Switching costs are manufactured
An agency owner calls us. She's been on the same platform for eleven years. She wants to switch. But she has 30,000 bookings, a decade of commission history, payout records for forty advisors, supplier relationships she's built one by one. All of it lives in a system that was never designed to let it leave.
That's the real switching cost in travel tech. It's not the new subscription, it's the terror of losing the work that took years to build. It's also not an accident. Legacy platforms benefit from that fear, and the harder it is to leave, the less the product has to improve.
In most industries, moving your data means exporting a file and importing it somewhere else. In travel, data lives in proprietary formats, undocumented schemas, and desktop-era systems that predate the cloud. There's no export button. There's no standard format. There's just years of business-critical data locked behind walls that someone else built.
We believe your data is yours. Choosing a platform should be about which product is better for your business, not how much pain you're willing to endure to leave the one you're on.
The race against the clock
Switching an agency to Tern involves three parallel workstreams before go-live: account setup, training, and data migration.
Brooklyn leads the enablement motion: configuring accounts, training agency owners, and building playbooks tailored for every agency type and size. The goal is to get agency owners and their team comfortable before switching to a new system essentially “overnight.” AI is compressing this work too: onboarding plans generated from discovery calls, auto-documented timelines, training content produced in a fraction of the time. Roughly 50% less back-and-forth coordination, which means agencies ramp faster and our team can run more launches in parallel.
Data migration is my focus, and it operates under a constraint the other workstreams don't share. The moment we extract data from a legacy system, the clock starts. Advisors are still booking trips in the old platform, and every day between extraction and go-live is a day where the two systems diverge. Data drifts, records fall out of sync, and the migration has to be fast enough that drift never becomes the problem.
Six months ago, we were entering bookings manually. About five minutes per booking, and even then we couldn't bring over client payments or commission data. Unscalable work that still left agencies with critical gaps in their financial history. That had to change.
Getting the data out
Legacy travel systems were not built with portability in mind. Some offer limited exports, but most don't. Data formats are inconsistent, fields are missing or mislabeled, and the same concept can be stored completely differently across two versions of the same platform. These systems don't compete on product, they compete on inertia.
We treat extraction as an engineering problem. For every source system we support, we've built reliable methods to get the data out completely and accurately. Most importantly, we do it on our schedule.
Understanding the data
Every system stores data differently. Some platforms have hundreds of tables with complex relationships between them, where a single booking might span three tables in one system and twelve in another. Commission structures sit in places you wouldn't expect, payout logic is buried in fields that aren't labeled, and none of it is documented.
Before AI, understanding a new system's data model was weeks of manual exploration. Now I generate hundreds of scripts to test hypotheses about how the data connects, validate assumptions against real records, and map relationships across hundreds of tables in an afternoon.
Some agencies have data spread across two systems. Maybe they used one platform for bookings and another for commissions. Maybe they switched partway through the year and have overlapping records in both. Now you're mapping two data models, figuring out where they agree, where they conflict, and which one is the source of truth. AI lets me rapidly cross-reference records across both systems, test matching logic, and surface discrepancies that would take days to find manually.
Thanks to these rapid iterations, we're able to learn and support new systems faster than it used to take us to export the data. Agencies that were locked into their previous systems now have optionality they didn’t have before.
Transforming and ingesting the data
Every source system needs its own transformation pipeline. When an agency spans two systems, the pipeline has to reconcile and merge before it can standardize. I bring the judgment calls about what the data should look like. AI brings rapid experimentation of the business logic, iterating on transformation rules in minutes instead of days.
A complete migration now covers the full back office: trips, bookings with supplier details, confirmation numbers, client payments, travelers, and costs. Suppliers. Commissions received. Commission payouts. The entire financial history an agency needs to operate without looking back. Import tooling runs with fallback logic throughout, auto-detection of anomalies, and validation checks at every step.
Thanks to these transformation pipelines, we've standardized outcomes across all systems, guaranteeing that what we import will yield what we expect - a complete back-office setup for business as usual from day one.
