>This seems like the buffet or gym model where they want people just mindlessly paying and then not using it that much.

I see this as the point to any subscription model. Of course they want you paying for more than you use/consume.

Right and this is the reason I don’t like subscriptions.

Id rather buy things.

I don’t want a car subscription. I want to buy a car.

I don’t want a book subscription. I want to buy a car.

I don’t a compute subscription. I want to buy compute hours.

Etc etc

Back when Netflix/Prime were the king of paid premium (ad free) streaming essentially being the only players, there were definite months where I barely watched any content and I was a pure source of subsidy for all the other viewers. There are other times, where I swear they (Netflix) start throttling my use to sub-VHS quality. Macroblocking the size of your fist that looks like a 320x240 image scaled to the size of my TV.

>>I want to buy compute hours.

Or a computER.

> Or a computER.

I barely know 'er.

Cars are the one thing I don't want to buy. It's not an asset. It depreciates.

I'd like to pay a reasonable monthly subscription with a small fee per mile once I go over some minimum for my tier.

Once autonomous vehicles are stable, I'm presuming I'll be able to hail a car in 5 or 10 mins via app. I guess the only question is what to do in full-on emergencies. For example, when the zombie apocalypse begins, or the aliens final attack. How do I get out of town?

Isn't that a lease?

Not really. Another issue I have with ownership (or leasing) is usage. While I'm sleeping...the car sits. I'm at work...the car sits. In fact, a whole lot of resources and effort goes into something that spends most of its life under utilized.

Car on demand would mean less cars per person. Perhaps the UAW sees this coming?

Similar use case that actually happens: hurricanes. Entire large cities evacuate. The Miami and Houston metro areas are both over 5 million in population.

Forecasts give days of notice, so I guess an autonomous vehicle ride-hailing service could have a million extra cars drives themselves in overnight. But refueling them would be a challenge. It's already a challenge for normal cars today (without using more fuel by rearranging fleets of cars between cities).

Also, sometimes they change traffic flow for evacuations, like contraflow lane reversal (https://en.wikipedia.org/wiki/Contraflow_lane_reversal) or using the shoulders as extra lanes. I wonder how well autonomous vehicle software handles that.

Food depreciates after you buy it. Do you buy it? Or do you lease food subscriptions.

I use a car that it’s more efficient to buy it. Not because I love buying cars. Cars are just an expense and having a fixed cost is nice.

Id love to live in a city that doesn’t require a car. Maybe one day again.

Only if you're wasteful does food depreciate, but I'd argue that's not the correct word. Food is consumed after you buy it. It was used precisely for what it was intended. That's not depreciation, but I'm guessing you know this and trying to be cute?

My car that I bought was used precisely for what it was intended. And it depreciates through its life until it is no longer useful and its value is zero.

We buy lots of depreciating assets because we need them. That’s not bad, necessarily. But don’t buy a car as an investment.

I brought up food because it’s a depreciating asset. You buy it, you eat it, its value depreciated to zero. It’s not cute, just pointing out that “don’t buy depreciating assets isn’t a very useful life choice.”

Yes. But cars are expensive *and* for the most part they are unutilized. When you're slepping...the car sits. When you're working...the car sits.

An given society that's based on personal transportation (e.g., USA) at any given moment has a significant amount of resources sitting around doing nothing. In a world now based on not-so-infinite resources less cars (via, car on demand) is more than just a biz model. It's good for the world we live in.

Most of the depreciation comes from wear and tear of using it. If you let it sit for an inordinate amount of time that will have a significant impact, otherwise your car isn't going bad when you sleep at night.

Yes and no. Mostly no.

You can live without a car. You can't live without food.

You can take raw food goods. Prepare it. And charge more than the cost of the goods. Restaurants do it all the time. Or even if you prepare it for yourself, you saved money.

I think we can do better that car v food :)

>Cars are the one thing I don't want to buy. It's not an asset. It depreciates.

If you look at a car as an investment, "you're holding it wrong" is the best I can come up with as to how I feel about it. To me, a car is just a really specialized tool. I don't expect my wrenches and socket sets to appreciate in value the longer I hold them. Cars are made by the thousands every day/week/month/year. We don't expect our mobile devices/laptops/desktops to appreciate either. This whole depreciating argument confuses me. Maybe some people just have a hard time equating something with that kind of price tag as simply a tool?

There are very few cars that are investment worthy, and if that's the kind of car you're after, then so be it. But to the 99.99999% of people looking for a car, it is simply something to serve a purpose.

Fair enough. Sell in bulk but have some minimum credits per month requirement. More or less, use'em or lose'em. Not super painful, they are paying customers :) But enough to keep it fair to the vendor. This way there's a path to paying for those who don't want or need a monthly unlimited subscription.

