A customer onboarding workflow: a clean handover from sales, a confirmed setup, and a clear status

SynaptaGrid8 min read

#workflow#customers

In short

A customer onboarding workflow is the path a new customer takes from the moment sales closes the deal to the moment their account is set up and live: customer success checks that the handover has what onboarding needs, implementation sets the account up and confirms it, and the customer ends onboarded, sent back to sales or blocked. Most onboarding problems start at the handover, where what was sold, who the contacts are and when the customer expects to start are missing or wrong. A good workflow checks the handover before any work begins, gives implementation an honest way to say this cannot be delivered as sold, and writes the status onto a record so everyone sees where the customer stands. We built a small version in our own test workspace in which an AI tool, written for this job and used as an action in a normal step, reads the handover note and writes a short check onto the customer record before customer success reviews it. We ran it three times, once to onboarded, once with the handover sent back to sales and once blocked at setup, and checked the customer records afterwards.

What a customer onboarding workflow is

Onboarding is the work between a signed deal and a customer who is using the product: collecting what was sold, creating the account, configuring it, and telling the customer it is ready. A workflow writes down who does each part, in what order, and what happens when something cannot proceed. Without one, the handover is a message in a chat, nobody owns the start date, and the customer asks where things stand.

The workflow does not set up the account itself. It makes sure the handover is checked before work starts, that the people doing the setup can stop honestly when the deal cannot be delivered as described, and that every customer ends in a status the whole team can read.

The version we built

An AI tool, used as an action in a normal step, checks the handover note and the workflow writes the check onto the record. Customer success then reviews the handover and can send it back to sales. Implementation confirms the setup or says it cannot deliver as sold. Automatic steps create the customer record and set its status and live flag at each ending.

The steps to include, and who owns each

Record the customer
When sales hands a deal over, a record is created with the customer name, a status of handed over from sales and a live flag that is off. Every later step reads from and writes to that record.
Check the handover note
This is where an AI tool helps. It reads the note sales wrote and returns a short summary and a list of what the note does not say, such as missing contacts or an unclear start date. The next step writes the check onto the customer record, so customer success sees it with the handover instead of reading the note cold.
Review the handover
Customer success reads the AI check against the note and decides whether the handover has what onboarding needs: what was sold, the contacts, and the start date the customer was promised. This is the step most processes skip, and the one that prevents the most rework. The tool does not decide; a person does.
Send it back to sales
A handover with missing information goes back to sales with a clear status, instead of onboarding starting on guesses. Sales supplies the missing detail and hands over again, which is a new request.
Confirm the setup
Implementation creates the account and the settings the customer bought, then confirms it is ready. The person who does the work confirms it, so the live flag reflects the real state.
An honest way to stop
Sometimes what was sold cannot be delivered as described. Implementation can say so, the record is marked blocked, and the problem is visible instead of hidden in a late project.
Mark the result on the record
Each ending sets the status and the live flag: onboarded and live, returned to sales, or blocked. The workflow does this itself.
Tell the customer
Welcoming the customer and sending login details usually happens in another system or by a person. Decide who does it and add that as its own step.

Where approvals and checks belong

Put a check where a mistake is expensive and cheap to catch. The cheapest place to catch a gap is at the handover, before anyone starts configuring, and the most expensive is a customer who was promised something that was never possible. So the handover is checked first and the setup is confirmed by the people who did it.

Name each owner by role, such as customer success and the implementation team, rather than by person. If larger customers need a kickoff call or an extra approval, add that on your own policy; we did not build it.

The old way and the new way

The table compares onboarding handled by chat and a spreadsheet with the same customers run as a workflow.

The handover from sales
The old way: A message with whatever the rep remembered
The new way: A step that checks it before work starts
Reading the handover note
The old way: Customer success reads it cold
The new way: An AI tool summarizes it and lists what is missing first
A handover with gaps
The old way: Onboarding starts and the gaps surface later
The new way: Sent back to sales with a clear status
A deal that cannot be delivered as sold
The old way: Discovered late in the project
The new way: Blocked at setup, visible on the record
Whether the customer is live
The old way: Ask whoever did the setup
The new way: The record’s live flag says so
Finding a customer’s status
The old way: Search chat and the spreadsheet
The new way: Open the record and read its status
Judgement
The old way: Still customer success’s and implementation’s
The new way: Still theirs: whether the handover is complete, whether it can be delivered

How we decided

We built the version in the diagram in our own test workspace in SynaptaGrid Automation, with a small record type in SynaptaGrid Data for the customer, and ran it three times in November. The AI part is a custom tool we wrote for the job: a prompt that says how to review a handover, an input that takes the note, and an output with a summary, a list of what is missing and a ready flag. In the workflow it is an action in an ordinary step, set up like any action that calls another system, and the next step writes the summary onto the record. In the first run customer success accepted the handover and implementation confirmed the setup. The workflow created the customer record with the status Handed over from sales and the live flag off, the AI tool checked the note and its summary was written onto the record, then customer success accepted, implementation confirmed, the same record was updated to Onboarded with the flag on, and the run finished with the onboarded result. In the second run customer success sent the handover back for missing information: the run skipped implementation, marked the record Returned to sales and ended. In the third run the handover was accepted and implementation said it could not deliver as sold, so the record ended as Blocked.

