OP3 Charts #60

Thanks for taking a look. It should work with other shows, but i was tinkering with how I lookup showid and showguid last night so i might have broken that when you look a look. I just checked and it was working working other show guids,

I am fetching the data from the standard OP3 API but caching it so it's a little quicker for repeat requests to the same show data and reduces load on the main API.

where is the show stats page url? I just see a movie on the home page

I re-architected a little, so you'll need to register. Once you register you can search for and add any podcasts and it'll generate the stats page for each. It then auto generates every day.

I was hoping to do it without requiring registration but some of the large shows with millions of records take 20-30 minutes to process, so I wanted to schedule it to avoid making people wait.

Just ocurred to me that you can see the stats pages without login, here's ours:
https://podstats.net/view_stats/929c7e866c694ee08910a9cd8ce7df77

very nice! great job

...as a PS: a future enhancement might be an email alert when something unusual happens. "Your latest episode is getting over 20% more downloads than the previous one", or "Your trailer has seen a lot of downloads over the last 24 hours", or "your podcast seems to be a hit on Pocket Casts", or "your podcast grew more than usual over the last month" etc.

(I bet that without that, it'll be easy for a customer to forget that they have set this up, and therefore won't feel they're getting good value).

Thanks so much James, both for the push and the feedback.

It's give me just what I needed right now which is data. When I first started with this it was for our podcast, which has a few thousand downloads. It turns out that you can write extremely inefficient code that appears to work really well when you only have a few thousand records to parse :-)

I've spent the last week or so making sure the whole thing doesn't choke on large pods, on the day you shared it, it was processing 100,000 records every 1 minute, now the various processors can chew through 1.5m records a minute.

I also wrote a SQL Query Interface which lets you interact with your data via SQL queries, it's under advanced in the dashboard settings view.

Next up is e-Mail alerting and more analytics. I have a rough AI model that's working quite nicely based on the aggregate data we use to build the dash. It's been remarkably insightful and great at picking out changes in trends etc.

Anyway, huge thankyou for the support!

@jamescridland I found a bit of time over the last few days to work on the weekly analytics report, this would be the one I e-Mail out every week to registered users. I've been using PWR as a bit of a test case so i thought i'd share where i'd gotten to and see if you had any feedback.

https://dev.podstats.net/view_weekly/ddb353f16b4b471ca547300e55bbf918

The email is good (here it is - I'm good with PWR stats being public; shout if you don't want this visible though and I'll remove it)
[podstats] Weekly Insights for Podnews Weekly Review.pdf

If a show has released one episode in the past week, I'd ignore it, and focus on the release the previous week. "The recent episode, "Don't just listen to podcasts - watch them, says YouTube CEO" (released June 20), has performed solidly with 467 downloads in the last 7 days" - it's had 467 downloads in the last two and a half days.

The "significant weekly trends" are interesting, but I'd wonder about the actual numbers there, and whether they're actually significant. Is a 112% increase in listeners from New Zealand just three people, for example?

Thanks James, working with an AI model for the e-Mail has been a steep learning curve. I'll keep chipping away at it over the week. After sleeping on it overnight I think the problem is that these models don't actually know what day it is, so I need to provide more context about now() and when each episode was released. Luckily, I have all of the data at hand, so shouldn't be hard to get on top of it.

You can drill down into the key trends table on the dashboard and it will show you download counts per day for that particular event. I am going to experiment a bit more with showing download numbers rather than percentage. The maths is easier working with percentage but it's not as intuitive. I just really wanted to get something out at the weekend to get some feedback.