Hacker News new | past | comments | ask | show | jobs | submit login
Launch HN: Kraftful (YC S19) – UX Analytics for IoT Products
98 points by yana on April 26, 2022 | hide | past | favorite | 52 comments
Hey HN, I’m Yana from Kraftful (https://kraftful.com). We’re an analytics product that helps IoT companies make their apps and displays more usable.

As you’ve probably noticed, everything is becoming internet connected: home appliances, furniture, exercise equipment, toys. But it’s often done in a clumsy, unusable, or unnecessary way—IoT toasters, anybody? Hardware companies haven’t yet figured out where connectivity makes sense and where it doesn’t. At the same time, some great use cases are beginning to emerge: elder care, micro mobility, preventative care, energy saving, and more.

I ran into this problem as head of product at IFTTT, working with the makers of hundreds of connected devices. Unfortunately, the apps and displays that control these products tend to have terrible usability, especially for less technical people. Hardware companies don't understand how to use data to track how their products are used or where they have usability problems, let alone how to fix them. The usability methods developed at software companies haven't made it into their universe. In particular, they hardly ever use analytics tools because those require lots of configuration, and IoT PMs have more than software and UX on their plate. If they use anything, it’s usually Google Analytics because it’s pre-configured. But when they want to answer IoT specific questions (like where users get stuck when they try to connect the device), PMs need to export the data and query it outside of Google Analytics. Most of them don’t know how to do that and don’t have time.

When I was at IFTTT, I would hear from IoT PMs that even when they’d sell millions of connected devices with a recognizable brand, only a few thousand people would ever use the product as a connected device via the app. I wanted to help them get to the root of the problem: How many people tried to connect the device? How far did they get? Which connected features did users care about most and are those features easily discoverable? I was surprised that hardware companies, big and small, couldn’t answer such questions.

I’m sure some of you are thinking: do hardware companies really need more data? But often they’re stuck in the worst of both worlds: their product design enables invasive data collection, while they're not even using non-invasive anonymous events to improve the experience. That's a lose-lose situation, which can and should be turned into a win-win: less invasive analytics and better usability.

Our goal is to provide IoT companies with everything they need to make their products usable by everyone without collecting unnecessary data beyond anonymous events. I’m particularly passionate about protecting user rights. In a prior life, I drove the rollout of HTTPS by default across Wikipedia and worked with privacy at EFF. As we continue developing the product, I want to think through how to nudge the IoT industry in the right direction.

Today, we provide a pre-configured dashboard with metrics that matter for IoT experiences, like device connection success rate. The dashboard includes usability recommendations for how to improve the experience based on the data and industry benchmarks to show how the experience compares to other connected devices.

For example, our dashboard shows where users drop off when using a device for the first time. If users struggle connecting the device in the app, the PM can read about how to simplify the connection flow (like avoiding app permissions not needed to connect the device and any configuration before the device is connected). They can track whether adoption goes up after they make improvements. They can also track how frequently people use different features in the app and learn where to move features so the experience becomes more intuitive. If they need to do a drill down on something beyond our pre-configured charts, they can create a custom chart directly in the dashboard. No need to export data or write queries.

The dashboard is based on anonymized events tracked by an SDK embedded in mobile apps that control the hardware. It does not analyze events from the hardware itself, like sensor data.

We also surface feedback from app store reviews to contextualize the data. IoT PMs can analyze reviews of their own apps or other apps from their industry and get GPT-3 powered summaries of common complaints, features requests, and how users’ general sentiment changes in response to app releases. The idea behind this feature is to give PMs more product insights and identify areas for improvement without burdening users with UX surveys or waiting until things are so bad that complaints bubble up from customer support calls. (Part of this is still in development and will launch shortly.)

I’m excited to get your thoughts on what we’re building! I recorded a short video to show you how the interaction data shows up in the dashboard: https://www.youtube.com/watch?v=hndsQzowic0. If you have a connected device app to try out with the SDK, feel free to pick a free trial on the plans page. Eager to hear everyone’s comments and feedback in the thread!




I am the co-founder of a connected fitness startup and our particular consumer electronics product brings to the table a range of software & hardware problems such as mechatronics, manufacturing, low-latency sensors/processing, consumer safety, content production, content delivery network, fleet management, hell even optics... which already take 100% of the brains of our engineers & developers (if not more)... Even if we all know that down the line, what ultimately is going to make or break us is how much our user base likes our experience how good we get as a company at fine-tuning said experience to create something viral. To that aim, getting, storing and visualizing usage is paramount, but we paradoxically have very little bandwidth to invest in making it happen. Plus I have seen other early stages companies trying to implement their analytics platform... and fail miserably at doing so. I have to personally plead guilty for having implemented some custom ELK-based over VPN telemetry madness in the past... Complete failure.