We checked the three records in Data afterwards. Each carried the AI check: a summary of the note saying that contacts were not provided and that data had to be moved from the customer’s spreadsheets before go-live. The onboarded one showed Onboarded with the flag on. The returned one showed Returned to sales, and the blocked one showed Blocked, both with the flag off. The AI tool did not work first time. Our earlier attempts to call it failed until we named the model in full in the tool’s settings, and our first version did not use the tool’s output in a later step; the version described here is the one that ran three times.

We chose a handover check, a way to stop that is not a rejection, and a live flag on the record because those are the places onboarding usually goes wrong. The choice of steps is ours and a judgement; it is not a standard.

Six limits, so this guide does not claim more than the runs showed. The tool saw one handover note, so we did not test how its check changes for a complete note, and it flagged the missing contacts but did not say who owns the spreadsheet import. We wrote its summary onto the record but not the list of what is missing or the ready flag. One test account held every human step, so the runs show the sequence and the branching, not hand-offs between different people. We did not create an account or settings anywhere: the workflow works on a record about the customer, and the live flag is a stand-in for the customer really being live. We did not build a kickoff meeting, a link between a returned handover and its resubmission, or a check that the contract matches what was set up. We did not test reminders when a step waits too long or notifications to the customer. And the template carries one test account as every approver and the actions tied to our test record type and our test AI tool, so to reuse it you would change each step to a role and point the actions and the tool at your own, which we did not do.

Standards and accepted practice behind this

These are the practices the guide rests on. We align with them. This page does not claim a certification.

A defined handover
Teams that pass work between functions usually agree what a complete handover contains. The review step is where that agreement is applied to each customer.
Single owner per step
Each step has one accountable owner, so a customer’s status never depends on who happened to be copied on a message.
Human oversight of AI
Where AI helps with a decision, accepted practice is to keep a person accountable for it and to let them see the AI’s input and output. Here the tool drafts a check and a person decides whether the handover is complete.
Source: EU AI Act (Regulation 2024/1689), Article 14: human oversight (opens in a new tab)
Segregation of duties
The person who sold the deal, the person who checks the handover and the person who sets the account up are best kept distinct, so each step is a real check.
Source: NIST glossary: separation of duty (opens in a new tab)
Records of customer data
Customer details are personal or business data and fall under your privacy and retention rules. Check yours before you decide what the record holds and for how long.
BPMN
The standard notation for drawing a process. The diagram above is a plain flowchart, and a BPMN diagram of the same steps would show the same decisions.
Source: OMG: BPMN 2.0 specification (opens in a new tab)

Where SynaptaGrid fits

SynaptaGrid Automation lets you write the steps as a workflow, assign each decision to a person, use an AI tool you wrote as an action in a step, pass its result to the next step, create and update records in SynaptaGrid Data as part of the run, and see where every customer stands. You can start from the version above, change the steps to match how you onboard, and keep the result as a template.

Check the limits above before you rely on it: whether each step is assigned to a role rather than a person, whether the setup must reach your product or billing system, whether larger customers need extra steps, and who should be told when a customer waits.

Frequently asked questions

What steps should a customer onboarding workflow have?
Record the customer, a review of the handover from sales with a way to send it back, a setup confirmed by implementation with a way to say it cannot be delivered, and a status and live flag written on the record.
Why check the handover from sales?
Because most onboarding problems start there: missing contacts, an unclear scope or a wrong start date. Checking before work begins is much cheaper than fixing it afterwards.
What should happen when a deal cannot be delivered as sold?
It should be marked blocked on the record so the problem is visible straight away, and the people who sold it and the people who set it up can resolve it together.
Can AI check a handover from sales?
It can draft the check: a short summary and a list of what the note does not say. In our test run it summarized the note and flagged that contacts were missing. A person still reads it against the note and decides.
Can a workflow update the customer record for me?
Yes. In our test run the workflow created the customer record at handover and set its status and live flag at each ending. It does not set up the account itself unless you connect it to your product.
Share

SynaptaGrid

Build your customer onboarding steps

Start with the steps above in a test workspace and run one customer through each outcome before you use it for real.