(Leaflet cofounder here) - much appreciated; we spent a lot of time iterating building different versions of apps before Leaflet that were more explicitly around shared social spaces and communities for learning and collaborative projects, which informed some of our aims in making Leaflets both simple docs and shared spaces (lots we can improve there e.g. bringing in live presence features!)
For the RSVP block, yes we're using Twilio - right now it automatically uses SMS for US/Canada numbers and WhatsApp for int'l, bit more cost effective…I think it wasn't a huge lift technically, though took longer than expected mainly b/c we had to wait a while / re-submit some documentation etc. for approval.
I wasn't the one doing the back and forth on this directly but I think it was at least a few weeks, lots of threads with support etc. Partly on us for something like business address records not matching in different places…we get it b/c I'm sure they're dealing with a ton of spam mitigation but the process could definitely have been clearer / more startup friendly.
Nice! I like the look of the resulting map. I've visualized gentoo packages and their dependencies before [1], but the result is no where near as aesthetic as this.
I find it incredibly frustrating if something breaks and my experience from attempting to repair it is that it's impossible to disassemble. I'm currently sitting with a broken coffee grinder that I can take apart with a brick :(
I frequently dream of openly licensed appliances that are repairable!
I've wanted to do Nand2Tetris for a very long time, but it just never reaches the top of the pile of "things I'd like to do".
Many years ago I built a linux system following the Linux From Scratch (LFS) guide. My experience doing it corresponds with a lot of what the author mentions.
In all this there is a slightly uncomfortable realization, at some point you need to trust the layers below, knowing everything is simply not possible.
This makes sense, if someone isn't using your service for a month, chances are good that they are going to cancel soon. Maybe they'll keep on paying for another few months, but if they're not using it, they're not getting any value from it.
So rather than getting them to cancel, pause their subscription. You don't have to deal with cancellations, and if/when the user does return, you are one step further than you would be with a new subscription.
Furthermore this generates goodwill, and I'm guessing goodwill has some % that converts to conversions and lower churn.
> More importantly you can tell investors you have even more accounts than you do and your churn rate is very low.
I know you're being a little facetious but it is actually a benefit. Many companies have implemented subscription pausing to reduce churn. The reason is pretty straightforward good business: it's easier to reactivate customers who lay dormant for 1 or 2 months than it is to let them churn and have to re-sell the product to them from scratch.
I've had places offer to just not bill for a few months while still allowing full access or offer some steep discounts in the workflow for cancelling in attempt to reduce churn.
If you ever dealt with the business side of SaaS you know how important churn is.
Growth = New Customers Acquisition - Churn. New customers are expensive for many businesses to get, they have marketing, sales, and promotion related expenses. It makes sense to spend money to reduce churn too, because it’s a cheaper way to boost your growth rate.
If you offer deals to reduce churn, you need to focus on if those deals are just delaying inevitable churn or if they are actually winning back customers. Delaying churn is just a game of spending money to make your books look better for a quarter.
Yes but it’s also a chance to convince the customer that you’re worth keeping around. Depending on your product, you might also have customers that dip in and out. Keeping them from churning, while sacrificing a little cash today, helps keep that customer from shopping around for alternatives, too.
Also, in my experiences most of those services which offered free time or heavily discounted alternatives to cancelling tend to be things which have a pretty low cost per subscriber (e.g., news sites, simple applications, not much per-user storage).
I agree with your statement but your general calculation of growth is off imho, it should be:
Growth = (New Customers Acquisition + Install Base Growth) - Churn
This also assumes that churn = Customers that leave completely and those that downgrade plans
IB growth can fuel business growth, especially during times of low new customer acquisition and or high to moderate churn
Fair enough, but as far as your investors are concerned you're changing the definition of churn. I'm sure if they ask, they'll be provided with "pause" metrics, but such data will never see the light of day in any marketing materials.
Correct, your churn would be artificially lowered, so you'd probably want to define a point where a "paused" subscription is effectively inactive and count that against churn.
It would still show in other metrics, however, as you'd have monthly active users (or accounts), which would take a hit. You'd also see a drop in MRR when an account is paused.
Bear in mind I'm assuming a business that wants clear, accurate metrics so the executive team know what actions to take; not simply a business looking to scam investors out of money ;-)
A thoughtful investor will be more interested in paid active users and cohort analysis of retention (in terms of actual usage) rather than subscription numbers/revenue in absolute terms.
Engagement can always be monetized better in the future.
And they don’t usually ask for number of subscribers, but number of paid subscribers. Startups have played enough games with this that investors should be able to hone in on the right numbers.
