Scoping a first gateway pilot: one sender, one line of business

Scoping a first gateway pilot: one sender, one line of business

Craig Hangartner

Saba Gobal, CPCU

A carrier that has decided to test a data exchange gateway still faces one scoping question before anyone writes a ticket: how much of the business goes into the first pilot. Take too much, and the pilot becomes a project. Take too little, and it proves nothing.

This guide is for the Guidewire architect or technology lead who has to brief a team. It covers a pilot of one sender and one line of business: how to pick both, what goes into the scope, how the pilot runs, and how to call it passed.

It concerns a carrier-side data exchange gateway, not Guidewire’s own Integration Gateway, which is a developer framework.

Why limit a first gateway pilot to one sender and one line of business?

A pilot of one sender and one line of business gives you a single, complete path from sample payload to first accepted submission. Every failure then traces to one cause, and your team can name it.

That clarity is the point of a first pilot. The pilot tests a route, not a rollout, and a narrow scope keeps it that way.

What a first pilot has to prove

A first pilot answers four questions, and each needs a yes before a second sender is invited.

Four-step infographic on a first gateway pilot: map to Guidewire, confirm mapping, return bad data, contain changes.
  1. Can the sender’s payload be mapped to your Guidewire structures?

  2. Can a named person on your team review and confirm that mapping?

  3. Does incomplete data go back to the sender instead of into your queue?

  4. Does a changed payload reach one place, not every sender?

Why one sender

Two senders double the variables. Each brings its own field names, its own value labels, and its own admin approving the mapping.

When a test fails with two senders in play, you cannot tell whether the cause sits in a payload, a mapping, or one of your own rules. One sender removes that guesswork.

Why one line of business

Mandatory fields and valid values differ from one line to the next. One line gives your team one rule set to write down and one set of test cases to run.

A second line adds a second rule set before the first has been proven. Save it for the follow-on.

Which sender and which line of business should you pick first?

Pick a sender that has a sample payload ready and an admin who can approve a mapping, and pick a line your team can describe on one page. Readiness matters more than volume.

Choosing the sender

The best first sender is the one that leaves your team the fewest excuses. Look for these four traits:

Four-step infographic on picking a first pilot sender: sample payload, admin approval, sandbox testing, matching line.
  • A sample JSON payload ready to share

  • An admin who can review and approve the mapping

  • A willingness to test in a sandbox before sending real data

  • A line of business that matches the one you chose

An underwriting workbench vendor whose product already produces structured JSON fits this profile.

Choosing the line of business

Choose a line whose mandatory fields and valid values your underwriters or claims team can state on one page. If the rule sheet runs longer, the line is too wide for a first pilot.

Avoid a line that is mid-change. A moving rule set makes it impossible to tell a mapping error from a rule that shifted.

Policy submissions or claims first notice of loss

A first pilot targets one Guidewire system, not several. The choice is between policy submissions and claims.

A submission pilot sends submissions into Guidewire PolicyCenter. The sender registers, maps its payload, and tests clearance, account creation, and submission creation before any live send.

A claims pilot sends first notice of loss into Guidewire ClaimCenter. Creating a claim takes more than one call, such as creating a draft and then submitting it, and ClaimCenter’s own rules decide whether each submission becomes a claim.

Those rules include a policy being in force on the loss date and a required validation level. For a first pilot, pick one route, because claims intake carries rules that a submission pilot never touches.

What goes into a pilot scope document?

A pilot scope fixes six things in writing: the sender, the line of business, the sample payloads, the mapping reviewer, the test cases, and the exit criteria.

One page is enough. Each item needs evidence that someone collects, so “done” is never a matter of opinion.

The One-Sender Pilot Scope

The table below turns the six items into a checklist your Guidewire team can sign before the pilot starts.

Scope item

What you fix

Evidence you collect

Sender

One named sender and its admin

Admin registered in the self-service portal

Line of business

One line, with its mandatory fields and valid values

A one-page rule sheet

Sample payloads

One complete JSON sample and one with deliberate gaps

The gap sample returned to the sender

Mapping review

A named person who confirms the mapping

Mapping signed off before any test send

Test cases

Clearance, account creation and submission creation, or the claim draft and submit

Results verified in the sandbox

Exit criteria

What counts as a pass, written before the pilot starts

A dashboard record of what was sent, arrived, and rejected

The sample payload

The sample payload is where most pilots start, and where scoping errors surface first.

Ask the sender for two samples: one complete JSON payload and one with deliberate gaps. The complete sample proves the route. The gap sample proves that incomplete data goes back to the sender.

For a submission pilot, the complete sample carries the account and the submission in one payload. The test can then run clearance, account creation, and submission creation from a single file.

Mandatory fields and valid values

The rule sheet is the carrier’s half of the scope. It tells the sender what a complete payload looks like for your line.

List the line’s mandatory fields and valid values in a payload specification the sender can read before it maps anything. Source values then convert into the codes your Guidewire system expects, so a partner’s label arrives as the correct code.

