Update Contactto add the tagsurvey-done. - Check every path ends in an action. Read the
Canvasfrom left to right. - Click
Testwith 3 sample contacts: a low score, a high spender, and a happy small customer. Watch each one's path. - Click
Publish. Then check it works: fill in one real survey yourself and readHistory→View run history.
The survey follow-up: IF, ELSE IF, ELSE
- Survey CompletedAny survey answer starts a run
- Score Less than 3?YesCreate CRM Task, HIGH: owner calls today. Send NotificationEnds hereNoGo to the next test
- Total spend Greater than 1000?YesCreate CRM Task, MEDIUM: thank-you callNoSafe default: Update Contact, tag survey-done
- Every survey leaves a markCount runs in History
Worked example: Cedar HVAC
Cedar HVAC has 9 techs on Growth. It signed up with ServiceAgent directly, so it is a client-signed account. Cedar completes about 150 jobs a month. Each customer gets the post-service survey. In a normal month, about 10% give a low score. That is 150 × 10% = 15 customers.
Each owner call takes about 10 minutes. That is 15 × 10 = 150 minutes, or 2.5 hours a month. Without the workflow, those 15 customers would wait until they posted a bad review. With it, the owner calls each one the same day.
Cedar HVAC owner calls
The first draft had an empty last ELSE path. A happy customer who spent $400 failed both tests and got nothing. The safe default tag fixed that. Now every survey leaves a mark, and you can count them. In Contacts, use Filter by Tags to list everyone tagged survey-done. That count tells the owner how many surveys came back each month.
The $400 customer before and after the fix
- Score 5, spent $400
- Fails both tests
- Falls to an empty path
- Gets nothing, leaves no mark
- Score 5, spent $400
- Fails both tests
- Falls to the safe default
- Tagged survey-done and counted
Common mistakes
- Mistake: An empty last ELSE path. Fix: Give it a safe default action.
- Mistake: Putting the thank-you test before the complaint test. Fix: Put the most urgent test first.
- Mistake: Comparing a text field with Greater than. Fix: Use a Number custom field for amounts.
- Mistake: Sending the review link only to happy customers. Fix: Send it to everyone. A low score only adds an owner call.
- Mistake: Publishing without testing each path. Fix: Test one sample contact per path.
Do this now
Draw your branch on paper first: IF, ELSE IF, ELSE. Build it in your practice account and test 3 sample contacts, one per path. Deliverable: a screenshot of the Canvas with every path ending in an action. Time: 25 minutes.
Practice the conversation
- Owner I only want happy customers to get the review link.
- Operator I understand. But Google does not allow that, and the FTC bans it.
- Owner So angry customers get the link too?
- Operator Yes. Everyone gets it. The difference is what else happens.
- Owner What else?
- Operator A low score creates a task for you to call them the same day.
- Owner That could save the review.
- Operator Often it does. And your big spenders get a thank-you call.
- Owner And everyone else?
- Operator They get tagged, so nobody falls through.
Quick check
- IF sentiment is negative, task a callback. ELSE IF spend tops $1,000, task a thank-you. The last ELSE has no action. What happens to a happy customer who spent $400?
- They take the thank-you path.
- Nothing happens. Give the last ELSE a safe default action.
- They take the callback path.
- The workflow fails and retries.
- A client completes 200 jobs a month and 8% give a low score. Each owner call takes 10 minutes. How many hours of calls is that a month?
- 1.6
- 2.7
- 16
- 2.0
- Which field type should "Total spend" be, so a `Condition` can test it with Greater than?
- Text
- Long Text
- Number
- Multi-Select
ServiceAgent UniversityLearn to run a service business on autopilot. Free courses, Blueprints and certification for Service Operators.