The Texas Tribune

Interview with Liam Andrew and Rodney Gibbs

February 2020

The Texas Tribune is a digital-first nonprofit news publication supported by a growing membership. Founded in 2009, it focuses on statewide issues, and provides its content for free to other news organizations throughout the state. Rodney Gibbs was, until last year, the Tribune’s Chief Product Officer and now leads their Revenue Lab experimenting with and testing new revenue ideas. Liam Andrew is their Director of Engineering, taking the lead on content management, distribution and analysis.   

What’s the most important technical issue facing you at the Texas Tribune right now?

Liam Andrew: It’s always the meta problem of buying versus building. We’ve had our own CMS for around a decade and it’s been built on the same CMS with consistent improvements. The most important thing is to make sure we continue improving it, keep ourselves up to date, and minimise our technical debt.

What made you decide to build your own CMS?

Liam Andrew: The Tribune at the time really wanted to have in-house tech talent and expertise, in part because they wanted to do a lot of data work. So I think it was a nice extension to say, well if we have good tech in-house, then we’ll also have good data platforms. It made it a little harder to just run with something like WordPress, because they wanted to publish stuff outside of the traditional story format.

Rodney Gibbs: Also, we got a big grant back then from the Knight Foundation to build a CMS. And I don’t think this was quite explicit, but in some ways it was like it was felt we’ll save journalism by building this tool that will be useful not just for the Tribune but for the whole news ecosystem. So it set us down this path, and we’ve certainly considered many times as to whether we get off it. But how do you do this today? If we had, we definitely wouldn’t be on the exact path we’re on right now. 

So 10 years later, how do you run the tech strategy in your organization? Is there an ongoing multi-year goal of where you want to get to in terms of tech? Or are you basically working on a year by year basis?

Rodney Gibbs: More year by year. We have some bigger objectives about tech features, but it’s a struggle for us in that when we get beyond maybe two or three quarters out, it gets pretty ill-defined. Part of the struggle is that we on the technical side have definite ideas and objectives we want to hit; but it’s an ongoing balance of trying to coordinate this with other departments and get them to be willing partners.

So what does this look like for 2020?

Rodney Gibbs: We developed a user portal last year, which we’re going to expand this year. It allows our donors and members to have more control over their details in our system, about payments, newsletters, and how they engage with our platform.

Liam Andrew: We’re starting a big partnership with ProPublica with our investigative department. I think this will expand the way that people want to visually tell stories. And it’s going to break the limits of our CMS here and there, so we’ll have to figure out the best ways to fit this in. We’ll also be onboarding a lot of people in the next year and this will require them to not only get IT and security onboarding, but also teaching them our tools, our internal tools for project management, task management, our Wiki, that sort of thing.

This kind of tech strategy sounds like it comes from business as well as editorial. Do you develop them in conversation and collaboratively with these departments?

Rodney Gibbs: Collaboratively. We have pretty low walls between departments at the Tribune; there are around 70 people right now. For the most part it’s quite collegiate and cooperative.

Liam Andrew: Two years ago we released the company strategic plan up to 2025 and this is what every department is working together to get to. It’s been a good guiding force when we do have issues across departments or questions about what to focus on. 

So does this plan include tech strategy or is it more based on business and audience goals?

Liam Andrew: We said we don’t want tech to dictate the future of this place. We want tech to be adaptive to the goals that the editors have. But it was trying to figure out a tech strategy that would be adaptable and reflect the way that we want to produce. 

Thinking about the last major tech decision you made, what was it and what informed that decision?

Rodney Gibbs: The user portal last year and it was a major decision because we hadn’t had a system like it before. We wanted to give our business and membership teams ways that they could make self-service tools for their constituents or users. Changing credit card payment details, addresses, and email contacts used to take a lot of phone calls or emails and it was really cumbersome. We had a really primitive login system for our site that was essentially just a comment on an article. But it was clunky and ill-used so we needed something more state of the art. It’s nice that users can update their credit card details now. And it plays into the bigger strategy of trying to move towards more personalized content. But the really important part now is it’s a very solid foundation for future features and ways that users can customize their newsletters and data. Or over time, if we can get them on the site more, to perhaps feed them content that we know they like whereas now, it’s one size fits all.

And was this based on any frameworks or platforms or did you build this from scratch?

Liam Andrew: We’re using Auth0 for authentication but then we built most of the platform from scratch on a flask-based Python app. We use Salesforce and a lot of Salesforce integration, and we’ll be integrating with MailChimp and other things, other user preferences. 