Plus, you get to stay in touch and advertise via a monthly email: “You didn’t do any searches so we’re giving you next month for free, here are all the cool things you could do:…”.
You can do that, but if they don’t interact with your email either, you may be better off not doing so. You get worse delivery rates if you constantly send email that gets ignored.
They send by default an email every month when the payment time is approaching, so I assume they may just send the same email and state that the payment amount is covered by the previous month. They have really good practices at that.
Yet Kagi is one of the very very few services I've seen that sends a monthly reminder that they are ABOUT to charge you... And you have time to stop it. Typically I only ever see something like that with shipped goods where they might have to deal with a return + cancellation
That's really nice sounding and comforting. I've been a little on the fence, but Google has become such absolute garbage lately that I've had to frequently use Bing in order to find something that should be one of the first results. This may just be a better model.
People also forget that they're subscribed to something (or that they signed up for emails either explicitly or implicitly), so they just report to Google (or whoever) as spam. I'm sure that's why I get a lot of email (mostly in my Gmail Promotions or Updates tabs that are from companies that scanned me at some event or whatever or I ordered something from.
A huge benefit to companies with subscription services is that people forget they have them and keep paying for Ancestry.com or whatever for months (what? no, of course that never happened to me...) after they've stopped using it. Kagi is voluntarily giving up that benefit. This just seems like a consumer-friendly move to me, not sinister at all.
I somehow ended up with two simultaneous Audible subscriptions in different regions. It seemed like a coin toss which one the website would send me to. I only found out because I canceled one of them, and then still got billed. Support was very understanding and refunded the double-sub period.
Ans any difference in pricing is made up for it by increasing rates to cover the lost revenue that wasn't automatic. And people are happy they don't pay in non use months but the company still makes the same momey.
I've seen the opposite of this at a startup once in the streaming music industry. We _knew_ our "free trial for a month" was being abused by 500k+ users because the sign-up flow neither required a valid payment on file or email validation. In fact there was an entire industry built around generating false throw-away accounts.
The company used stats including non-paying users to demonstrate demand for our service was high, even though we knew they would highly likely never spend a cent with us.
If the streaming services did this, I'd probably have pretty much all of them. Then, instead of paying monthly, you essentially have a tab open with everyone, and you pay for whatever you stream.
Indeed, this would make me way less annoyed at the thousand and one streaming services popping up like mushrooms after a rainy day.
I... really like this idea. It's an interesting problem and something that a challenger service could possibly use (assuming they can resolve potential cash flow issues around content licensing). From an incumbents' perspective, it's less desirable since the fact I already have Netflix, Crunchyroll, Amazon Prime, Apple TV makes me less inclined to add Disney+ or a new service to my list of monthly outgoings.
There's a separate cash flow problem on doing this, a big chuck of the expected revenue comes from people that setup the subscription and forget about it
I tend to care lot less on keeping something like dropout even if I don't use it all the time (I like to think I contribute keeping it afloat, and watch it whenever), but I cancel other subscriptions a lot more aggresively (I've unsubbed/resubbed to Gamepass plenty of times specially when playing random stuff with friends, there's something nice of the exploratory playing you do when you don't need to care who has bought which game)
On the pattern I use dropout, they'd get a month or two of revenue out of me (Binge a couple of their limited runs and catch up on their staples) and zilch for the rest of the year even if I'm a happy customer
It's an interesting point, because it gets to when streaming services' content costs are paid.
Are they upfront? Or do they pay per view royalties at the end of a streaming period? Or likely mix of both, varying by each piece of content terms?
Hypothetically, if a streaming service structured most of their obligations in the form of post-view true-up's, they wouldn't have any problem doing this. And could make bank on the float between customer payment (first of month) and paying for their content (end of period).
For movies and series, there's a process called "Minimum Commit" where the media companies have to commit to spend $X for a bundle of content that includes both the content you want to sell and a random selection of crap to pad out the offering. Then it depends on the deal, but often the amount that each item is watched then contributes to the royalties cost. The relative exclusivity you might want will then dictate the overall cost and royalties returned.
For sports it's different, typically you bid for the right to own a geographic market and the games are sometimes split into bundles where you can bid for one or all bundles depending on how deep your pockets are. You'll then get to keep exclusive or non-exclusive rights for a certain number of years. You'll then pay annually/quarterly for that right at the total bid package for the term of the deal (e.g. 5 years). Depending on the contract you might be able to re-sell that right to other companies as well, which dilutes your audience, but may increase the distribution overall.
Then aside from the rights, you end up paying infrastructure costs both fixed and variable. You also generally commit to CDN capacity for distribution based on a forecast of how much you think your customers will watch in any quarter.
