Blog article: Quarterly Update (Q3 2026)

Article text

At Toronto Open Data, we work in the open. To that end, here’s a summary of how the Toronto Open Data Portal is doing and what our team accomplished between July and September in 2026. 

Performance and Key Metrics

Traffic and Consumption

We’ve noticed our usual seasonality with viewership and downloads this quarter; July and August saw less traffic than most months, and September saw an upwards spike. As in the past, we’ll expect to see traffic stay high until December. We have historically attributed this to the academic year, but wonder what else might cause this.

Since this April, we’ve tracked API calls to consume data and have seen those numbers climb monthly:

The chart above counts HTTP requests to our datastore_search API endpoint, which returns records and attribute-level metadata. This endpoint is what users and apps could use if they wanted to programmatically consume records, but didn’t want to ingest an entire file. Its upwards trajectory makes us think more users are building programmatically on top of our data in addition to one-off data consumption via file download.

All the raw traffic data we use for this analysis is public, if you want to analyze it yourself.

Requests and Publication

This quarter saw a spike in requests originating from members of the public: 53 individual requests! Last quarter, for reference, we received 19.

This, for us, is a good sign: it means more users are seeing the value of open data, and want to engage with the program. It also means that our Public Requests Portal is being used, which is validating for us.

Now, that said, less than half of the public requests we actioned in Q3 led to anything meaningful (by meaningful, we mean a new dataset, an update to an old one, or work leading to one of those 2). This is a flag for us, as it means the Open Data Program may need to think on how to equip requesters to make more-likely-to-succeed requests. Being open, this may be difficult for us at this stage, because we don’t know what data can and can’t be published before we investigate. We’re still trying to find patterns here.

Unfortunately, this doesn’t go well with the recent state of our “queue”; the backlog of requests we have. Our queue has been exploding (growing faster than we can handle) for a year now:

This puts us in a funny spot; as the program gets more popular, we expect more requests. That’s good, and would be great if we were confident we could handle them all. That said, we’re trying a few new approaches over the next 6 months to deal with requests; we’re hopeful this will help address the problem.

The raw data for the requests we work on is public, if you want to analyze it yourself.

Product & Platform Enhancements

We upgraded our gallery! We developed it last quarter, but it took us this whole quarter to deploy it.

We’ve also been working on a CKAN extension to alert users when a dataset gets a breaking change, (like when a dataset’s schema changes). We’ve been asked for this from users for years, and we hope what we make is useful.

We haven’t put any code public as of writing this article, but we are designing this based off of what users have asked us in the past, and have opened a conversation on CKAN’s GitHub about the details of the extension. All we know now is that, in its first iteration, the extension will be able to call out and track several kinds of changes (changes to a dataset schema, to its metadata, and some other details). We also know that, in its first iteration, it will not be able to send emails or webhooks, but will likely communicate via an API endpoint or a feed.

Community Engagement

Engagement

We continued our monthly open data jam sessions (we did them in July, August, and September, and have more planned) where we meet with different open data users, listen to how they use data and what pain points they have. This is important for us; we don’t require user accounts on the open data portal, so we don’t have any strong way to track who is using our data and what problems they might be experiencing. 

We ran a session in Etobicoke, a session in Scarborough, and a session at City Hall in partnership with Toronto Police Services.

We learned a few things here from these:

We aren’t good at drawing a big crowd outside the downtown core. That, for us, is alright; we aren’t measuring the success of an event strictly on the number of people in the room. We also acknowledge that there’s a balance to be struck here; while we don’t measure success on the number of attendees, we can’t say an event series is a success if turnout is consistently very low.

We also learned that attendees want more events with other city staff. After our event with Toronto Police Services (shout out to their Analytics and Innovation unit who are so deep into open data they have their own portal with raw data, maps, and reports), attendees repeated to us that they wanted to hear from other corners of Toronto’s large bureaucracy.

Social Media

While we value our face-to-face interactions with users most, we don’t ignore how we engage with the community over social media.

These past 3 months have seen trends in impressions and engagements similar to the 3 months before; these numbers are highest when we post about a recent or upcoming event, as opposed to when we post about a new or improved dataset.

Impressions this quarter ranged from 200 to over 2000 per day, similar to what we saw last quarter. We have seen our online following slowly and consistently grow across all platforms, and have continued to see the most engagement and impressions on LinkedIn.

Spotlight and Insights

Insights

Toronto Building saw continued high levels of consumption of the data they publish both through file download and API. Their most commonly consumed datasets are related to building permits; the permits that give formal permission to begin construction, demolition, or similar work in the city. Municipal Licensing & Standards saw increased downloads of their business licenses dataset, while Technology Services’ Address Points dataset continued to be the highest consumed-via-API dataset on the portal. These datasets are below.

Spotlights

What’s Next?

The next 3 months for us are straightforward: we’re going to pay off more technical debt, finish our alerting extension, run 2 more events and make a roadmap for the future.

We also started accepting submissions for the Open Data Awards! If you’ve used our open data to build something this is your chance to share your work and be recognized. Learn more here.

Thanks for reading and see you next quarter! As always, you can reach us at opendata@toronto.ca or tell us about datasets you’d like to see via our requests portal.