Creating a free, Strava-like, app

I have several regrets in life, and one of them was me dropping the ball on a fitness tracker with gps segments that I started creating in 2004. My issue back then was that getting hardware with GPS built in for fitness tracking was near impossible.

But anyway, it is not easy in the sense that there is way more engineering than you think when it comes to the backend of something like Strava. Even managing the storage of all those gps files is not easy.

The algorithms to do segment tracking and evaluating whether any given route goes over any of the millions upon millions of segments out there. Dealing with inherent drift that GPS watches have, all the different models and handling automatic sync.

I'd love to be involved in such a project, but it is not trivial.

I have pretty limited programming knowledge, really only 1 class, so I'm probably underestimating how difficult it is.

Still, it seems like if you have the GPS mapping already, then you would just need to run something that would look for segments near where you ran, and look to see if you traveled within x feet of it. GPS tracking is also still super accurate. Life360 shows where I am within 10 feet or so almost all the time, so just having a small margin of error on segments seems like it would fix the problem.

Strava stores millions of new activies every day. Where do you propose your free app is going to store all of that data, for free?

As someone who has several apps in both app stores, it takes a lot of work and money to do what you are asking for. It would take a year just to be able to properly sync data from all of the third parties. Unless you have a full stack expert and a great mobile dev you aren't going to get very far.

Even ignoring the development costs, storage and bandwidth cost money. The running2win guy had to start charging money even though he wasn't interested in making it a business and his app is 100 times simpler than Strava.

There are ways to monetize without making people pay up front. I suggested ads earlier, and that seems to work for a lot of free apps, though I do understand it may be difficult to collect enough revenue. I just don't know how much space the data for activities would take up.

Here are the two reasons I wouldn't bother trying to replicate strava:

- They must accept data in all different kinds of formats, from all different kinds of sources (manually uploading various file types, interacting with the APIs for various run tracking services like Garmin Connect).

- You have to store all that data somewhere.

These are the factors that take it out of the fun-pet-project domain for me

I’d guess the time, cost and headache is not worth $60 per year.

I've written a little running/mapping/gps app (incomplete, but it does the basics). The problem is, you get so far, and then the extra stuff spirals.

I think conversion from GPX etc formats is fine, there's normally a library out there to do that stuff. Once someone has written it in another language, you can convert as well.

The fiddly stuff is user accounts, storage. Then there's privacy policies, authentication into various other systems (like people would like to upload to strava, again there's normal a lib for that already, but then they go and change something and it eats up your time).

Then you probably want a messaging system, segments etc. Again it eats up too much time (as a sole developer).

Having said that, I do think the way to go, if you ever want it complete, is to open source it on github and try to get some contributers out there. Many hands make light work, and it can be fun collaborating with others, and who knows where you end up.

So all of this (apart from the basics) sort of relies on volume of people. How do you get to that stage of getting all these people to start using it, and not getting annoyed when feature x isn't as good as Strava or Runkeeper or something.

But, again, it's fun learning and trying :).

Rocket Mann wrote:

...an app that had all of these things without a paywall, and maybe ads or something, ...

Any thoughts?

I'd much rather pay a few bucks a month for Strava than have a free version with ads.

An open source version is not a bad idea, and depending on tech stack used, I think that's something I'd be happy to join.

I have thought about the opensource option. The first question would be, unline a lot of other opensource things, this app actually requires some place to store data. So, it's simply a no-go without servers. That's different from an opensource text editor or compiler, which is inherently just source code that can be pulled from the repo and used. So, all anyone ever needs to do is maintain the source code.

Or maybe I'm missing something here. Feel free to correct me

JobbyHogger# wrote:

I have thought about the opensource option. The first question would be, unline a lot of other opensource things, this app actually requires some place to store data. So, it's simply a no-go without servers. That's different from an opensource text editor or compiler, which is inherently just source code that can be pulled from the repo and used. So, all anyone ever needs to do is maintain the source code.

Or maybe I'm missing something here. Feel free to correct me

You're correct, but that is infrastructure and your dev stack will help determine that. I'd think that no-sql / document db would be the right idea for this. Google Cloud Firestore doesn't even charge by document size, only by number of reads/writes. MongoDB has cloud versions too, and I think they are reasonably priced as well.

So then we talk about what should the backend look like? Front end? Apps for Android and iOS...

I have plenty of experience, just passing 20 years in my current job, and the last 18 of those years is .Net and various web technologies. I'd even suggest a Kubernetes cluster as a starting point for the backend. The tech doesn't really matter as long as it can be dockerized, which just about everything can be.

If there is anyone that is serious about this and has some experience, I'd be happy to at least talk through some ideas to see if it could go anywhere.

You never design a solution before doing the requirement analysis

1) You are storing personal sensitive data, which, in wrong hands, could be potentially very badly used (track people, inferring their schedules, gaining or losing sponsors, etc, etc). Just on the security side, that's something that just cannot be stored in a public cheap cloud

2) third party providers for street maps? your own? public open ones which do not offer any reliability or availability SLA's?

3) your app will be used only in USA? all over the world? need to handle timezones? translations?

And many many more...

Yeah but then your app sucks because people constantly see ads on it.

Strava is awesome and costs 7 bucks a MONTH. Is that really an egregious price? Sheesh....

That's the cost of one fancy frapumochachino latte at Starbucks

lcl2006 wrote:

Rocket Mann wrote:

...an app that had all of these things without a paywall, and maybe ads or something, ...

Any thoughts?

I'd much rather pay a few bucks a month for Strava than have a free version with ads.

+100000000000000000000000

Oh Please wrote:The algorithms to do segment tracking and evaluating whether any given route goes over any of the millions upon millions of segments out there. Dealing with inherent drift that GPS watches have, all the different models and handling automatic sync.

Have to say, the FlyBys option on Strava amazes me how quickly it can find all the other people who you crossed paths with on a run. You could pick one from three years ago and it'll load that info in 5-10 secs.

1) I think it depends if one is trying to replicate Strava, or do something a bit different. Some things could be stored locally, but naturally there is at least something you need to store elsewhere, which costs. But, if it's an opensource solution, that's down to whoever to take the code and figure that.

2) Again, that's down to each solution anyone wants to implement. It's trivial to plugin in a url for a tilemap server of your choosing. Openstreetmaps can work fine as a demo.

3) timezones aren't really a big problem, translations is a bit of a different issue though.

Uhh wrote:

Oh Please wrote:The algorithms to do segment tracking and evaluating whether any given route goes over any of the millions upon millions of segments out there. Dealing with inherent drift that GPS watches have, all the different models and handling automatic sync.

Have to say, the FlyBys option on Strava amazes me how quickly it can find all the other people who you crossed paths with on a run. You could pick one from three years ago and it'll load that info in 5-10 secs.

And yet people still feel they are entitled to access this service for FREE. It is truly baffling how people can be that entitled. Strava deserves our money (those of us who use it regularly) and then some. It's a great app.