Streaming services mostly pay for licensed content up front. That’s why video streaming is much more like a traditional tech company - high up front costs with very low marginal costs compared to streaming music companies where their costs stream linearly with their revenues
Pay for what you use streaming exists. I don't think it would work here. Paying a large sum and dividing it up by what you watch is basically back to cable tv, just with a little smarter more immediate billing -> analytics feedback for hollywood.
A change I think is necessary for consumers however is deduplication of content payments. If you subscribe to multiple services, youre paying for a license to some content multiple times, sometimes many times.
What I would like to see is more like Kagi Fair Pricing, a master payment account (like prime or movies anywhere) that has access to all your accounts, cross references where you are paying for a title multiple times, and offers a refund or credit.
The problems with per-watch pricing most come down to requiring a conscious decision each time they use your service, when you'd prefer that they be indifferent at most margins so that they don't have to be regularly reminded that oh, yeah, the amount of money they're spending on this darn this thing actually varies with how they use it.
Largely this breaks down into two salient factors:
- the friction of the transaction itself, which you largely shed when the consumer already has already agreed to be billed on usage, and
- metering aversion, which can be alleviated with a wide range of cheap tricks, e.g. using very coarse quantization: think not "rent this episode for just $0.99", but "rent up to 50 episodes this month for just $9.99". But the extreme of this is what you actually see: one price for any usage of the service at all, which ... is a popular pricing model at consumer scales because it works?
Give it a reasonable enough price (0.99 per episode is too expensive) and add a monthly cap (can be a bit higher than the current subscription) and you've got me as a customer.
I make the streaming services work like this already. I have neither the time nor the interest in watching anything on streaming most months. I have all my subscriptions cancelled all the time. When I want to watch something badly enough on service x, I sign in, re-activate, get charged for one month, then immediately cancel the subscription. Then watch the thing. Then not get charged again until I want to watch something else badly enough in another 6 months' time.
I do this when we have visiting relatives who would spend all day watching a particular cable news channel that gets them all riled up. Before they arrive, I subscribe to a different streaming service with a package that doesn’t include that channel and put its icon front and center on our TV. I then immediately cancel it so it doesn’t renew.
I do something similar and that's why I usually try to subscribe on my phone or iPad. iOS makes it so that I can start or stop a subscription in about 3 seconds. Sometimes it costs a little more because the services have to pay Apple, but for me, it's worth it.
Please tell me that you have at least a Bash script or Selenium workflow for this. It actually sounds like a fun project to abstract and make pluggable. It would probably require maintenance, though.
Yes a lot of people does this manually. But the services predate on the people that is too busy or lazy to unsubscribe and I'm afraid they are a much larger group.
A big problem with micropayments is that the transaction costs tend to dwarf the actual payment, which isn't good for the buyer or seller. I don't think it is an unsolveable problem, but there are significant network effects that would need to be overcome.
AFAIK it never pans out really. People turn out very stingy if they're faced with a decision to pay or not to pay for every article, so the revenues end up a lot lower than what the subscription model would pay.
When people are confronted with the actual cost, they tend to say no.
With a subscription, their head tells them that for 10 €/$ they get an infinite number of articles.
No, they get the articles that _you_ provide. But if _you_ provide only 50% of the interesting articles, as does every other provider, then approaching the ability to access 100% of interesting articles get very expensive. Just getting to 90% of the articles would cost 40€/$. And pushing that to 99% will cost 70€/$.
You are really underestimating how many users just forget they have such unused subscriptions, and how much of subscription based company monthly revenue is those that are not used at all.
I'd prefer a pay-as-you-go / per API call / search pricing model... to something that if I use it just once, I pay full price for a month. Same rationale for AI in my IDE, I'm waiting for the pricing models to change
The downside is that price per unit will increase as minimum unit size goes down. You're essentially buying at a "retail" price whereas the monthly bundle is wholesale.
But the tradeoff would be worth it for sparsely used applications.
Plugins will never match the possibilities of a deep integration. Those who build or fork VS code want the monthly fee because they make way more per customer
I dunno. The thing is, Kagi isn’t really that much better than Google. When they still had a free tier, I tried it every once in a while, and it quickly wastes a lot of searches even while just entering queries, and then the chance to find something better than Google is mediocre. Perhaps a prepaid model might make more sense, especially if it’s designed not to blow through queries quickly and transparent about how many searches were actually done.
Compare to ChatGPT, which is much more expensive, but the value relative to Google is pretty obvious.
I have the opposite experience: I use Kagi a hundred times a day with always relevant results while the GPTs always hallucinate random crap. I guess it depends on how you search.