Why this matters for your business
Here's what a slow or incomplete migration actually costs an agency. If your bookings come over without commission history, you can't reconcile new commissions against what's already been received. If client payment records are missing, you don't know who's paid and who hasn't. If payout history doesn't transfer, you can't accurately pay your agents. These aren't data points on a spreadsheet, they're the operational backbone of your business. Without them, you're not just learning a new system, you're flying blind while you do it.
That's why speed matters too. The longer a migration takes, the more your advisors are booking in one system while your data lives in another. Commissions come in that don't match, payments get tracked in two places, and the drift creates real problems that someone has to untangle manually. We plan every migration around payroll cycles so your data is as fresh as possible and you have maximum time to get comfortable in Tern before your next payout run.
Six months ago, that's exactly where we were. Migrations took weeks, came over incomplete, and left agencies with gaps in their financial data that made the transition painful. Today, agencies with over 50,000 bookings go live in a single day with their complete financial history. Trips, bookings, client payments, commissions received, commission payouts, all of it. At five minutes per booking manually, that's over 4,000 hours of work compressed into nine, a 99% reduction. We went from migrating a handful of agencies per quarter to over a dozen, and we have plans to scale that significantly.
Let’s be honest, no one ever said, “I can’t wait for implementation day!” The faster we can migrate your data, the faster you can get to the fun part of your job - selling magical adventures.
Your advisors aren't stuck reconciling gaps or rebuilding records, they're selling travel on day one.
Your content library comes too
Switching costs aren't just about back-office data. Every advisor has a content library they've built up over years: email, form, activity, and itinerary templates. Preferred suppliers and media content. We used to have to prioritize back-office data over library content, which meant advisors had to rebuild their libraries from scratch while also learning a new system. Another manufactured barrier.
So we built a Chrome extension that lets advisors import content from any webpage directly into Tern. AI parses and structures the page automatically. Thanks to our accelerated engineering practices, I was able to build two new AI features specifically for the extension - email and form template imports - and I did it in two days. Your library comes with you.
We've seen rapid adoption of the extension and are continuing to iterate, with the ability to import activities and itineraries directly into trips now live.
AI doesn’t replace expertise. It makes it scale
I'm not a travel industry veteran. I've been in this space for a couple of years. But I've spent that time in the weeds, day and night, talking to every agency owner who will give me their time, watching them navigate their current systems, working through their data together. That's how you build the judgment to know what the lifecycle of a booking looks like end to end, from planning to commission payout, which edge cases matter, and what "complete" actually means for a given agency.
AI can't give me that judgment. Without it, AI can actually amplify bad or incorrect assumptions. But once I have it, AI lets me act on it at a speed that wasn't possible before. I can test a hypothesis against 50,000 records in seconds, and we can iterate on transformation rules in minutes instead of days. That combination of domain knowledge and AI speed is what turned migrations from a months-long bottleneck into something we can do in a day.
What's still hard
Despite our rapid improvements, there are still challenges ahead.
Every system is different, and every agency and advisor uses their system differently. That creates an explosion of edge cases, and no amount of tooling eliminates all of them.
Some data might not come over perfectly. We tell you that upfront, before the migration starts, because we'd rather be transparent about a tradeoff than have you discover it after go-live. We're fortunate to work with clients who treat this as the partnership it is. They know what their data represents and they're grateful for the effort we put into respecting it.
Time has been the biggest constraint, and our goal is to offer this level of service to all our agency partners, no matter their size or legacy system. To accomplish this, we're constantly learning, adapting, and closing gaps in hours that used to take weeks. We’re training migration-certified Lucia Copilots that can support end-to-end migrations, and building toward a world where agencies can review and clean their own data in-app before go-live. We’ll have more exciting news on this soon.
Even with a perfect migration, getting your data into Tern is only half the story. The other half is human: getting every advisor bought in, changing workflows that have been muscle memory for years, building confidence in a new system while still serving clients. That work doesn't compress the same way a data pipeline does, and it deserves its own deep dive.
Up next in the series: Abby, our Head of Customer Experience, on rethinking enablement, support, and customer success across the end-to-end user journey - how we're making the human side of switching as smooth and delightful as possible.
The walls are coming down. If you've been staying on a legacy platform because you're afraid of losing your data, that's not the tradeoff it used to be. Once switching costs disappear, the only thing that matters is which product is actually better for your business.