I really liked the suggestion from somewhere else on this topic of paying a set fee for an amount of credits to be charged against. When the balance gets low, allow them to re-up. Of course, there shouldn't be some BS type of expiration date like food products or airline miles. Pay as you type of plans. In that way, you are paying for exactly what you use, and not just donating each month like it's a charity.

In this case expiring is not the same as airline miles. The product doesn't run itself. If you only want to use your credits on the odd numbers months and you expect it to be there and ready, well it still has to be there and maintained on the even numbered months.

I guess you can bake that into the credits price. But if you say, purchased 100 credits today and don't use it for a year...well, who is paying for that year? Adding a small monthly fee, oddly enough, keeps the customer engaged.

Airline miles are a perk.

True, but there's also the reality that maintenance costs increase over time for software projects as upstream vendors change their pricing models and codebases grow in size and complexity. I can see why a product like Kagi wants to keep their traffic tied to recurring revenue instead of selling a bunch of credits upfront.

People would need to buy more credits. I’m not asking for an eternal membership where I pay once and am done.

There would be many customers buying on different schedules and that would aggregate into a predictable cash flow.

This is a solved problem as companies have sold software for 50 years.

Subscriptions are more profitable, that’s why companies do it. Not because there’s any realistic business reason (other than wanting more money for less work).

The incentive for the company is also different depending if they sell subscriptions or credits/usage. Kagi doesn't have competition in the paid search so space so they shouldn't need to go the subscription route yet as a differentiator.

Using dark patterns like subscriptions at this point seems to me that kagi doesn't trust their product or its users.

I think the point is having stable recurring revenue. Giving you unlimited usage means on average you'll be paying more than you use, but, unless you're seriously abusing it, you can technically cost them more than you pay.

Just like you can go to the gym every day - if everyone with the membership did that they would not be able to function. But it doesn't mean you can't.

It’s like “unlimited pto.” It’s a dodge.

Unlimited isn’t real, especially for something like search where it’s not like I need unlimited searches.

The best gyms are pay per session. Gyms are also special because people who buy memberships and never use them subsidize real gym goers who would have to pay more for 20 sessions per month.

The best gyms aren’t pay per session in my experience - and I’ve been to a lot of gyms around the world.

Sure, most gyms let you buy a day pass but compared to the monthly membership it’s very expensive.

My current gym is £40 a month which gives you access to every gym in the country as well) or £10 for a day pass just for that single gym for that single day.

The fanciest gym I’ve ever been to was the equivalent of £30 for a day pass and like £300 for a monthly membership.

Also you example of gyms being special isn’t specific to gyms, that’s also how insurance works, spread the cost out over a lot of people and everyone pays less.

I agree Unlimited PTO is a con though, just give everyone an allocation and let them choose how much they use (most will use the maximum)

I hear you, but it's also really hard from the service provider's perspective because - for many people at least - the allotment of searches starts to become a hinderance on using it.

By making it feel a finite resource, some percentage of the users will start to ration their use of your service and/or do some deliberation before using it ("I kinda want to look that up, but I don't know if I want to spend one of my searches"), and introducing that kind of usage friction can even lead to a subtle resentment of your service.

Yeah, exactly. I looked at Kagi last week (first time I had heard of it), when the $10/month plan was "only" 1,000 searches a month and "unlimited" was $25. While I think that I'll probably stay well within 1,000 searches, I'm also not sure, and it's just not something I want to have to worry about (it's ~33/day on average and I doubt I'll ever hit it, but still...)

One of the reasons I decided to skip Kagi for now.

> it's ~33/day on average and I doubt I'll ever hit it

That's a bit misleading. If the search engine really is that good, you should use it more, and then all of a sudden you'll hit the cap. I probably don't search that much now but that's because it's not worth the trouble.

I just don’t search that much.

And I don’t want to think about how much I search.

Comically, the biggest problem I have with signing up since $10 isn’t that much is that I don’t want to have to log into kagi to prove my account. I search on my phone, work computers, random terminals, etc etc. Having to userid and password to all these places disrupts my current UX of 1) open browser, 2) type search (maybe go to ddg.com first).

Changing to 1) open browser, 2) log into kagi, 3) search increases my time by 100x

I usually use the same browser, but I use incognito or fresh browsers at least once or twice a day.

They have a solution for that, a session link. They pass a token through a query parameter that validates to your account. You set that as the default search engine for each browser, no need to repeatedly sign in.

https://kagi.com/search?token=<some-token>

> If the search engine really is that good, you should use it more, and then all of a sudden you'll hit the cap.

FWIW, I searched less with Kagi, mostly because I needed fewer tries ;)

> If the search engine really is that good, you should use it more

In practice the opposite happens. Because Kagi is presumeably so good, people use it less, because they find stuff faster. Something that would require 2-3 searches on Google is just one search on Kagi.