I don't doubt you but this life experience is so far from my own I struggle to understand what you use that much search volume for. I maybe search for 4-5 things a day (based on my one stint paying for Kagi and their usage reporting and trying to use it everywhere, and this was before AI products were able to search on your behalf) which is what led me to cancel, I was usually not getting the paid plan value from it. A large amount of my searches today are often just fancy autocompletes for specific URLS on already known domains that I probably could have accessed without a search engine at all.
Do you pay for ChatGPT with built in web search? I’ve been paying for ChatGPT for two years. I just started using the ChatGPT extension for Chrome for search and it is so much better than Chrome.
My experience is completely different - I get much better results from Kagi. And one of the things I really like is the ability to entirely block domains, so for example I never get any Pinterest links cluttering up the results the now. I also love the fact that you can enter a ? at the end of a query and it'll give you an AI-generated summary at the top of the results. That's a great shortcut.
> And one of the things I really like is the ability to entirely block domains, so for example I never get any Pinterest links cluttering up the results the now
Note you can do this on Google using the uBlacklist extension [1]. You can select domains but also use patterns to match specific URLs, like `somedomain.com/someprefix/*`.
I've been a paying Kagi user since the beta - and thats because I get good value from it. Out of the box the search results were, and are, much better than Google or Bing. The ability to raise, lower, or block the priority of sites adds to that and gives me a very personalised tool. I'm very happy that it exists.
On the other hand, the fact that we're having this discussion does point to how difficult it is for Kagi to explain its value proposition and differentiate itself from Google.
As for chatgpt - I'd say its functionality relative to google search is obvious, but not it's value.
I'm finding Kagi gives me relevant results much more readily than Google, where I have to wade through all those sites which take technical content from other sites and repost it for ad revenue. I'm on the lower tier plan and haven't hit the monthly search limit yet... but I'll consider upgrading if I do, because wow it's so much better for me.
It does seem likely though that it's not going to be better for absolutely everyone, other than in terms of having their business model being "give good search results" rather than "give people adverts we can charge advertisers for".
According to my usage statistics, I use Kagi around 20-50 times a day.
Date (UTC) AI Tokens Searches
Feb 5, 2025 0 64
Feb 4, 2025 0 43
Feb 3, 2025 0 19
Feb 2, 2025 0 24
Feb 1, 2025 0 19
They don't seem to track any form of history, only the number of searches (since some of their plans have a quota). I pay for unlimited searches, but the stats are still interesting :)
Similar stats for me. It’s become an invaluable tool, sometimes I’ll use another browser that’s has Google as default and immediately notice how much worse it is — all the ads, irrelevant cards, etc. Kagi is like the way Google was 10 years ago, which is MUCH better… with the benefit of more personalization
you can get that on DuckDuckGo. The main problem with Google is that the search is garbage. Kagi wasn't able to convince me that their better within the free searches (I have an account since 2022). Now that I can't try them anymore, they can't ever convince me they're better - so their pricing model perhaps isn't very smart.
People see no tracking and just trust it nowadays? I'd much rather use a public SearX/NG instance than to trust something that claims to have no tracking and isn't open source. Same thing with DuckDuckGo.
Oh yeah, in this political climate I'm definitely going to voluntarily tie my and my children's search results to my credit card. As long as people continue to gush about how amazing this service is, I'm going to gush about how ridiculuous this proposition is.
yes I do think there is a important difference between google triangulating data, trading data with others and attaching a name to an ip adress by their own efforts without me voluntarily giving them that information for free. And you seem to forget that Google lost a class action suit about incognito mode. And I'd rather sue Google than Kagi.
Plus, like 23andme, when times are tough I don't want to think about what a smaller company in dire straits will do with my dafa.
Today, Kagi has a negative incentive to even historically track user search data (if discovered, their business would be cooked). Consequently, it's very likely they're being honest and don't.
Furthermore, they're building a sustainable business around subscription revenue.
In the event any of the above changes, they still won't have any historical data to share.
As opposed to Google, who keeps things in their vaults until the heat death of the universe.
> And I'd rather sue Google than Kagi.
Ha! You and what European data authority supporting you? Because that's the only way you'd have a chance of making headway.
Thank you for agreeing with me. Why would I bother using a VC-backed search engine today that forces me to login to use it routinely only to receive an email later saying, "An Update to our Terms of Service". And whose only way to convince me that they do not store my data is to tell me that I can "trust them." Even if I trusted them, I wouldn't trust their investors or their random late stage C suits.
>As opposed to Google
Are you willfully ignoring what I wrote in bad faith? Google had to settle a class action law suit that forced them to delete "billions of user records" and still allowed them get sued for individual claims down the road. Use kagi to search for the winston strawn summary of the case.