On top of being labor-intensive, most analytics tool solutions out there are strongly biased towards the app/webapp SaaS paradigm and do not fit our use case very well, and that’s when they run at all on our custom Linux platform (options tend to get scarce when you leave the Android / iOS ecosystems, API is the way to go for us).

Glad to see Kraftful finally tackling these issues, the IoT and larger hardware scene really needs a tool like this.


Having spent most of my work in software before having an opportunity to work with hardware, I think it’s hard for software folks to fully appreciate all the different engineering challenges that go into connected hardware. So it completely makes sense that UX analytics gets little attention and doesn’t get done well, if at all. Kudos for articulating the problem so well and for juggling all this :)

Look forward to getting you the analytics you need to optimize the user experience!


Man I really want better analytics and UI for my IoT apps at home. Right now the apps we use are in their infancy compared to what they could be. I feel like we are in web 1.0 of the IoT apps. And the issue is I end up relying on such apps quite a bit. For eg I used the awair app to change the ducting for my home because of its air quality sensor. Hoping companies like Kraftful can disrupt this space.


Renat from Kraftful here. Great to hear! Awair is actually one of the companies that just started using Kraftful Analytics in beta.


Congrats on the launch! Really great beta and access to analytics for better UI is definitely a need. Many IoT companies seem to focus on iterating the hardware side and lag on the app UI updates. It's easy to spot since these smart companies keep building and launching new products and connect it to their brand's existing app (which is usually looks the same as the time they launched the previous product).

As for the video, the demo is really straight forward and imo easy to follow. Looking forward to watching Kraftul's progress and hopefully adopting the dashboard on our end when we grow.


Glad you found the demo helpful! Would love for you to try out the dashboard and get your thoughts on the experience.

You bring up an excellent point about multi-device apps. When companies launch additional devices, the app is usually an afterthought at best. If there's any new features that go with the new hardware, those features get the most prominent location in the app (usually on the home screen). It doesn't matter whether they're actually things that users frequently want to use. There's so much room for usability optimization in these apps.


Totally agree! So many IoT solutions in particular for the home space are anything but intelligent and user friendly. I live in Germany and use „Smart Friends“. They should definetly be a customer of yours! I am wondering whether Kraftful could also be used in for IoT in factory environment?


Thank you! We’ll definitely look into Smart Friends.

We do want to support industrial and commercial use cases eventually, but they’re further out on our roadmap. The dashboard right now is optimized for the consumer use case. It would need some adjustments for industrial use cases, e.g. good visual representation of interactions with a large number of devices.


My first thoughts on the linked video are that a well written and followed script would improve it a lot, the flow of consciousness thought pattern in your speech makes it pretty hard to follow the concrete features and reasons for the product being worth using (lots of aa's and umm's, tripping over words, changing topics too quickly), it's definitely understandable as it's just a self-made video for HN but I found it pretty difficult to follow.


Great feedback! I’ll see if I have a moment to record a new one later.


Also in the analytics space but not related to IoT apps, curious do you do use batch or stream processing or combination of both in lambda architecture?

Just on a high level what components are you using behind the scenes? I ask because I love this work!


Renat from Engineering at Kraftful. The MVP has a different architecture, but we’re in the process of switching to using batch processing. For now the main components are Segment and BigQuery.


I’m curious how are you using Segment? I used to work at Segment and we didn’t have any IoT customers due to the price being too high


[Context: I worked with this team previously] Data here is coming from mobile apps, and is similar in volume to typical mobile app user analytics. I can certainly imagine that Segment would be a poor (or maybe just costly) fit for device telemetry in most IoT use-cases.


It hasn’t been an issue so far, but we have our own ingestion pipeline on our roadmap that will address this and other issues. (Renat from Kraftful team).


Congrats on the launch! There is a definite need for a tool like this in the IoT space. As a homeowner with well over 80 connected devices, I'm hoping more manufacturers adopt your tool and improve their software!


Thanks Jared! I hope so too. Many IoT teams are still shipping software as if it’s hardware: they develop it once, ship it, and move on to planning the next product. This is particularly true at companies that traditionally made non-connected hardware. Having access to analytics will hopefully make them pay attention to where people struggle with the existing software and improve it.


Hii , I am the founder of a beauty services Startup . Our web platform bring the all Beauty solutions . We have find out quality beauty services is not available in every small towns of India . So we have made a online platform Quickquery to shortout this problems . First we Partnership with small and unbranded salon owners who have potential after we will trained them by our artists team so they develop more beauty skills, work performance and new beauty services . Our artists team provide all services in every location . Artist will be available at marriage places and resorts , there is so many programs , people can get as well as skin suitable service from top artist . We have a training centre for those people who want to build their career in beauty industry , we will trained them according to our brand services so after they can Partnership with us and work as an employee .


