Subscribe
Sign in
p/deep-dive-duckOur AI reports show key changes to competitor sites
Start new thread
pete bray

Election 2020 Web Change Monitor - We crawl candidates' sites and use ML to surface key changes

by•

We're tracking changes to US Election 2020 campaign sites: Bernie, Biden, Trump, Warren, etc. We trained a model based on key changes detected in prior campaigns. Explore code, content, and visual changes for each candidate's site, with key changes highlighted

Add a comment

Replies

Best
pete bray
Maker
We're really excited about the data we've captured so far! We've detected key changes from Bernie Sanders in terms of immigration, from Elizabeth Warren in terms of revenue, and from Donald Trump in terms of his website's Terms and Conditions. (Take a look at the links near the top for key changes detected to-date.) Here to answer any questions you might have!
pete bray
Maker
We had an eventful PH launch ;) We somehow ended up getting launched unexpectedly at midnight vs 6am... not ideal, but ok. *But* it turned out no one could upvote us. Fortunately, PH staff reset us to launch at 6am... but perhaps w/o the "freshness boost" (as we had already used ours up while live at midnight). 😂😅
tela
Congrats on the launch, Peter! This will really help with keeping on top of candidates’ changing positions. I’d love to see this represented visually as a timeline
pete bray
Maker
@tela great idea, Tela, thanks. Yeah, an overall dashboard is something we have thought about... finding a way to visually represent is tricky. I suspect it might be like the little "Github flare chart" that shows commits over the past year on a day-by-day basis. This way, users could spot trends and so on.
tela
@petebray I think a timeline view with a filter toggle for changes, and another for news events would be an easy way to do it. I think the value here is the context of these changes in the broader political milieu, and less about the data view
pete bray
Maker
@tela thanks, that's good insight. The political monitoring in many ways is a fun example we use simply because they change their content so frequently and so notably (e.g., some of the issues pages). Most prospective customers want to monitor competitors, which mainly means tracking exec bios, locations, new products, pricing, and so on. Often, they want this coupled w/ other competitive insights, such as social media, blogs, LinkedIn, and so on into "battle cards" or the like. So we're thinking of how we might wrap other services together to supplement into a dashboard that provides change analysis, but all the rest too. Tricky row to hoe...
Nathan C Bowser
Wow, As a marketer I've had about 5 epiphanies of different applications of this data. From competitive SEO monitoring, to content governance, to optimization workflows I'm so inspired. The fast visualization of the content updates is so helpful. Nice Work!
pete bray
Maker
@nathan_c_bowser Thank you Nathan! Placing our bets in the right industry and use case is one of the trickier things we're facing... we also think there are a lot of uses, but zero'ing into one is challenging. We also want to make sure we build enough of a "moat" as the core use case, while technically actually quite difficult when you dig into it, often can invite more competitors. So we're thinking about how to up-level the data we capture with machine learning, metrics, and other valuable data stories for in-the-weeds analysts *and* C-level execs. That's the hope anyway :)
joshua bradley
This is potentially a very interesting tool and shows a good use case for how to use it to monitor competitors. It does look like the ML floating relevant things to the top might need some work, as it points out changes on an important page for Warren, but it's just a change in the copyright year.
pete bray
Maker
@airjoshb which one are you seeing? we are only highlighting the ML surfaced data in the list near the top. all of the other changes are just the raw changes that you can explore... many of these, as you noted, aren't necessarily going to be super interesting. There's a lot of chaff in political campaigns in terms of copyright year changes, promotional banners, and so on... we still capture all of these, but the ML does help find the "wheat" (to continue my metaphor haha)... the scores are available on the backend, and we haven't actually surfaced these in the front-end, except for the curated list near the top. (The page-by-page scores will be revealed in a future front-end update, we are just trying to find a way to reflect them in a meaningful way (e.g., perhaps a letter grade.))