Here is an excercise: Open a three letter browser starting with the letter T, go to google.com and search for the life expectancy of ALS. Now close the browser.
Now tell me what google can deduce about about the real-life ethbrl with certainty and how they came by that information.
Are you hitting "New Identity" in the Tor browser, removing all cookies/sessions and creating a new circuit for each search?
In that case I guess there is not too much they can deduce aside from the type of device (desktop, mobile).
But of course, if you make more search queries without hitting "New Identity", they can piece together a lot more than that, including exactly who you are with enough time between new identities.
If you're going so far, you can use Kagi from Tor as well. There is even a Hidden Service for it [1], so you don't even need to hit the clear web at any point.
If you're concerned about tying your credit card information to your searches, you can just use a prepaid debit card or crypto to pay [2].
>If you're going so far, you can use Kagi from Tor as well.
I have to remind you we're talking about preventing Kagi or Google from tracking you. This suggestion makes no sense when you're forced to sign-in to Kagi to use it meaningfully as your default search engine anyway no matter where you're connecting from.
Your first two paragraphs describe a use case that is way more convenient than your last paragraph, and most crypto wallets have most likely come into contact with exchanges that have the user's kyc data to begin with.
Again, you seem to be missing the point here. Those "billions of user records" are the users who thought they were not being tracked by using incognito mode. All the other users who didn't care about being tracked one way or the other are irrelevant to the use case we're discussing.
Your point seems to be arguing in favor of how a company whose core business product is tracking... makes you more comfortable you're not being tracked than an alternative with a subscription model?
>Your point seems to be arguing in favor of how a company whose core business product is tracking
I'm doing no such thing. I'm merely pointing out that a user still has tools in their disposal to prevent a company whose core business is tracking from tracking them as long as said company does not require the user to sign in with PII info.
When you sign in with Kagi, your only protection is to "trust them". Kagi's next move should be to allow mail-in-cash for account activation to back up their privacy intent if they require user sign-in, like some other privacy-focused services allow.
> there is a important difference between google triangulating data, trading data with others and attaching a name to an ip adress by their own efforts without me voluntarily giving them that information for free.
You’ve specified the difference. One company is actively trading your data as its core business, for profit. One isn’t.
I find your position baffling.
No, one says it isn't at some specific point in time. Some people here seem to want to believe the last decade of bait amd switch VC backed startups never happened (often times through no fault of the founders).
>one is trading your data
As I mentiomed in my other comment, the user has tools at their disposal to prevent google figuring out its "your" data. No such tools exist when you're forced to sign into Kagi with your credit card.
Your position is baffling. You think that Google, who maintains every search you’ve ever made, including correlating them across every Gmail account you have, and who routinely provides this search data to authorities, as well as sells access to this data is somehow more safe to use than a company based in Europe, who are funded off of a subscription model, whom are not VC funded even though you claim that, and who’s entire sales strategy is that they don’t sell your data or even retain it.
Nothing in the comment you're responding to says anything about me using Google search or Gmail. In other comments I'm simple comparing the use case of using google search with an obfuscated connection and without ever signing into a google account with the use case of having to sign into Kagi. In that respect, I have no idea what you're talking about. If you're responding to another comment of mine please respond to that comment so I can better understand your point.
Nothing in any of your comments indicates that you are using an obfuscated connection to search Google at all. In that case there is little difference to using Google signed in or not, you are still trackable across numerous devices and your searches are correlated. So being signed into kagi has little difference besides them now being less incentivized to sell your information or track you in any manner.
Kagi is entirely dependent on giving the best search. Without it they would lose pretty much all customers.
"Privacy minded" customers is not a foundation for a business. They spend all their time complaining and accusing, and then after some time they cancel their subscription because spending $10 per month keeps them awake all night.
There are lots of benefits for sure, but you have to weigh them against the users who can't be arsed to cancel their subscription and keep on paying. You'll miss out on those.
I think this is maybe a bit simplistic. People forgetting to cancel means the price can be lower per-person. It's a little bit like how insurance is priced.
There is no reason for number 10, it could very well be 15 or 20. O just chose a number instead of saying multiple of many.
It stops once someone picks up a call. I will put that info in the landing page.
Ahah! I was expecting this question. Indeed, the interest is lower for me (but not null!). I mean that I based it on my own way of thinking, and I don’t think I’m the only one.
But after all, my wife and I still have a goal just for fun! And I’ve already charged myself! In the end, I still pay the Stripe fees! :D
Also interesting to see that you chose WhatsApp for the RSVP element. Did you use Twilio for the integration and what was the level of effort needed?
reply