Anupam Kalita

Why complex payroll customers take months to activate when simple ones take weeks

My POV of how Deel can close the gap between how fast simple and complex customers activate their payroll.

DeelPayroll ActivationComplianceAutomation
Global payroll platforms

Problem statement

  • Simple customers self-serve their payroll activation in weeks. Complex, multi-entity customers wait on a person working a compliance playbook by hand, inconsistently, depending on who's assigned.
  • 57% of global payroll professionals rank local compliance as their single biggest challenge (PayrollOrg, 2025), and that's exactly where activation stalls: one country and one entity at a time.
  • Built from public reviews and commentary on how global payroll platforms handle onboarding, not internal data, but a pattern worth acting on.

Who feels it

Priya, a VP of People Ops at a 600-person company formed through three acquisitions, now runs payroll across five entities. She isn't a payroll or compliance expert, was told her multi-entity setup needs a manual, phased rollout, and has heard from peers that the experience "depends on who you get."

Where it breaks down

  • Signs up, sold on a fast, self-serve activation timeline.
  • Flagged as complex, since a multi-entity structure from acquisitions triggers manual review.
  • Gets a manual, uneven process, where the playbook is followed by hand and quality depends on who's assigned.
  • Activation is delayed, well past the timeline simple customers get.

The proposal

  • Encode compliance rules per country and entity type, so the system handles the deterministic parts with the same outcome regardless of which manager is assigned.
  • Route only flagged exceptions to the implementation manager, so the customer sees consistent, real progress instead of a black box.
  • Rules first, AI next: a rule can be audited and a model's compliance judgment can't, yet, so AI layers in only once the rules have earned a track record.

Why this, not something simpler

  • Adding more implementation managers doesn't fix inconsistency, and the cost scales linearly.
  • One self-serve flow for everyone breaks down exactly where multi-entity complexity begins.
  • Fully auto-approving every setup is fast, but a wrong compliance call isn't just a UX issue.
  • Chosen instead: turn the static compliance playbook into rules, automating what's deterministic and routing judgment calls to the implementation manager.

Shipping it

  • V1, rules: encode the highest-hour country and entity-type combinations first, run deterministic checks with the same outcome regardless of manager, and have the manager review only what's flagged.
  • V2, AI layer: add AI extraction and classification on top of the proven rules, expand across every country and entity type, and improve accuracy from real setup data.

Business impact

  • Faster activation means faster time-to-revenue on the hardest-to-serve accounts.
  • A consistent experience removes the "depends who you get" risk seen in real reviews, a trust issue, not just a speed one.
  • Lower cost to serve, since implementation-manager time is the constrained resource and standardizing it protects margin at scale.

What I'd watch

Time-to-activation for the complex segment measured against the timeline simple customers already get, the % of playbook steps auto-completed, the spread in activation time by manager as a consistency proxy, and complex-customer activation NPS to see whether the simple-vs-complex gap is closing.