Is there a software decision you’ve made or products you’ve used that you wish you hadn’t?

Rodney Gibbs: This is ancient history, eight or nine years ago now, but we started with a bad CRM before we moved to Salesforce. We stuck with it for about five years and that was a bad decision. When I joined the Tribune, it was one of the first things I wanted to change. But the person who ran the revenue side at the time, the membership side in particular – I believe her words were, ‘over my dead body will we do this’. It wasn’t so much that she loved the system, but it worked for her, and she was scared to death of a new system jeopardizing her ability to hit her numbers. It took two or three more years to get the political wherewithal to move forward and jettison the old system. There was a very tangible business reason in that we realized we were leaving money on the table because we just weren’t tracking our users well. We were sometimes even insulting our users in that we might have had an ask out to a major donor for $10,000; and the other side of the system that’s soliciting $10 donations wouldn’t know this. So they would send that same person an email saying, ‘why don’t you give us $10?’ We had egg on our face when people called us to say ‘I thought you wanted $10,000, now I’ve got this email saying ten dollars, I’ll take the ten’. It was a year-long project to change systems, not just to build a new system but to clean the data and get everyone to lock step onto it. It was painful.

We also did a major CMS upgrade over a year ago and it was very analogous in that we had to set expectations – it’s going to be painful, and it’s going to be slow, and you’re not going to get a lot of extra work out of us for a long time because everyone’s heads are down on this. But we’d built trust enough to say that when you come out the other side, you’re going to be delighted. We needed a lot of political capital to make it happen though. 

Liam Andrew: And one thing since then is having this new framework has helped us push features and improvements faster than we used to, which is sort of a meta help just to explain to everyone, look, that thing that used to take us two weeks to fix, well now it only takes a day.

How has the implementation of your tech strategy changed over the years? Have you always focused on a couple of big jobs or has it moved to a more continual testing environment?

Rodney Gibbs: We’ve been able to do continual development and gradual improvements over time. Every Monday morning there’s a weekly editorial meeting. We have a slide as part of it that tells them what we did during the last week, and pretty much every week there’s something small we did for an editor or reporter.

Can we just stop you there? Because that sounds incredible. The amount of newsrooms that have a slide from the tech team explaining how they’ve improved the technology every week, you can probably count on one hand. Have you always done this at the Texas Tribune?

Rodney Gibbs: Maybe the last two or three years. It wasn’t always this way. I think it reflects a bigger change at the Tribune in walls and silos coming down. It’s happened between editorial and revenue. It’s happened between technology and everything. When I joined the Tribune there were huge walls and the engineers in particular were siloed by choice. There was very little communication, very little process. And so to get to the point of having a slide in the weekly meeting, I think our Editor-in-Chief at the time really saw the value in collaborating and communicating. So even though it’s an editorial meeting, it’s also our de facto weekly staff meeting because almost everyone’s in the room. I think it reflects an ethos here of communication and trying to break down walls between departments. And you know, thinking back, there have been some times over the years when I had complaints from the Editor-in-Chief or CEO about not knowing what the tech team’s doing and ‘you guys speak a different language’. And so this was an easy way to try to explain what engineering was doing and often with very tangible benefits like, ‘that button that always bugged you, it’s fixed now’. So it’s a way to show people what’s going on.

Liam Andrew: We also do a tip of the week. So we’ll tell people about a feature in the CMS they might not have noticed. We also try to throw in things that might not be affecting them, but so they know what we’re doing. For instance, if we spent all of our week on a big upgrade that they’re never going to notice, we’ll still tell them so maybe later they’ll notice improvements. So they have no idea what Python is, but they’ll say ‘oh you were working on that Python thing last week’. We’re a Python shop. Our CMS is built on Django.

Does this mean that you’re getting more people from the editorial team asking for tech improvements?

Liam Andrew: I think so. It depends on how busy they are and depends on the individual. But there are not a lot of walls here, so people walk right by and just stop and ask us about something going on.

Rodney Gibbs: I’d say yes but not enough. A lot of the reporters just want to be left alone to write, which I get, but on the product and engineering side, we’re always looking for partners in crime. I wish more of them were like squeaky wheels saying ‘I don’t like this, can you fix it’, or ‘hey, I have an idea, I want to try something crazy, can you help me?’ We have a few of these people, but it’s a minority of the editorial staff. 

What’s your relationship to Facebook and Google and the dominant tech platforms? How important are they to the relationship you have with your audience and your distribution strategy?

