[forcer]

I am not sure why this old news is surfacing here today but I can give my 2 cents, since I sold speedchecker.com last year and were directly competing with Ookla.

The main business is selling the data. You use Speedtest.net to troubleshoot your connection but metrics captured with the test alongside location data give telcos invaluable insights on where they should improve their networks. Telcos pay 6 figures annually for this data and we have a few hundreds of of those big MNOs globally. This market is pretty big. Accenture is in trouble with their main consulting business due to AI so acquiring data business is one of the smart strategies they can implement to stay relevant.

To all commenters who think they can code it over the weekend, yes you are right. I coded my first speed checker over the weekend in 2008 but it took me 18 years to grow the user base , figure out entreprise sales strategy and exit. Its not easy as it seems.

[cortesoft]

As someone who worked at a CDN for years, I imagine the code is the easiest technical part. Managing the infrastructure, network connectivity, load balancing, and capacity planning would be the harder parts, outside of the sales and marketing bits.

If you don’t get all of those parts right, you are going to end up measuring your own bandwidth rather than the client’s.

[ksec]

This.

The website, and backend code for the test. 10% of the software work. Which is what everyone seems to think.

The code to managing the infrastructure, network connectivity, load balancing, and capacity planning is the 90% of the software part. But even then it is only 10% of the technical thing.

Getting all the ISP onboard to have your server in their network / exchange and to deal with you, takes more time and effort then all the software part. But even then it is only 10% of the project.

The remaining 90%? Non technical part for Sales and Marketing and getting user traction.

To put that into perspective, the website can be done in a weekend was only 0.01% of the work.

[mbesto]

> but it took me 18 years to grow the user base , figure out entreprise sales strategy and exit.

The audience here has never wanted to admit that the codebase doesn't really matter. Now that codebases can be created in a weekend, people are opening their eyes to this sentiment - the hard part is the sales, the code is easy.

[elxr]

Great code is still not easy. Choosing the right stack/libraries/billing and getting everything to work together (for cheap) is still something barely 10% of devs can actually realistically do.

Sales is hard, yeah, but look at everyone claiming to be building something amazing and it ends up 9 months behind schedule or just being an buggy, untested version of something that already exists in the market.

[hello_moto]

Great code has never been the requirement of success.

Throughout my career no software that hits $100m annual revenue was born from great code. That’s 2 fortune 100 hi-tech companies with other medium sized companies with revenue close to $1B.

There was one company that had better codebase than the others, unfortunately that company struggles to hit $2M MRR…

Looking back, it was painful to admit that code quality was not how the company succeed: it was overall strategy and luck.

[anotherevan]

> the codebase doesn't really matter

Sigh.

I’m sure the viewpoint from being in mergers and acquisitions is quite different (and to me, often comes across as quite callow). I’ve been a software developer for 35 years (closer to 45 if you include my pre-professional life, aka adolescence) and have deliberately stayed “on the tools” in my career with working in codebases and product development as I’ve found that is where I am happiest and can make the best contribution, rather than move up the managerial ladder to my level of incompetence, to quote Peter.

To create a successful product in IT, or any industry really, it takes a lot of different skills, facets and (often competing) priorities. And those priorities do change over time. I’m sure by the time a product or service crosses your desk, the codebase quality is not as big of a priority. Earlier in the life cycle a shit codebase makes for a shit product that is a lot harder to grow and maintain — so much so that most of them have probably folded before they reached the stage of looking to be merged or acquired. I’ve dabbled in sound mixing for live performance and when training others I’ve mentioned the fact that it very hard to make a bad singer or musician sound good, but very easy to make a good singer or musician sound bad. Same goes for trying to make what would otherwise have been a good product or service with a bad codebase. That’s really hard and creates a hell of a lot more work for every part of the business.

I’ve had sales people tell me to my face that they are the most important part of the business and the actual product or services is not that important. And in my more callow stages of life experience I’m pretty sure I’ve reciprocated with words like useless and parasitic, and that I could replace them with a small bash script. But in reality what we all do is important to the complex endeavour of developing and maintaining a successful product or service. The existential threat of AI is moving up the ladder of incompetence and changing the face of what we do. It may even jump a few rungs in the process. But it’s not there just yet. Keep making good sales, keep making good mergers, good products, good acquisitions, good services, and good codebases.

— No tokens were harmed in the production of this comment. —

[bruce511]

Your argument is sound. It certainly takes a good deal of skill to create good code. And yes, good code makes it easier to create a better product.

