Ten Weeks to Election Day: Building KQED's Election Hub

Background

KQED is Northern California's leading public media organization: an NPR and PBS member station with a daily newsroom, podcasts, and a growing digital audience. Uptech Studio has been KQED's mobile development partner since 2021. We rebuilt their iOS and Android apps in Flutter, added CarPlay and Android Auto, and keep shipping features that give people more reasons to open the app.

Elections are one of the biggest moments of the year for a newsroom. KQED already publishes a lot for one: a voter guide covering every race and local measure, live results from the AP, and wall-to-wall coverage on air and online. The question for 2026 was what that work looks like inside the app.

Four KQED app screens: the Election 2026 card on Home, the top of the Election Hub with the voter guide, the election alerts toggle and live radio search, and a Forum election-night radio clip
Left to right: the Election 2026 card on Home, the top of the hub, election alerts and live radio search, and a Forum election-night clip.

The Challenge

When we started planning in late August, Election Day was ten weeks away. That sounds like plenty of time until you account for how mobile apps ship.

The content existed, but it didn't reach the app. The voter guide and results pages lived on kqed.org. Tapping one of those links in the app bounced the reader out to Safari, because the app's deep-link routing only recognized standard article URLs. KQED's two biggest pieces of election content were one tap away and still outside the app.

App releases set hard deadlines. Anything that needs a new app build has to be through App Store and Play Store review and installed on people's phones before it matters. Election alerts had the earliest deadline of all, because we wanted to prompt people to opt in throughout October, and you can't prompt someone to turn on a toggle their app doesn't have. That build needed to be in the stores by September 22. Everything else had to be in by October 13. Missing a cutoff meant the feature didn't exist this election.

The baseline data pushed back on the plan. Before picking features, we set success metrics against a four-week baseline, and the data didn't flatter the original pitch. Two-thirds of weekly active users came to the app for live radio, and only about a quarter read articles. Installs barely moved during the June primary. "Elections drive installs" was an assumption, not something the app's own history showed.

Our Approach

Fix the cheapest, highest-leverage thing first. Teaching the app's link routing to handle election pages was a couple of days of work, and it brought content KQED already published into the app. It also made "parity between web and app" mean something concrete.

Sort the work by what needs an app release. We split everything into two groups. Above the line was work that ships in an app build and is gated by the store deadlines: link routing, the alerts toggle, the hub screens, and in-app results built on the AP's API. Below the line was work that ships from a server or the CMS and can change the morning of the election: Airship opt-in prompts, push alerts, and the email itself. That split kept the release-gated list short, so we spent our risk where the deadlines were real.

Design set the schedule, not development. When we estimated each piece, design time outweighed development on almost every item. AI-assisted development made writing the code fast. It didn't speed up the questions that set the schedule: how native the results experience needed to be, where alerts should live, and what goes at the top of the hub. It also didn't speed up code review, QA on real devices, or store review, so we treated development estimates as floors and kept buffer in the calendar.

Lean into what people already use the app for. Since live radio is why most people open the app, we put live radio search in the hub, so people could find, play, and share clips of election coverage they'd missed on air. Election alerts got their own opt-in list, separate from Breaking News, with an on/off toggle right in the hub, so people could get election updates without signing up for everything else.

Tell people it exists. A news app isn't something most people open every day, so shipping the hub wasn't enough. We sent the app's first product announcement email through Airship a week before Election Day, and nearly half of the people who got it opened it. Links opened the hub directly in the app, with a fallback to the app stores for people who didn't have it installed.

Results

Election week, compared to the average of the previous four weeks:

  • New installs: up 53%
  • Active users: up 35%
  • Active members: up 24%
  • Live radio listeners: up 26%
  • Article viewers: up 80%
  • Donation clicks: up 21%

Election week cleared the goals we set at kickoff for installs, active users, and article readers.

The hub built up to Election Day. Daily hub traffic grew steadily through October. It jumped the day the email went out, and on Election Day it hit nearly ten times a typical October day.

The voter guide did most of the work. About two-thirds of everything people did in the hub was opening the voter guide, more than three times as often as they opened articles. "Voter guide" was also the most-searched term in the whole app.

Radio search found a long tail. On Election Day, live radio search taps ran at seven times a normal day. More than 90% of clip plays came from Forum, KQED's daily call-in show, and one October episode kept getting played for 18 days after it aired. Radio normally disappears the moment it airs. Making it searchable gave it a second life.

Opt-in beat broadcast. Pushes to the election alerts list were opened at more than three times the rate of election pushes sent to everyone. Across all pushes, breaking news and results got the most opens, and reminders ("there's still time to vote") got the fewest.

Takeaways

  • Find out where your content is getting lost. KQED didn't need new election content. It needed the content it already had to open in the app. Check your routing before you build anything new.
  • Sort work by what's release-gated. Keep the list that depends on app-store deadlines as short as possible. Move everything you can to server-driven or CMS-driven delivery, where you can change it the day of.
  • Set metrics before you pick features. Our baseline challenged the premise of the project, and it changed what went into the hub. Better to find that out at kickoff than in the retro.
  • AI speeds up the code, not the decisions. When development gets faster, planning and design become the bottleneck. Plan for that.
  • Let people opt in, then send them news, not nags. A smaller list of people who asked for alerts outperformed a broadcast to everyone, and people tapped on news far more than on reminders.
  • A launch needs an announcement. A feature people don't know about doesn't help them. For a product people don't open daily, the announcement is part of the feature.

Planning for a big moment of your own, whether that's an election, a launch, or a season? Let's talk.

We make great products
Looking for a partner to help you create a successful business and amazing software products? Get in touch with Uptech Studio today.
Get Started