Liam Andrew: They’re very important, especially right now. Google is important for us, like the SEO implications, and running our own CMS. We sometimes feel like we’re flying blind and not sure exactly where to integrate with the platforms the most, or the best. It’s weird to me how reliant we are on Amazon, and in particular AWS; that we’re reliant on these platforms for hosting as well as for our audience. We do have some other servers – we have Heroku, we use Google cloud publishing. We use Google for all of our internal tools, like our email or documents. And, for instance, our partnership with ProPublica; a lot of the time Google Docs is the common denominator when you’re doing partner work. You don’t have access to their CMS, they don’t have access to yours, so you work in Google Docs together. Especially as we’re trying to ramp up partnerships with other newsrooms, I feel like Google integration for story production is increasingly something we’re talking about. It feels weird to me how reliant our CMS could become on Google.

Have you ever deliberately chosen an alternative to one of the big tech platforms?

Rodney Gibbs: This isn’t quite the same as the Googles and Facebooks, but our initial CMS was more of a Swiss army knife that had a homegrown email system, RSVP system, and newsletter system. And over time we realised we needed a steak knife not a Swiss Army knife, so we stripped out a lot of those things and now we use MailChimp, Eventbrite, and other services that specialise in these things. Then we integrate with them via API or other methods. So we’ve deliberately said we don’t want to own an RSVP system, let’s trust someone who does this for a living to do this. But there are trade offs with that. We’ve had to switch vendors a couple of times in the past because of privacy or technical concerns.

We have a love hate relationship with Facebook. I think everybody does. We certainly support it and share all of our content on it. We’ve done some innovative stuff with private Facebook groups, We have a love hate relationship with Facebook. I think everybody does. We certainly support it and share all of our content on it. We’ve done some innovative stuff with private Facebook groups, engaging people with political content. We did go all in on Facebook instant articles for a while and then we quickly realized that the juice wasn’t worth the squeeze so we backed out of it and went back to standard feeds, which made us in the engineering world very happy. We also tried many times to play with them with the fundraising aspects; but because they don’t share data with us and we’re big on transparency – so if you give us a dollar, we have to know who you are because we put your name up on our website and say thank you for your dollar – we decided not to play with them with any type of donation. This is hard to do as the potential seems great; there are so many people there and we could raise a lot of money. But it’s always been a deal breaker that if they won’t tell us who they are or how to contact them, then we can’t take the money.

Could you tell us about your decision outlined in your strategic plan about being a product not a story culture?

Rodney Gibbs: Over time, we made a lot of progress improving the relationship between editorial, engineering and design, to have a much more collaborative partnership. And our Head of Newsroom at the time saw that the reporters weren’t staying up all night to get something done, the stories looked better, they weren’t as buggy. And the engineers were happier because they were being treated as partners not as service people who just do what they’re told. So as we were looking at a strategic plan for the company as a whole, editorial and engineering are really embracing this product thinking and asking good questions; but we’re not doing the same for the revenue side. So the product thinking was a good framework and reference point to show how we’re going to permeate it through the whole organization. And the revenue leaders, the people who generate our money, wanted this. And the quirk of the Tribune is that we have two women, one who leads on corporate giving, the other on foundations and major donors, who are miraculous people and generate the vast majority of our revenue. But their working methods could be more efficient, and they work themselves to death. It’s not replicable or sustainable. So I think both of them realized that we’ve got to get better processes and better methods in place or we’re going to keel over. And more importantly, and more altruistically, the Tribune is not going to sustain itself. So we need to figure this out over the next few years.

Liam Andrew: When we think about products instead of stories, it does lead editorial to think about packaging news a little differently. This includes thinking harder about our newsletters and our podcasts, not just stuff that goes on our website; and the taggings and the headlines and all the things we’ve given short shrift to over time. But we don’t own different platforms in the same way that we own our story producing platform. For instance, we use MailChimp for newsletters. They’re not on our website, they’re not produced by our tech. So if we have a problem with production or analytics, we might be out of luck depending on what it is. Same with Eventbrite. It’s something we’ve been thinking about. So on the one hand we’ve done a really good job at going from the Swiff Army knife to the steak knife, and just being really good at making stories. But I sometimes wonder if we are going to wish we did a little bit more in-house to see what takes off in the next few years. 

Portrait of Liam Andrew

Liam Andrew

Director of Engineering at Texas Tribune

Portrait of Rodney Gibbs

Rodney Gibbs

Revenue Lab Lead at Texas Tribune