🌸 Jenny Ho 🎨 Design work 👩🏻‍💻 More about me 🌞 Resume 🌊 Say hi 🌸 Jenny Ho 🎨 Design work 👩🏻‍💻 More about me 🌞 Resume 🌊 Say hi 🌸 Jenny Ho 🎨 Design work 👩🏻‍💻 More about me 🌞 Resume 🌊 Say hi 

Setting up & scaling integrations

Research, product design, data visualization, UX writing



The challenge

Archipelago is a platform that centralizes data for insuring trillions of dollars of property. While we have construction and occupancy data covered, there's still a lot we don’t have.

Climate change predictions, natural disaster tracking, and tailored engineering recommendations – urgent data for property owners, insurers, and brokers that we don’t provide.

The solution

What if we built integrations with other trusted software companies in the industry? We can give users a more complete view of their portfolio by adding new building resiliency data.

My role

I conducted research with product and business development stakeholders, led the design (including supervising a mid-level designer), held design reviews with PMs, and did QA with engineers.

Impact

We ended up piloting 2 integrations with 35 customers. Our users saved anywhere from a few hours to a few days preparing a portfolio for the insurance market.

The pilots set us up for partnerships with Origami, Unqork, Verisk, CoreLogic, and PwC later on.

What do users want?

To find out which integrations to build first, we needed to know what kind of data would be most helpful. Here, I worked with the product strategist, product manager, sales director, and head of account management to sell the concept of integrations to customers.

We presented slides pitching hypothetical new features on customer calls, and we sent out follow-up surveys. I contributed a few ideas for these potential features and created mockups so customers see their value (as opposed to forgettable blocks of text).

Hypothetical features that users would be excited about. Manage your integrations. Compare building valuations with market rates. Predict climate change risk. Track climate and weather perils. Auto-populate missing building data. See sources and timestamps on an enhanced portfolio. Building valuations (dark green) was the top pick from the survey.

Future-proof the MVP

My design process was to start more complicated, scrap what we didn’t need, and re-introduce scrapped features as needed. The PM and engineers appreciated this method, since it’d future-proof the feature while reducing the technical complexity.

✂️ What got cut
  • Support multiple integrations. We might have a scenario where a partner company has multiple integrations.
  • Self-service functionality so users can run an integration whenever they want.
  • Provide advanced customization settings for users to choose the exact properties and data they want.

👌 What we can do
  • Build for one integration at a time.
  • A white glove approach where Archipelago admins set everything up. Customers don’t see these screens.
  • Limit customization options.

Determining scope and user flows. Mapping how new data types fit into Archipelago's data models. Wireframes of how managing integrations happens. Wireframes of how the valuation integration would work.

Reviews are in!

Our first integration imports building valuation data into Archipelago. Users can compare their own values to market standards, so they are better prepared for price negotiations.

We showed 4 brokers a prototype before building, and they were thrilled! It’d save them 6-8 hours preparing an account. After launch, we saw similar productivity gains among users.

1. The account manager enables integrations for a customer. 2. Go to the portfolio. 3. Run the integration.

4. When it's done, see the results. 5. Check how your portfolio's value compares to the market. 6. Review specific properties for red flags.



Adding the next integration

Months later, we had a second partnership: we can fill in missing COPE building data as long as there’s a valid address.

Our most frequent use cases were owners acquiring new properties. They’d typically fill 50% of blank data per property with a few clicks. This saves them 2-3 days compared to manually researching and inputting data for each building.


1. As an MVP, all integrations have the same usage limit. 2. Choose which integration to run. 3. Run it.

4. See results when it's done. 5. QA new data before publishing. 6. See updated property details.

Explaining API limits to users

One ongoing challenge was describing API limits to users, since integration partners have their own usage rules.

With help from a technical writer, I made API limits easier to understand by:
  • Using terminology that fits users’ mental models.
  • Editing UI copy to focus more on impact and less on transactions.
  • Adding a history page to track usage.

Estimates before running an integration. The results afterwards. View usage history.



Post-script

We were set to add natural disaster tracking by summer 2023, so customers can get ahead of hurricane season. However, the project was cut short by layoffs.

As of August 2023, Archipelago had 10 more partners and 3 new data types in the pipeline: natural disaster tracking, climate change, and fire protection & manufacturing recommendations. 

Eventually, we added intergrations with Origami, Unqork, Verisk, CoreLogic, and PwC.

🗂 All projects