How onboarding works, and what to expect

Different systems can have different credential owners, and that's fine. But for the day-to-day of onboarding, working with our team and moving the project forward, you need one solid leader. We've seen onboardings stall when three or four people rotate in from the customer side, each with a different answer to the same question. We need a senior leader, someone available to engage throughout the process and empowered to make critical business decisions, who owns communication with us from kickoff through go-live. They can pull in IT, finance, or credential owners as needed, but the decisions and the answers come through them.

Timeline impact: every extra voice in the room adds coordination overhead, not clarity.

• Name one senior leader before kickoff, available to engage throughout and empowered to make business decisions. Everyone else routes through them.
• Credential ownership can stay distributed across systems, that's normal. Project leadership shouldn't be.
• If your leader changes mid-onboarding, tell us immediately and get the new person caught up before the next session.

CorralData connects to 600+ sources, but not every system has a usable API. Some platforms lock data behind their own dashboards. Others require a formal vendor request to enable external access. This is a conversation we'd rather have in week one than week four.

Timeline impact: varies, days to indefinite depending on the system

• API we haven't built for yet: custom connector build, typically 2 to 4 weeks
• Limited or undocumented API: we assess, give you a realistic range, and build if feasible
• No API at all: we explore alternatives, scheduled exports, manual uploads, or middleware, and tell you the tradeoffs honestly
• Vendor approval required: we initiate the request, but the vendor controls the timeline. Submit immediately, don't wait for kickoff.

Credential delays are the most common early bottleneck. Access to a system often lives with one person, and when that person is out, onboarding pauses. Every day of delay here pushes your go-live date.

Timeline impact: 1 day of delay equals 1 day pushed. It compounds.

• Map out who holds credentials for each contracted system before kickoff
• If someone with access has upcoming travel or is hard to reach, flag it to your TAM immediately
• If IT approval is required, initiate the request before the kickoff call, not during it
• Your onboarding lead needs a direct path to every credential holder

Finance says revenue is net collections. Operations says it's scheduled services. Marketing says it's what got booked. All three are defensible. None of them are the same number. We can't build until there's one definition, and we can't make that call for you.

Timeline impact: unresolved definitions means nothing gets built

• Before the discovery call, review the prep questions we send and identify where your team doesn't have a shared definition
• Get the right stakeholders in the session to reach a decision. Hold a pre-meeting to align first if needed.
• Your onboarding lead needs the authority to lock a definition when the team disagrees, or a fast path to whoever does

A definition that changes after a model is built means the model has to be rebuilt. Definition changes cost time proportional to how far downstream the change is. A change before we build is fast. A change after dashboards are live can be time intensive.

Timeline impact: depends on timing, earlier changes cost less

• Treat the discovery sign-off as a real decision point, not a draft
• If a definition needs to change, tell us immediately. The sooner we know, the less rework.
• Limit who can trigger a definition change. If anyone can casually request a recalculation, expect a slow onboarding.

Replicating another tool's logic is doable, but only if we understand the logic, not just the output. A screenshot tells us the target number but not which records were included, how edge cases were handled, or what filters were applied. We can match your numbers. We need to see how they were built.

Timeline impact: fast with documentation, slower without it

• Pull any underlying SQL, formula docs, or calculation notes and share them in week one
• If the person who built the original reports is still available, a 30-minute walkthrough is worth more than any screenshots
• If documentation doesn't exist, tell us early. We'll reverse-engineer what we can and be explicit about where we're inferring.

Some metrics require connecting records across two systems that don't share a common ID. Building that bridge requires someone on your team to create a mapping file. This is your work, not ours, and specific metrics are gated on it.

Timeline impact: affected metrics blocked until mapping is delivered

• We'll flag the affected metrics in Phase 2 and provide a mapping template. Your named owner fills it in from your systems.
• We build everything else in the meantime and hold the affected metrics. We won't present placeholders as real numbers.

We deliver v1 and need your confirmation before moving forward. If feedback doesn't come, or comes as "something looks wrong," we can't make progress. Silence isn't approval. And "looks off" isn't a bug report.

Timeline impact: every day without feedback is a day of delay

• When we deliver v1, set aside time to actually look at it, ideally within 2 business days
• Useful feedback: "April revenue shows $142K but our billing system shows $138K for the same period"
• If you don't have a benchmark to compare against, say that. We'll help you reason through it.
• If a number looks right, say so. Confirmation is as useful as a correction.
• Feedback isn't a ticket you file and wait on. We sit down with you and implement it together, so you see exactly why a number changed.

Training delivered to the wrong audience means the people who need to use CorralData day-to-day never learn it. They either never adopt it, or they come back with questions weeks later that require re-running sessions. This is the most preventable outcome in onboarding.

Timeline impact: no immediate delay, but creates support burden post-onboarding

• Before Phase 4, identify who will actually use CorralData: operations leads, clinic managers, the person who currently runs reports
• Get them in the live session, not a recording they'll never watch
• Have them bring real questions. Best training is live, on actual data.

AskCorral's accuracy is a direct reflection of how well the underlying data is defined and governed. Vague definitions from Phase 2 or unresolved issues from Phase 3 surface here as inconsistent answers. This is rarely an AI problem. It's almost always a definition or data quality issue upstream.

Timeline impact: fixable, requires tracing the issue back upstream

• Flag unexpected answers specifically: "I asked X and got Y, I'd expect Z"
• We'll trace the answer back through the definition and data model to find where it diverged
• Best prevention: clean definitions and resolved validation issues in Phases 2 and 3