Buddee · Product case · Sander Stricker

Onboarding a new HR and payroll customer

Stage 1Goal

1Goal description

New customers operational faster, with less work for them and for us, and working independently afterwards without much explanation.
Original brief · Situatie

Buddee is een HR- en salarisplatform. Klanten nemen onze HR-software, onze salarisadministratie of beide af; bij salarisadministratie verzorgen wij de verloning. In deze case gaat het om een nieuwe klant die beide afneemt.

Een klant kan zelf een account aanmaken, of we zetten dat klaar tijdens het salesproces. Daarna moeten bestaande bedrijfs- en personeelsgegevens worden overgezet en moet de omgeving worden ingericht, voor HR én salaris. Nu krijgt een klant hulp bij de inrichting en een onboardingscall. Daarna moet de klant grotendeels zelf de weg vinden, en krijgen we nog veel vragen. We willen dat nieuwe klanten sneller operationeel zijn, met minder werk voor hen en voor ons, en dat ze daarna zelfstandig met Buddee werken zonder veel uitleg nodig te hebben.

2KPIs

The goal is translated into a few measurable KPIs, each tied to a part of it. Product does not set these alone: they are agreed with the rest of the management team, with a baseline and a target.

Less work for us Buddee hours per onboarding
Today8 h
→
Target4 h
-50%
Time customer success and payroll spend per new customer, from contract to first correct payslip.
Less work for them Customer hours per onboarding
Today31 h
→
Target15 h
-52%
Time the customer spends on data, setup and questions, from contract to first correct payslip.
Independent afterwards Support questions in the first 90 days
Today9
→
Target4
-56%
Questions per customer by email, chat or phone in the first 90 days after go-live.
Stage 2Inputs

3The current flow

Map every step from contract to first correct payslip: who does it, in which tool, the hand-offs and the waiting. Based on the call with customer success, the Valuecase portal and my own signup.

HR onboarding

#StepWhoWhere
1Contract signed, portal sentSalesBuddee, Valuecase
2Company details filled inCustomerValuecase
3Employee template uploadedCustomerValuecase, Excel
4Template checked and importedCustomer successBuddee
5Leave, roles and workflows setCustomer, customer successBuddee, call
6Employees invited (go-live)CustomerBuddee

Payroll onboarding

#StepWhoWhere
1Contract signedSales–
2Payroll data deliveredCustomer–
3Payroll rebuilt in NmbrsPayroll specialistNmbrs
4CAO questions and checks doneCustomer, payroll specialistCall
5First payroll runPayroll teamNmbrs
6First correct payslip (end of onboarding)CustomerBuddee

4Funnel data

Put numbers on the flow from step 3, starting at contract signing: how many customers reach each step, how fast, and at what effort. Not finished within 3 months counts as lost.

Key measurements

8 hBuddee hours per onboarding (4 HR + 4 payroll)
31 hCustomer hours per onboarding
39Median days, contract to first correct payslip
4%Share lost (not finished within 3 months)

HR onboarding

StepReachedMedian daysLost hereBuddee hours
Contract signed, portal sent100%00%0.5 h
Company details filled in99%3 +31%0 h
Employee template uploaded97%16 +132%0.5 h
Template checked and imported97%24 +80%1.5 h
Leave, roles and workflows set96%33 +91%1.5 h
Employees invited (go-live)96%36 +30%0 h
Back in Valuecase after go-live (within 6 months)12%–––

Payroll onboarding

StepReachedMedian daysLost hereBuddee hours
Contract signed100%00%0 h
Payroll data delivered98%14 +142%1 h
Payroll rebuilt in Nmbrs98%22 +80%1.5 h
CAO questions and checks done97%26 +41%0.5 h
First payroll run97%36 +100%0.5 h
First correct payslip (end of onboarding)96%39 +31%0.5 h

5Internal input

Talk to the people who run onboarding every day, customer success and payroll, and read the meeting transcripts and internally reported tickets in Jira. AI clusters the questions and pain points per step.

Key findings

  • The real onboarding happens outside the product: portal, email and people.
  • Biggest hurdles: employee data, then leave balances, then roles, permissions and workflows.
  • Payroll is a separate track with its own specialist and a rebuild in Nmbrs.
  • His one wish: "fewer choices and more explanation in the system".

Examples of input

6Customer interactions

Read the emails, chats and calls with customers during onboarding and their first 90 days. Each contact is tagged with its step, so AI can count topics per step and per week.

Examples of input

7Observations and assumptions

Go and look for myself. I signed up as a new customer and went through the Valuecase onboarding portal, in a test space set up by customer success. What I found are observations and assumptions to test, not a list of fixes.

