What a Good Onboarding Call With a New Data Vendor Should Cover | NPLUS Global

Most onboarding calls with data vendors are wasted. Someone from the vendor's customer success team walks you through a slide deck you've already seen, asks "any questions?", and thirty minutes later you're both back to your inboxes with a vague sense that things are "set up." Then six weeks in, you find out the file doesn't map to your CRM the way you assumed, nobody explained how NPI matching actually works, and the person who ran the call has already moved to a different account.

This is a working meeting, not a courtesy call. Treat it like a technical handoff, because that's what it is.

Before you start: Come to the call with your own data dictionary or field mapping, even a rough one. If you don't know which fields in your CRM or MAP the vendor's file needs to populate, you can't evaluate anything they tell you — you'll just be nodding along. Also decide internally, before the call, who owns this vendor relationship long-term. If that's not you, get that person on the call now, not after the first bad file.

1. Get the data dictionary reviewed line by line — not walked through, reviewed

Ask for the field-level data dictionary at least a day before the call, and actually read it. On the call, don't let them narrate it top to bottom. Instead, go straight to the fields that matter most for your use case — specialty codes, license status, affiliation type, whatever drives your segmentation — and ask how each one is derived. "Specialty" can mean self-reported, claims-derived, or board-certification-derived, and those produce meaningfully different lists. If the answer to "where does this field come from" is vague, that's information too.

2. Ask about sourcing and refresh cadence, field by field, not as a blanket claim

"Updated monthly" is a marketing sentence, not an operational fact. Different fields refresh on different schedules almost everywhere — a provider's practice address might update quarterly while license status updates in near real-time from state boards. Ask which fields lag and by how much. If a vendor can't break refresh cadence down by field type, they're either bundling everything under one number for simplicity, or they genuinely don't track it that granularly — and you should know which.

3. Walk through match and dedup logic against your own systems, not theirs

This is the step teams skip most often, and it's the one that causes the most downstream mess. Ask specifically: what key do they match on (NPI, name plus address, internal ID), what happens when there's a partial match, and how does their file handle records that already exist in your CRM under a different source? If you're layering this vendor's data on top of an existing database — which is almost always the case — get a real answer on how conflicts get resolved, not "our matching is very accurate."

4. Define the exception process before you need it

Every data file has bad records — wrong emails, stale addresses, duplicate NPIs. The question isn't whether errors will happen, it's what happens when they do. Ask directly: is there a replacement or credit policy for records that bounce or are demonstrably wrong? What's the process for flagging a bad batch — a support ticket, a shared spreadsheet, a dedicated Slack channel? How long does resolution typically take? Get this in writing, or at minimum in a follow-up email, so it's not a memory six months from now when you're arguing about a file that underperformed.

5. Pin down usage rights and compliance boundaries in plain language

This matters more in healthcare data than almost anywhere else. Ask what you're actually licensed to do — can this data be used for outbound email, direct mail, paid media matching, resale to a partner agency, enrichment of a third-party platform? Ask what compliance obligations pass through to you versus stay with them (this comes up constantly around provider consent, opt-outs, and state-level marketing restrictions). Don't assume "B2B data" means unrestricted use; healthcare-adjacent data often carries more contractual nuance than general business contact data, and it's cheaper to clarify this now than to find out later from your legal team.

6. Set an early validation checkpoint — not a 90-day one

A lot of vendor contracts default to a quarterly business review as the first real check-in. That's too late. Ask for a working session at two to three weeks in, specifically to review the first live file against your own data — match rates, field completeness, obvious anomalies. This is also the point where a good vendor will proactively flag if something in your file usage doesn't match what they expected from the sales conversation. At NPLUS Global, we push for this early checkpoint deliberately, because most real problems surface in the first file, not the fiftieth.

7. Confirm the human infrastructure, not just the technical one

Get a named point of contact for day-to-day questions and a separate, named escalation contact for when something breaks and the first person is unresponsive or out. Ask what their actual response time commitment is — not "we're very responsive," but a number, ideally in writing. Vendor relationships degrade quietly when the original salesperson disappears and nobody replaces them with a clear owner. Ask who that person will be before you're relying on them in a crisis.

What to watch out for

Be wary of a vendor who wants to schedule the onboarding call before the contract is fully signed and treats it as a formality to keep momentum going — that's a sign the relationship is being managed for the close, not the delivery. Be equally wary of answers that stay at the marketing level no matter how specific your question gets ("very accurate," "highly responsive," "industry-leading") — that's usually a sign the person on the call doesn't have operational visibility into their own data pipeline. And if nobody on the call can tell you, concretely, what happens when a file has a problem, assume that problem will eventually be yours to solve alone.