Name any carrier-specific fields in the same sheet. A gateway that supports tenant-specific mapping holds those fields in the mapping itself, so the pilot needs no separate build.

How does the pilot run from sign-up to the first live submission?

The pilot runs in four steps: the sender registers, maps its sample payload, tests in a sandbox, and sends its first live submission after your team reviews the results.

Register the sender

The carrier invites the sender by email. The sender registers through a self-service portal with a one-time passcode, receives an API username and key, and selects the line of business it will send.

No ticket to your integration team is needed at this step. That is the point of the submission intake route: partners register and test on their own.

Map the sample and review the mapping

The sender uploads its sample JSON payload. A syntax check runs before mapping begins, and insurance-trained AI suggests how each source field maps to your Guidewire structures, flagging any field it could not match.

A person then reviews and confirms the mapping before any test call. The scope names that person in advance.

The pilot also answers a design question: whether the mapping sits with the sender or with the gateway. Vendor-side or gateway-side mapping is easier to judge once you have watched one sender complete it.

Test in the sandbox

The sender runs the test cases in a sandbox before any real data moves. For submissions, that means clearance, account creation, and submission creation. For claims, it means the full route, including the draft and the submit.

Your team checks each result against the rule sheet. Run the gap sample here as well, because a payload with gaps goes back to the sender to complete rather than entering your queue.

Send the first live submission

After your team signs off the test results, the sender sends its first live submission. A dashboard shows what was sent, what arrived, and what was rejected.

That record becomes the evidence for the exit criteria.

How do you decide the pilot passed?

A pilot passes when the complete sample reaches your Guidewire system in the expected format, the gap sample is returned, and one deliberate change is absorbed without a rebuild.

Write these criteria before the pilot starts. Criteria written after the first test bend toward whatever the test produced.

Set the pass criteria

Four checks decide the outcome, and each is a yes or a no:

  1. The complete sample arrives in the expected format and matches the rule sheet.

  2. The gap sample is returned to the sender.

  3. Every flagged field is resolved by the named reviewer.

  4. The dashboard shows a record of each request.

Test one deliberate change

A pilot with no change in it leaves the hardest question open: what happens when something moves.

Add one deliberate change, such as a new mandatory field on the line. A gateway built for this raises an alert for the new field, and a default value keeps submissions flowing while the mapping is updated in one place.

That matters because a change to a payload or a Guidewire release otherwise reaches every sender’s connector.

Carry the scope to the next sender

If all four checks hold, the scope document, the rule sheet, and the test results become the template for the next sender on the same line.

Add a second line only after the first sender is live. The same scope shape applies, with a new rule sheet.

How InsOps Helps

Yazh (Yaal), the InsOps Integration Gateway, is a data exchange platform for carriers on Guidewire. A sender maps its payload once, and Yazh translates it to each carrier’s Guidewire format. LiLa, our insurance-trained model, assists the mapping, and a person reviews every mapping before it is used.

Yazh brings inbound data into Guidewire PolicyCenter, ClaimCenter and BillingCenter. A self-service portal, a test harness and a sandbox let a pilot sender register, map and test on its own, and incomplete payloads go back to the sender.

Guidewire’s own Integration Gateway is a developer framework. Yazh adds a place beside it where outside senders map, test and send without a separate build for each partner.

If you are evaluating how to scope a first pilot without committing your Guidewire team to a custom connector, contact us to talk through what a one-sender, one-line pilot looks like for your operation.

Frequently Asked Questions

What is a gateway pilot?

A gateway pilot is a limited test of a carrier-side data exchange gateway with one sender and one line of business. It runs from sender registration to a first live submission and ends with a pass or fail against exit criteria set in advance.

Why start with one sender and one line of business?

One sender and one line give every failure a single traceable cause. A wider scope turns a test of the route into a project with many unknowns.

How do you choose the first sender?

Choose a sender with a sample JSON payload ready, an admin who can approve a mapping, and a line that matches your pilot. Readiness matters more than volume.

What should the sample JSON payload include?

Provide two samples: one complete payload for the chosen line and one with deliberate gaps. The complete sample proves the route, and the gap sample proves that incomplete data goes back to the sender.

Who reviews the field mapping?

A named person on the team reviews and confirms the mapping before any test call. LiLa suggests the mapping and flags fields it could not match, and a person reviews every mapping before it is used.

What should you measure in a pilot?

Track whether the complete sample arrived in the expected format, whether the gap sample was returned, whether every flagged field was resolved, and whether the dashboard shows each request. These are pass or fail checks, not rates.

How long does a gateway pilot take?

The length depends on how complete the sender’s sample is and how fast your named reviewer confirms the mapping. Fix the exit criteria first, and the pilot ends when they are met.

QWhat are the common pilot pitfalls?

Adding a second sender, using a sample with no gaps, leaving the mapping reviewer unnamed, and writing exit criteria after the test starts. Each one blurs the result.

Craig Hangartner

Saba Gobal, CPCU