And yes it's easier to build a better company on a better product.

But history is littered with "worse products" that won in the marketplace.

It turns out that all the attributes you name are helpful but not necessary. Good marketing trumps good product. We see this over and over again.

The best combination is good marketing and good product. If I can only get 1 of those then I'll take hood marketing. Equally if you have a good product but bad marketing you don't get many (if any) users. The "ask" section on this site is littered with that.

So, assuming we can all make "good enough" code, the code doesn't matter. It's all good enough. The distinguishing feature is the marketing, because that leads to market share, and that's all any company is really selling (once it sells for a lot).

I'm upvoting you because your comment is well made, and certainly common, even if it is incorrect:)

Having been involved in multiple different acquisitions, on both sides of the table, I can anecdote that the code quality had no impact on any part of the acquisitions. The players are not buying or selling the code.

[vanviegen]

> So, assuming we can all make "good enough" code

Your entire argument hinges on "good enough". Problem is: you can never know if something is "good enough", except in hindsight for those products that succeeded.

I'm upvoting you because your comment is well made, and certainly common, even if it is nothing more than a tautology :)

[bruce511]

The bar for "good enough" can be set quite low. In general, consumers can be convinced to buy almost anything. And their resistance to good marketing is very weak.

The problem with presenting good examples is that decades of sustained marketing is hard to overcome even with facts which are immediately obvious. Indeed good marketing has already negated those facts.

For example smoking is objectively terrible and yet was (and is) very popular for decades. Tobacco might be out, but vaping is still cool; same message as before.

From outside its easy to spot US examples because their absurdity is obvious to outsiders. It's harder to see examples in one's own society (because we have our own marketers.)

In software land there is obviously lots and lots of complete rubbish. Most of it gets no marketing at all. But is Windows the best OS? Is Chrome the best browser? Is Google the best search engine? Is Facebook the best social network?

Or do each of those have a competitor with "better code" that has no marketing and gets no traction?

When IBM hooked up with MS was it because of good code? When Sun bought MySql was it for the customer base, and Brand, or the code?

Did Facebook buy WhatsApp for 18 billion because of the code? Fo you think they compared the code to some other messenger with 100 users, or did the 400 million people using WhatsApp matter more?

In truth every product you ever heard of, and ever used, was good enough. Github is full of projects with really great code and no users.

There's a fundamental disconnect between business people and codesmiths. The programmer wants another year to craft perfection. The business needs to start selling and earning next week.

Good code lasts longer, and is better for the company in the long run. Engineers know this. Companies know they have to ship, and sell and earn, to survive at all. Engineers sneer at marketing, the product should be good enough. (Tell that to Amiga.) Marketeers are frustrated by Engineers who want to build forever and never ship. (Any wonder they want to replace us with AI.)

Yes AI products are objectively worse. But if history tells us anything; that doesn't matter.

[vanviegen]

I agree with most of what you're saying.

But was Google the best search engine when they got popular? Most definitely! Chrome the best browser? According to most, yes! When MySQL got popular, that was also due to it being the best free product out there at the time.

That is to say, a good product can easily be the key factor to growth. Especially in the critical early phase. But it'll have to be a lot better than the well-known alternative.

[throwaway7783]

But doesn't your counter point assume all products that failed were not good enough?

[vanviegen]

Well, at least the combination of code, marketing and luck wasn't good enough.

[tormeh]

I've been in an acquisition where code quality was important. But it was probably an edge case since the buyer just wanted to turn the company into a feature, and ease of integration into the buyer was important.

[smugma]

> The players are not buying or selling the code.

Counter-point: many acquisitions are not for the code itself but for the engineers who designed and wrote the code. Acquihire is almost exclusively used to describe acquiring engineering talent that can… design systems and write code.

[consp]

Marketing without a product is called a scam. Marketing with a "good enough" product to sell it is the same.

[anotherevan]

Sigh.

Bless your heart.

[mbesto]

> I’ve had sales people tell me to my face that they are the most important part of the business and the actual product or services is not that important.

Same here. This happened when I was 22 years old I didn't want to believe it.

BUT, I've seen far more shitty codebases win marketplaces with strong sales & marketing, than I've seen stellar codebases with shitty sales & marketplaces win marketplaces.

It's unfortunately really that simple.

[jbs789]

I think this perspective benefits from experience, the ability to step outside one’s self, see that the world is complicated, then focus on the thing you enjoy.

As much as I agree with you now, I also accept that younger me wouldn’t have!