8Competitor analysis

How the closest competitors in the Dutch market take a new customer from signup to the first payslip, based on their public websites and help pages.

Stage 3Analysis

9Key problems and opportunities

Every input from steps 3 to 8 is distilled into problems and opportunities, each with its sources, and grouped by the KPI it affects. I then use it for follow-up questions, to challenge each finding and to bring up input we missed. This is not a process product does in isolation: it is a discussion and a collaboration with all stakeholders involved. Nor is it a manual task: AI distils the essential information from all of it.

Buddee hours per onboarding 8 h → 4 h

Customer hours per onboarding 31 h → 15 h

Support questions in the first 90 days 9 → 4

10Focus

From eight findings, we pick three to start with, compared on three criteria.

1Impact on a KPIHow much it moves one of the three targets.
2Customers affectedHow many new customers run into it.
3FeasibilityHow realistic it is with the team, tools and partners we have.

Focus issues

Stage 4Ideation

11Solutions

For each focus issue, a few possible solutions, from small changes to bigger ideas. Quick to sketch, cheap to throw away.

Note. In practice, many small improvements add up to a big impact, and they deserve to be found and shipped. The solutions below are deliberately bigger: they suit a case discussion better, and the small details need deep product knowledge that I do not have yet.

Data import is labour-intensive

Onboarding is overwhelming and split across tools

Users still have questions after go-live

12Solution validation

Before building, test whether a solution solves the problem. Prefer real customers and real data; mockups and pitches are cheaper but weaker evidence.

A simple version, as an A/B testStrongest evidenceShip a minimal version to part of the new customers and compare with the rest on days and drop-off. Real customers, real data.
Fake door testStrong evidenceShow the solution in the product, for example a button or a banner, without building it fully. Measure how many customers click, then explain and follow up.
Mockup shared with customersMedium evidenceA clickable mockup for five to eight recent customers. Watch where they hesitate and what they expect next.
Pitch to a customerLight evidenceDescribe the solution and ask for a reaction. Quick and cheap, but what people say is weaker than what they do.
Stage 5Rollout

13Prioritization

Here we decide which solutions to build first, scoring each on effort and expected impact. In practice the scores are not hard facts, as neither can be measured exactly in advance. They are a sanity check: what makes sense, where to invest, and what feels right for everyone involved.

SolutionEffortExpected impactScore
Data import is labour-intensive
1AI-guided data setupHighVery high13
2Start from your last payslipsMediumHigh13
3AI scan of your current systemHighHigh10
Onboarding is overwhelming and split across tools
4Personal to-do list inside BuddeeMediumVery high17
5Proactive AI onboarding assistantHighVery high13
Users still have questions after go-live
6Onboarding content stays available in BuddeeLowHigh20
7Ask Buddee about your companyHighHigh10

First 3 months

  1. 6Onboarding content stays available in Buddee. Low effort, high impact: the quickest win.
  2. 2A proof of concept for starting from the last payslips. Mainly to test how viable this use case is: how much can we reliably read from real payslips?
  3. 4Personal to-do list inside Buddee. Something we can build up step by step, one onboarding step at a time.

What we deliberately do not do

  • No fully self-service payroll setup from day one: payroll specialists stay involved for CAO questions and the first run.
  • No custom integration per old system up front: start with the AI scan and the payslips, and build integrations only where volume justifies it.
  • No new, separate onboarding module: reuse existing Buddee screens and bring in the content that already exists in Valuecase.
  • No big-bang switch away from Valuecase: content moves into Buddee step by step, using the Valuecase API where possible, and the portal stays as long as it is needed.

14Build

1Spec / ticket
2Design
3Build MVP
4Improve

15Release

  • Release gradually: first a group of new customers, then everyone.
  • Tell customers and the team what changed, customer success and payroll first.
  • Update the help content in Buddee at the same time.
  • Follow adoption, so the new way of working actually lands.
Stage 6Outcome

16Results analysis

  • Compare the KPIs with the baseline: Buddee hours, customer hours and support questions.
  • Set up dynamic dashboards, for example with Claude's dashboard feature, so the numbers stay current without manual work.
  • Use AI to find correlations in behaviour we were not tracking but that turn out to matter.
  • Watch for side effects: a change can perform well and still raise more support questions, or create work elsewhere in the company or the funnel.
  • Use decision points set in advance: scale up, adjust or stop.

17Iteration

  • Feed what we learn back into the solutions.
  • Pick the next problem from the list.
  • Run the cycle again against a new baseline.