Anecdotal evidence: https://news.ycombinator.com/item?id=37008132

Wouldn't today's change make you reconsider, because you now no longer have to worry about how much you search for $10/month, it's just unlimited?

I am not the gp but yes. I had the same concern last time I evaluated Kagi and now I expect to become a paying customer this weekend when I have some time to do all my reading and reconfigure my main devices.

> I am not the gp

what is GP?

(sorry, I am new to hackernews and don't yet know all the acronyms)

“grand parent” - it’s a way to reference the comment as opposed to user in a thread.

“OP” aka “Original Poster” is usually the first in a thread.

“PSA” aka “Please See Attached” references a link or attachment to the original post, and is used in the title of the first post, i.e, please look at this webpage I’ve linked to and let’s discuss it.

“TFA” aka “The Featured Article” is often used in the discussions and is the same object as the original poster referenced via PSA

Welcome! And thanks to comprev who has laid it all out above.

Yes, I'll probably check it out and at least consider paying for it, depending on how much I like it (well, once I have a job and some disposable income...) $25/month seemed too much regardless; I don't know the economics behind it but that it's now adjusted to $10/month seems I was right.

It me!

I literally looked at Kagi like a week or two ago from a link here. I really liked it, and I've really been hating Google search more and more each day. But I concluded that 1000 searches felt too limiting, and $25 felt too expensive, so I passed hoping a price change would come at some point in the future.

And voila! Just signed up for $10 unlimited. Probably won't even use 1000/month, but psychologically it just feels so much better.

Same here. Looked at Kagi a while back, thought about paying for it, but 1000 searches per month was way too little, especially considering what actually eats into your quota: load more results? another search. Looking for an image, so you search and then switch to Images? two searches. etc.

But as soon as "unlimited" became something I could reasonably buy, I was in :D

Well you should worry! I've heard that when people pass their search limit, the Kagi team shows up with big pick-up trucks to their front yard and start screaming insults, waking neighbours.

But seriously, I've seen this comment so many times now, that I'm curious: What is it that you worry about exactly?

I didn't post that comment, but I'd suggest it's not necessarily a rational worry at all. It's just human nature that if you tell somebody that something can run out, then some part of their brain will worry about it running out. :)

You can set notifications and hard limits. I had this reservation and 6 weeks in have gone to the $10 plan and have other family members liking not, so will likely upsize again.

I get that it’s hard, but figure it out.

Amazon doesn’t make me subscribe and I can buy 100 hours and use them over 10 years for all they care. I’ve had monthly bills that are a penny from glacier.

I don’t think this pricing is because it’s hard on Kagi. I think it’s a dark pattern that once people subscribe they just autopay forever. My dad subscribed to dial up aol until last year. He hasn’t had a phone line for 15 years.

You think kagi is going to not charge people if they do zero searches?

They’ve already made it a finite resource by charging $5/month for 300 searches. I’m already rationing. They’re just saying it’s $5 if you do 1 or 300. There’s already friction. But friction to dark pattern you into paying more.

I mean it’s their prerogative and they can charge whatever they want.

I just don’t want another monthly fee. I’d rather just pay once and be done. Sell search tokens or something.

Oh, I'm not defending their approach per se - I don't want another subscription at that price point either - I'm just saying that it's not as simple as it appears on the surface because the alternate approach has a direct impact on the behavior of a good chunk of the user base, in a way that's not really good for the business or the users.

I'd argue that in many cases - including this one - the issue with subscriptions is simply that they are overpriced from the customer's perspective. Reasons for this could be greed, a desire to get to a self-sufficient revenue point too early, worrying about the handful of users that will abuse any sort of unlimited plan. From the customer's perspective, if the cost-vs-benefits don't feel right, then they'll complain about a subscription, but it's not /really/ the main problem in most cases.

The most "fair" plan for e.g. Netflix would be if I paid for each thing I watch, but that is effectively a "tax" on usage and negatively affects my usage patterns (from both my perspective and theirs). For example, I'd be super hesitant to browse and try out stuff I might not like. On the flip side, Netflix has now passed 15USD/mo and while it's not a huge chunk of change, I don't feel like I get 15 dollars of value out of it, so I'm thinking of cancelling. If the subscription was an absurdly low 1USD/mo, I wouldn't care about the subscription at all.

Kagi's situation is unfortunate for them because they are trying to get people to pay for something that we normally think of as free, so they probably need to lower the price point, at least for now.

> Amazon doesn’t make me subscribe and I can buy 100 hours and use them over 10 years for all they care

Well, sure, getting a customer's money up front for something intangible that they may never use and whose cost to you trends towards zero anyway is something they're happy to sell you. :)

A simple paygo model with monthly caps at the same prices as the current tiers could work.

If I search five times a month, charge me $0.05 a search. If I search 2000 times, charge me $10. If I search 2001 times, just charge me ten dollars.