Very well said.

[Gigachad]

Most of them can't. It's just a few products like speed testers and social media that are at least surface level easy. You can't vibe code up a triple A quality game in a weekend.

[iugtmkbdfil834]

To be fair, triple A gaming lately has been something of a letdown. That does not invalidate your point exactly, but it does mean that the tastes of the gaming audience might be changing ( we kinda were in similar spots before ).

[BoggleOhYeah]

I'm not sure anyone was operating under the idea that a speed test website's code was the hard part.

[themafia]

> The audience here has never wanted to admit that the codebase doesn't really matter.

Are we talking about speed testing websites or the code that controls space vehicles? Perhaps extreme generalities do not provide useful insights.

> Now that codebases can be created in a weekend

Now that corporations are whitewashing copyright off of code so you can steal it without conscience.

> people are opening their eyes to this sentiment

Code is the product. Engineering is the discipline. That you can achieve high sales without good engineering is not a new idea. That it only provides short term benefits and leaves you irrelevant in the long term is the actual sentiment.

> the code is easy.

Coding has been easy since Perl was released. Knowing _what_ to code is the problem.

[bigfatkitten]

> Engineering is the discipline.

And even that is very rare in the field of software development.

[andix]

The partner network of Speedtest is also impressive. I don't know how many speed tests they need to handle in parallel, but usually it's always enough to do speed tests up to 5-10 Gbit/s. With more and more fiber connections also latency becomes very relevant. Otherwise the tests would be meaningless. Speedtest manages to measure less than 1ms latency on my fiber connection.

[forcer]

Once you have a good amount of users testing, its not that difficult to get free servers from the ISPs. The secret is that on-net servers show testers better performance than off-net so every ISP wants to contribute the speed test server. If they dont do it they are shooting themselves in the foot by routing their traffic to competitor networks and getting test results behind their peers.

Whats even worst then your competitors can claim awards for the Fastest ISP and your marketing people are furious!

[mapBasketWand]

That was a persistent conversation with ISPs when I was building a white label WiFi product (competitor to Eero).

Some ISPs wanted us to pin to their servers in our app to have the best possible results (we refused) while others wanted us to use their servers because they offered 10G service and none of the other servers had that much throughout. So their true 10G line would be limited by the server, not the line.

[andix]

Sure, but it's still a huge effort to set up all those partnerships and keep them alive. ISPs are often traditional and slow moving companies, it probably takes a lot of work to get those servers in place at the right locations.

[bigfatkitten]

And some cheeky traffic shaping by your competitors to make your results look worse than they should be.

[_fizz_buzz_]

When you first built it, were you aware there was market for the data? Or was this something you discovered afterwards? It makes sense, but I wouldn’t have guessed it.

[taneq]

> I coded my first speed checker over the weekend in 2008 but it took me 18 years to grow the user base , figure out entreprise sales strategy and exit. Its not easy as it seems.

The biggest surprise (imo) when you start a business is how little of running a business is actually directly about the product. Having a product is essential, sure, and having a good product is nice, but that’s just the tip of the iceberg.

[altairprime]

It’s less than ninety days old and it isn’t (2025), so I wouldn’t consider this as ‘old news’ yet. TIL, for example! But if you think it’s a dupe/repost and should be squashed, email the mods a link to both this and the prior post so they can evaluate.

[Raed667]

How come ISPs aren't providing that data internally from observing their own traffic ?

[k2enemy]

I'm sure they do, but data about the speeds of other ISPs is also valuable.

[hocuspocus]

I'm not sure about broadband data, as it can't be that useful. However on the mobile side, it's fairly valuable as a mobile app can collect A-GPS location and sensor telemetry that are unknown to the MNO otherwise.

[rajveerb]

@forcer would love to know how you "grow the user base , figure out entreprise sales strategy and exit"

[steve_adams_86]

Apparently it takes roughly the same amount of time as raising a child

[rajveerb]

haha sounds like it.

[iugtmkbdfil834]

Thank you for your service ( the product was/is -- haven't used in a while -- useful ).

I think this is the part that people do not appreciate. Sometimes it genuinely it is not the difficulty of the task from a pure programming perspective, but 1) getting the users and 2) getting people to pay for the service and 3) getting the right people to sign off on that.

It is very similar in banking. The products themselves are not super hard ( though the challenges are real ), buy just getting to talk to the right people is a hassle.

[saidnooneever]

thanks fornthe insights. inthought it was a wopping number but this makes totally sense. never realised this was gather valuable data for network operators. cool insights!

[Melatonic]

Any recommendations for similar tools to check network metrics other than speed ? Used to be a few free ones but would be nice to have an easy one to use

[esseph]

Most will show you speed / latency / packetloss with the toggleable option for multi-single threaded.

What would you like to see?

[8cvor6j844qw_d6]

ISP seems to give higher network priority to Ookla so I'm not sure how useful it is compared to actual experience.

[NetMageSCW]

Like all speed tests it should be considered the maximum you can achieve.

[dgellow]

MNOs => Mobile Network Operators

[brikym]

Are you able to share the details on how it was valued? Was it N times revenue or anything like that? I have tried to value a property several parties are interested in and found it quite mysterious.

[glaslong]

> To all commenters who think they can code it over the weekend, yes you are right... but it took me 18 years to grow the user base

I should get this printed and framed

[aryehof]

> Accenture is in trouble with their main consulting business due to AI

Is this something being seen across all outsourcers like Accenture, Wipro, Infosys etc?

[ricardobayes]

Telcos wanting to improve their networks is news to me. I always thought providing the bare minimum is basically their business model.

[zulban]

Providing the bare minimum when you're not is improving. ;)

[athrowaway3z]

Did you ever fear that any big player like Google, Apple, Valve (though fewer mobiles), Meta, OpenAI (nowadays), etc decide to get involved?

[qsxfthnkp2322]

Apple has their own internal speed app due to privacy concerns.

[qsxfthnkp2322]

https://theapplewiki.com/wiki/Speed

[kalleboo]

And on macOS Apple has `networkQuality` on the command line which runs a speed test against presumably the iCloud servers

[guessmyname]

Not iCloud servers per se, but close:

• ussea4-edge-fx-012.aaplimg.com

• mensura.cdn-apple.com/api/v1/gm/large

[qsxfthnkp2322]

Gotta collect that data on how the services perform. lol

Snark aside, at least they have tight data controls and don’t sell that information.

But I don’t really have a lot of faith in the whole story and/or some of the management inside. Yeah there usually is always two sides to the coin. But tbh It’s too secretive. There has to be something there outside of simply “protecting the surprise.”

[contingencies]

Just wanted to say congratulations!

[PunchyHamster]

Huh, interesting, I thought ISP would get enough metrics from their own devices

[frankdenbow]

what was the point where you had enough data to make it worthwhile for telcos?

[forcer]

I would say if you can reach 1% of the population in a given country every month then you are starting to be interesting for the telcos.

[krm01]

What about city specific data? Is that equally/more/less valuable? At what scale do you think?

[AndrewKemendo]

Why did you sell it?

[forcer]

Because I got a good offer :)

[sph]

Many such cases

[cortesoft]

Money without having to do more work > money that you have to keep working for

[arikrahman]

All it takes to defeat the business model is https://openspeedtest.com/

[jedberg]

That test was totally inaccurate for me. It got the download right but upload was only 1/12 of my rated speed and 1/12 of what all the other tests (and my actual experience) tell me.

[archb]

Same for me. The download speed was fine, but upload speed was 1.75x the actual one.

[girvo]

Same for me: download was okay, upload was completely wrong

[steve_adams_86]

My very limited experience in this space is that measuring uploads is actually quite a bit trickier and harder to nail down than download speeds

[arikrahman]

Perhaps you can audit the https://github.com/openspeedtest/Speed-Test codebase

[mbesto]

Or https://speed.cloudflare.com/

[arikrahman]

Moving from one proprietary solution to another defeats the purpose. That's why I linked a FOSS solution. The repo is here https://github.com/openspeedtest/Speed-Test

[swiftcoder]

Cloudflare's undercounts my download by 25% and my upload by 75% (versus both speedtest.net and my observed sustained data rates). Also reports double the latency.

[mbesto]

Undercounts it by what barometer?

[swiftcoder]

Versus versus my own measurement of sustained download/upload rates (this isn't rocket science - upload a 1 GB file to S3, download it again, time both). speedtest.net is within ~5% of my own test results, cloudflare isn't even in the ballpark.

[UqWBcuFx6NV4r]

evidently putting “open” in the name of a website doesn’t suddenly make it not-bad.

[arikrahman]

It is open. https://github.com/openspeedtest/Speed-Test

[hbwang2076]

Solid point on the infrastructure testing gap. Accenture/ookla collaboration is interesting - real world speed data meets enterprise rollout.