The best form of data is usage and third party data, since self-reported, first-hand data can be dirty, chocked full of biases and subjected to opinions. Connected devices are only just beginning to take over and we will see more of them in our spaces - from owned and rented properties to co-living and co-working properties. Excited to see where Kraftful will go!


Yes, there's a difference between what people say that they do and that they will actually do. Similarly, when people are given instructions to test a product, they behave very differently from what real users do. After listening to or reading instructions, users will usually carefully read everything that’s in an app. In real life they just click. That’s why it’s so important to use interaction data :)


Are the analytics only on the application side?

We're an eldercare IoT company and I would say the bulk of our issues are on reliable pairing and maintaining our network mesh. An SDK that I could add and implement either in zigbee2mqtt or directly in our Z-Stack coordinator + router firmware would probably be the most useful to us.


Had a quick look at what you’re building. Looks awesome! Looks like you don’t have an interface for the elderly individuals, so completely appreciate that usability of an app that connects hardware is not your greatest pain point.

This is only for the app side (or display if the hardware has one) because our solution is focused on making the interface user friendly. But there are other solutions for troubleshooting connectivity and related issues. Feel free to ping me at info@kraftful.com and I can put you in touch with folks who work on that.


Fair enough. Thank you for the response and the best of luck to you and your team :beers:


What are you building in the elder care space with IoT?

I am really interested in this space, and would like to see what people are doing.

Thanks


Congrats on the launch! Looking forward to seeing innovation in this space! We just yesterday decided to disconnect our home water boiler from the central programming. Just can’t seem to figure out how to optimize to reduce consumption whilst still having hot water when we need it!


Thanks! As we say, a connected product that isn’t connected…is not a connected device :)

(Heidi from Kraftful)


I'm confused, how would this be different than an IOT app developer adding mixpanel, amplitude, etc and tagging the various events in their app? does the sdk somehow automatically figure out when the device that the user is trying to connect actually connects properly?


Heidi from Kraftful here again. You’re right, it is possible to track all the right events and then send data to Amplitude and configure charts there. The way this is different is that we’ve already done most of that work for hardware companies. So they don’t need to figure out how to track IoT specific events. The SDK doesn’t do that automatically, but our documentation saves them time figuring out how to do it. And our dashboard is pre-configured so that IoT PMs don’t need to spend time configuring charts or figuring out which metrics matter for IoT. Amplitude also doesn’t have IoT specific benchmarks, usability recommendations, and app store review analysis. So this is just a more complete solution for everything an IoT PM needs to improve the experience in the app.


Thanks for responding! I'm not quite sure I follow what the value proposition (over a generic analytics solution) would be then if the SDK doesn't know how to automatically detect the 'interesting' events and I'd have to add them manually, unless I'm missing something? Why is 'figuring out how to' track IoT specific events tricky? And why would 'your documentation' be enough of a differentiator? It seems like the important events would largely be based on various APIs provided Android/iOS, no?

I guess, I'm not really understanding how kraftful would save users time. And those other product analytic services have tons of integrations, AND they could also use those solutions for their landing pages / logged in web experience too. Imagine creating a 'funnel' like

Home screen (referrer=HN) -> product purchased -> app installed -> device connected

where the funnel tracks the funnel of users from HN all the way down to device connection. Does that make sense?

I realize that my concerns are a little pessimistic, sorry about that! I do think that there's something there, but I'm not quite sure I understand the overall value prop. I also noticed your site has something like a no-code 'white label' mobile app product for IOT device manufacturers, that seems really interesting! how does that tie in with the analytics piece?


No worries, appreciate a little skepticism! Figuring out how to track IoT specific events is tricky because there are usually too many cooks in the kitchen (back and forth between product, engineering, data science, and sometimes even marketing teams). This back and forth is often a barrier to using a generic analytics solution. From talking with hundreds of IoT companies, they just don’t use these tools. Some of them try to use Google Analytics because it’s pre-configured, but it doesn’t show them the right metrics for IoT type interactions. That’s why they need pre-configured charts focusing on IoT metrics.

About our no-code UX platform: it was our first take at solving usability of connected products with an experience that improves based on interaction data. That product is live in the field and has for example improved the app store ratings for one of the largest HVAC providers in North America from a brutal 1 star rating to 4.5+. We got to dog food our analytics solution with that platform before productizing it with other beta users to make sure it can actually be used to improve an IoT product.