Still no need to think about it, no need to lock into a subscription, but you also don’t have to worry about blowing your credit card up if you have a research paper due or something.

Of course, no one does this because you’re right, it’s purely a dark pattern.

.

>By making it feel a finite resource, some percentage of the users will start to ration their use of your service and/or do some deliberation before using it

Exactly to a T how I feel about Khanmigo. I pay for it because I love the idea of a maths tutor in my pocket that wont make me feel stupid, even if by accident, for forgetting something simple.

But there's a "battery" that resets every day and it just makes me anxious. I'd easily pay 50-100$/m instead of 10/m I already pay if I could get unlimited access but there's just no option for that.

If anyone has any recommendations for something like Khanmigo with unlimited access please please let me know. I'd pay so much for a good personal private tutor in my pocket. For maths if that helps.

This is exactly what I did and I am so happy they switched to unlimited. I have historically done things like using my search as a calculator or searching for websites are commonly visit instead of using them as a bookmark. When I would forget to add g! at the end, I feel a pang of anxiety because I knew that I just wasted another one of my kagai searches.

You might want to look at rigging something up with their API. To use their API, you pre-pay for credits – 25$ for 1000 searches via the API.

Thanks. I was looking at this and it might be worth it as seems pretty straightforward.

Their api is slightly different. That is an api into their own index, but their search results are a combination of those results as well as a bunch of outside search results from places like google etc.

There's a reason most non-ad-supported, profitable SaaS businesses have switched to a subscription.

I don’t even make 10,000 searches in a year. I just did a search of my browser history: 8881 visits to DDG in the past 12 months, including many duplicates where I hit the back button to click a different link on the same SERP.

How much do you think it would be reasonable for them to charge for 10,000 searches? I assume this fee is supposed to sustain the company and fund all development and growth with a 100% ad-free business model, forever!

GP already said $5 per 300 searches. 10,000 would be $166.67 at that rate.

Given that they allow unlimited for $10 per month, that would mean it's the equivalent revenue of 16.67 months, which seems pretty comparable to your numbers.

I think you're being unreasonable toward GP.

I suggest that it might be different: a lot of kagi users - of which I am one - complain about feeling internal pressure to optimize their number of searches, which leads to higher mental cost. The unlimited model is nicely brainless, which I mean in a good way: you just use it however you want, and as long as you're getting>= $10 of value from it, you keep doing so, instead of having this per-operation mental calculus of "should I preface this with !g to save a search?"

I have this same unavoidable urge to optimize with other limited plans, such as cell phone plans (for voice , before everyone went unlimited anyway) and have found I'm much happier with an unlimited plan. It costs more and I value the saved mental energy more than the trivial amount of money.

That was basically their previous model. As a software developer I blew threw several hundred queries per day ... $10 for unlimited searches is the point I was waiting for to make the change.

It's not what you're asking for, but FYI, you can sign up for a month or a year and immediately cancel your subscription very easily on the Kagi site. You'll still have access until the end of the subscription period.

You probably just killed this "feature" since kagi employees are monitoring this forum :)

We do read the comments, but as far as I'm concerned this is a feature, not a bug :)

FWIW, I don’t sign up for any recurring subscriptions outside of major utilities, so this feature won $108 from me.

I understand not wanting another subscription to pay for, but I just can't agree this is something they should do.

https://blog.kagi.com/status-update-first-three-months#finan...

This was a comment about their financials last year. This isn't a high profit business making hand over fist and exploiting users.

I'm happy to pay monthly to get a service I see as valuable, especially if it ensures that this service is sustainable long-term and doesn't disappear because of money.

Subscription based services are horrible user experience and it's a show of bad faith from the vendor, IMHO. You want me to be locked into your service, wen lwell, then I don't trust you or your product.

This whole post feels dirty, how it's a direct link to a blog post that just reads like a press release, and there's a ton of sus responses promoting it in this thread.

Tin foil hat, are we having a reddit bot moment on HN?

What is a good way to charge for an ongoing service without a subscription?

Depends on the nature of the service - for dns, vpn and so on - thing you only really need one of - once chosen it makes sense - nextdns is 1$/month, backblaze 10 and i feel fine for subscribing. And not locked in in any way. I don't really need second one of any of those.

It doesn't make sense when your consumption is spread between many vendors - like news sites or streaming services or substacks. There should be option - done one time to someone on substack. There are authors I would like to send some money, but they just don't create enough value.

search engines and gpt chat bots are something in the middle - you usually have 1 + 1/2 backup. Goodle, then bing (google filter way too much some stuff), then yandex (since there torrents are not nuked). In this spot subscribing for each is too much money, but just limiting for one has real quality of life costs. Kagi seems to be somewhere here. I wish them luck - search space begs for disruption, but my hunch is that 10$/month is on the high side. On the other hand having those kind of people is valuable. A search engine for the rich will have easier time pitching value adding service.