All the best! and congrats on your launch. I used to work on smart watch idea, that collects data from human and let to control any electronic devices. Happy to share)


This is really clever. There are definitely huge gaps with existing product analytics tools once you veer away from their primary ICP – just ask any B2B PLG company about account-level analytics...

Looks like a great solution for IoT companies!


Thanks! That’s a good point. General purpose analytics tools can’t cover specific needs because that will blow up their product and make it even more difficult to navigate. That’s true when using them in other specific industries as well.

While the IoT industry is pretty broad – home appliances, wearables, scooters, etc. – the fact that there’s physical hardware involved means the apps are used in similar ways: connect the device, control the device, read data from the device, etc. In that context, pre-configured charts and industry benchmarks work really well.


> just ask any B2B PLG company about account-level analytics

Are you saying account-level analytics are hard to come by in Amplitude, Mixpanel, etc?


Yes, especially if you have account-level usage that isn't just a rollup of user usage, e.g. API usage that isn't tied to a user, but to an account that multiple users are members of.


Congrats on the launch! There's definitely a need for this. Rooting for you. Liked the video, charts and explanation- clean, easy to understand.


Thanks for the feedback and support!


Wow this looks like a great and much needed product! The demo shows a really simple and clean UI. Love the charts too.


Really appreciate you picking up on the simplicity of the dashboard. Usability is core to our product (both to help connected products to be usable and to do it with a usable analytics product). We definitely have lots of work ahead before we get there. But it's an exciting challenge :)


Amazing read, really loving points mentioned in the article and also the clean and friendly UI within the demo. I'm excited to see the impact Kraftful makes within the IoT industry and beyond. Congratulations on the launch !


Congratulations to team Kraftful for such a needed solution. All the best!


Heidi from the Kraftful team here. Really appreciate your feedback!


This has been the promise for so many years and yet have not been solved. Rooting for you to solve it. You definately look like you're on the right track!!


Thank you! Appreciate the support. The longer this goes unsolved, the greater the impact as more and more things are becoming connected.


Video is private. Can't watch this.


Congrats on the launch! Your tool is going to make many people's (and companies') lives easier!


Congratulations team Kraftful! I continue to be a fan of your solution and supporter of Kraftful!


Thank you! We appreciate your support! (Heidi from Kraftful team)


> I’m sure some of you are thinking: do hardware companies really need more data?

Right on.


Rant: Sooner or later some person will use all the data to something no good. I predicted some 8 years ago that criminals would start using encryption in the same way military does. It scared me. Encro happened; Criminal networks using encryption to plan murder, smuggle drugs, human trafficking etc. Data has been called the new oil, businesses are more and more finding ways to refine it. Will it be for their good? Yes. Will it be for your good? Just as long as it does not conflict with their good? By the businesses giving away features ("the first one free"); look at the ever better search results or all the long lost friends in social media; you are lured into giving away your data because it helps you. The problem is that our machines and even business models have become so complicated that it is hard for people to know all the logic and data flows. Can anyone without a comp sci University degree discuss strengths and weaknesses in various encryption methods? Journlists, politicians, public opinion? As engineers, programmers, mathematicians, nerds etc we don't see the problem because a) this stuff is paying our bills b) it's just so much fun to use your intellect to conceive new things no one thought about. All the code I've written has been mathematically proven? (Well...) Of course, I have waited for some 20 years for techageddon and it hasn't happend (that I know of?). Can we learn from history? What about mr Murphy, the guy with the law? Why is the debate so just...nowhere?

Well, never mind, back to sleep then, take the pills, leave the kids at school, wait for the paycheck, make another payment on the mortgage. By the way, the kids love the phones, filters are so much fun, facial recognition is soo practical. Why would anyone be intrested in my emails, phonecalls, age, preferred gender, facial patterns, voice patterns, habits, bad habits; I'm no politician or famous person. I bet their phones and mine are protected by some...kind agency. Social media is way better than school. If everyone can find you in your underwear in social media, you are so much not just a weirdo, just like everyone else. Gig economy is freedom. What was the name of that new Russian "rocket" - САТАН, what does that even mean? Why can't they just use 7-bit ASCII? Elon's got it all figured out for us. People usually don't take opportunities to make dishonest money. Don't worry, icy chills may be trained to be ignored. The dating app will help you conceive children of the appropriate IQ. Those paranoid brain cells are just wasting your energy. Life will be so much freer without. We'll blame big pharma later and then some network / streaming service may use it for entertainment.

TLDR: Of course you already know that the spelling of powerful in Swedish is kraftfull. Kraftful is so much more fun. It translates into power ugly


Congrats!! Fantastic!


This is awesome !




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: