Beat me to it!

Kagi also goes one step further and allows you to "pin" sites to the top. For example, I've got MDN pinned, so whenever I'm searching for web stuff they're the top result, even if there's an SEO'd blog post that normally would have come first.

Abandoning Google is a huge motivator for me, but this feature set is why it's my primary search engine. Google tries to guess what I want and just ends up feeding me the same garbage it feeds everyone else. Kagi allows me to correct it when it guesses wrong. That makes all the difference.

I used to just download the MDN docs to my computer so I can search locally. Offline search beats everything. You don't need the Internet to search.

Huh, $10 a month is pretty steep. It's great that they offer a free plan, but that comes with all the misaligned incentives again. Any reason they don't just do pay-per-use (1 cent per query)?

Pay per use is a great model for search and I wish we could use it. But I don”t think the world is ready yet.

In our survey 90% of users told us they preffered fixed fee over pay per use and feedback we got was that pay per use would make them anxious to use search. Also it adds additional friction in the signup flow (where the idea of paid search is already a novelty and then pay per search?) and so we decided to go with a fixed monthly price.

Sweet spot would be $15-$20/month but this way we would not have enough users, and less users equals leas feedback to build product. Our pricing is subject to a change, we had to launch with a price and we’ve chosen one that was good compromise.

We are likely to introduce pay per use first in our enterprise plan. Pricing Kagi is an extremely difficult intelectual challenege. (Kagi founder here)

Maybe you could do something similar to Google Fi's pricing model for data usage. You pick a monthly amount of data, say 2GB, for a monthly price of $40. Any GB you use in excess of that is $10 per GB up to a maximum of $80, and data usage is unlimited after that.

I would bet that my usage most months would not exceed $5, and I think many other prospective users would be willing to gamble on that as well. Food for thought.

Too bad a pay-per-search model was not adopted. But would it be a big trouble to let users choose?

* Say 1 search is 5 cents. So unless a person searched less than 200 times a month, the pay-per-search is cheaper....

* Or why not adopt the model where you pay 5 cents per search _for the first 200 searched_ and nothing afterwards (so the ceiling is $10 per month)?

Relevant Tweet: https://twitter.com/janhodl3/status/1535530318987509761

I don't think charging per query would work for most folks, for the same reason micropayments to bloggers or what have you won't work: it discourages use. If you know that every time you hit enter on a search query will cost you something, anything, you'll hesitate. You might choose to just use Google and use your data to pay for your search instead.

But that's kind of the point: It always costs Kagi that amount when you do a search (according to their pricing page). If the relationship between user and Kagi is not supposed to be adversarial, then indeed the "price vs value" tradeoff needs to be resolved on the user side.

At the moment, I'm either overpaying (because I perform less than $10 worth of queries per month), or the company is losing money on me. And with the existence of the free tier, the business model can only work if most paying users are effectively overcharged significantly. Right now they are operating at a loss in both tiers, if their pricing page is to be believed.

One would hope that costs amortize better with more users (e.g. scraping is pretty much fixed cost regardless of the number of users, but maybe that's already negligible) to push the price low enough for pay-per-use to not feel spendy. (When did you last think about how much one toilet flush costs you?)

Scraping and building their index probably costs way more than querying it. The way that db would scale is very friendly to replication (read your own writes isn’t anywhere near required for example) so the number of queries (times cost per query) needed to match the indexing costs is probably very, very high. I bet the 10$/month cost is meant to cover scraping and indexing costs, not the queries.

> I'm either overpaying

The company is aiming to have many users that everyone is overpaying with their $10, so that they make money, thanks to the reduced marginal cost. And the company hopes that the $10 is low enough that enough everyone knows they are overpaying they are still willing.

Yeah, $10 is steep, but I feel like I'd be happy with a bundle deal. Search, reliable email, a small bit of storage, and other small services for $20 or $30. I'm sure I'm underestimating how much I use search, but it just doesn't feel like an essential part of my life that I'd want to pay that much.

I'd honestly rather see Kagi focus entirely on search and not try to branch out too much. These days, I think startups try to chew up too many markets at once instead of really honing in on one.

100% agree. Maybe I'm missing something but where's the synergy between search and email + storage unless you're harvesting data?

The synergy is in the fact that email and storage are high margin products. We are currently basically losing money on search. Cost of providing email per user is negligible (compared to search) but you could double the price and make the economics work.

From their usage panel, I do 50-200 research per day. In 10 days my usage cost is estimated at 11.69$. It look like each query cost 1.25$ to Kagi. I don't want to be conscientious of the cost of my search usage, I fear it will inconsciously reduce my search usage and access to knowledge.

> It look like each query cost 1.25$ to Kagi.

Correction: they're saying 80 searches cost them ~1$.

>Why does Kagi cost $10/month?

>Our proposed price is dictated by the fact that search has a non zero cost. With other search engines, advertisers cover this cost. But it costs us about $1 to process 80 searches.

>Someone searching 8 times a day would perform about 240 searches a month, costing us $3. An average Kagi beta user is actually searching about 30 times a day. At $10/month, the price does not even cover our cost for average use, and we are basically betting that average use will go down a bit with time because during beta people may be searching more than normal due to testing etc.

>Our goal is to find the minimum price at which we can sustain the business. If it turns out that we have more room we will decrease it. But it can also be that we may need to increase it.

>The free plan will be limited to 50 searches a month (and this too has to be paid by paying customers which makes the above math even harder).

How... do 80 searches cost them a dollar? That seems insanely high unless they're counting fixed costs that'll go down fast (on a per-search basis) as they get more subscribers.

8,000 searches costing a dollar, in actual resource use? OK, maybe. Still seems a little high, but maybe. 80? Are they paying someone to manually look things up for you?

I think I remember somewhere that they said a very high percentage of searches are totally unique i.e. never queried before thus not served from cache. I don't think they reword searches like Google does for a higher cache hit rate.

I never thought about that. That could explain a lot. Although I also recall Google themselves saying a lot of their queries are totally unique anyway.

> Although I also recall Google themselves saying a lot of their queries are totally unique anyway.

Which is probably why search quality is going down. They're rewriting your query to a more common way of saying the same thing, at least according to Google.

Perhaps amortizing really high salaries. $1M/month for a chief metrics officer or something.

Kagi is completely bootstrapped. It has basically 10 developers and me doing everything else. No managers. The expense is low as humanely possible as still coming out of my own pocket.

I admire kagi a lot and will eventually pay you.

You don’t have to answer, but how do you have such a high per search price?

It is the cost of commercial APIs that we use. We can't get volume discounts because we dont have volume, not are ever likely to have volume similar to free search engines.

I'd challenge that the cost per search is actually very low. For just 1.25 cents, you can search the entire world wide web with any query, find what you need with no distractions and all that in just 500ms. The value of that search outcome for you can be $10 or even $100. It is probably the most amazing return on investment existing in the world today. Imagine for a second that Google and other search engine didn't exist, and someone offered you this deal today. I'd be you'd take it in a hearbeat. Heck you may even consider paying $1 per search or even $5. So the value is there.

But today this value is anchored in the price of 'free' search. So instead $1 per search sounding like a good deal, we struggle to accept that 1.25 cents is.

We hope that for a portion of people that 1.25 cents per search is going to sound like an amazing deal, especially as this is the only search relationship where the incentives of the information provider are aligned with the incentives of the information consumer.

While I agree that it probably have a lot of value, the issues is human brains.

I don't feel 1 cents per search is a good deal at all, while -logically- it probably will give me more than 1 cent of value out of it. People never thought about how much the knowledge they acquired is worth, most of the time they share it free of charge, people like to share, somes will even pay hosting to share knowledge !

That's why I don't believe in the pay-as-you-go I'm fine with a subscription, I don't pay to have more value out of my search, but because: 1. I dont like ads, I want to support an ad free alternative.

2. I pay for my privacy.

3. Better results will mean less time lost on searching, less frustration.

I never accounted the value I will get out of better search, I looked at the price and felt that I was a bit overpriced for my taste, but I want an ad-free, pro privacy internet.

I didn’t know Kagi was bootstrapped. As a fellow bootstrapper, it makes me like Kagi even more. Good luck in the future. I think you really have something here so I hope the monetization strategy works.

It's new software, features are always prioritized over cost efficency at the beginning when pressure to ship product overrides all.

From Kagi usage panel:

Date-search count-generated cost

>Jun 10, 2022 90 $1.12

They show conflicting numbers.

> I don't want to be conscientious of the cost of my search usage, I fear it will inconsciously reduce my search usage and access to knowledge.

Sure, but this whole adventure won't last very long if the company loses money even on paying customers. If your usage costs them about $30 a month but you only pay $10, who will pay the remaining $20? Someone has to finance your access to knowledge in the end...

Correct, we are betting that avg user will cost us less than $10 in the future. Our current userbase is skewing towards HN - heavy usage. If that does not happen, we will have to change the price. Cost per search is unlikely to (significantly) change without (significantly) jeopardizing the quality of results we are known for.

Most of my queries through the day aim a few hundred of website - programming docs- I wouldn't mind if Kagi automatically chose my lenses(current one are too limited) if it reduce research cost.

No, it's industry standard to operate at a deficit to gain userbase and subscriptions. They're certainly prioritizing shipping product right now over reducing COGS, but you can bet that if they're successful in the short-term that in a couple years they'll be able to significantly reduce cost per search.

A lot of services operate like that. France ISP offer unlimited data with very low price, because they know the average usage, and I'm perfectly fine with that.

You'll get used to it. Of course you won't limit your access to knowledge just to save $20 monthly.

How much does it cost to have a suboptimal search experience 50-200 times every day? Saving 5 seconds (on average) per search, that's something like $10 per day in savings (provided you search during work hours).

Sorry, how did you find that usage panel? I'm on my phone and I'm just not seeing it.

Nevermind. Looks like you have to have a subscription to see usage stats.

As someone who prefers to search in private tabs, I was wondering why do I need to create an account, until I saw the pricing. It's an interesting conundrum, either you search anonymously with bloat and ads, or have your activity pinned to your account maybe with ads, or guaranteed without ads for $10/month.

As much as it bothers me, I'd prefer to work around the first option.

they now have a browser plugin that allows search in private tabs

All searches are logged to your account which is tied to your credit card / kpi.

They are not. See kagi.com/privacy

They promise never ever to share your confidential info, but that promise wouldn't be necessary if they didn't have it in the first place.

I understand they need it for customization and monetization, but search queries are too private to ask for trust.

I'd love if more services worked this way. Same for streaming, YouTube etc.

I need to get around to signing up. I started a couple weeks back, but stalled about some question re: what search means to you, or something like that, went off for a while to think about it, and never went back.

Guess I need to just write "finding stuff" or something trite and get on with it.

(don't ask me open-ended questions as part of a signup process unless you want me to brain-lock and never finish the form :-) Though I just checked and it looks like the signup flow's very different and more normal now, so that's good)

You probably had to answer questions during their closed beta, so they could look for and invite specific user types as they ramped up. As of last week, IIRC, they're now in open beta. That could explain why their sign up flow changed.

Makes sense. I wasn't mad about it or anything, just noticed that it had that effect on me, and was like, "huh, weird". The three or four times I'd thought about going back, remembering that was there had stopped me, too ("eh, I don't want to get started and just end up abandoning it again, waste of time") until checking again after reading this thread.

...or use Grease/Tamper/Violent- monkey and the excellent Google Hit Hider by Domain[0] --which is free and [in spite of the name] works across all major search engines.

I feel like an evangelist for this script as I seem to mention it on an almost monthly basis here. But, as the saying goes "No connection. Just a happy user."

PS: as I said last time 'uBlacklist' was promoted here; whatever the merits of your product, trying to subconsciously associate yourself with the excellent 'uBlock Origin', by using a similar naming convention, is very shady.

PPS: reading the rest of the comments, I'm amazed at the number of people saying they've signed up for this, or intend to. $10 a month for something you could have for free --and we're supposed to be in recessionary times. It's true what they say about 'a fool and his money...'

[0] https://www.jeffersonscher.com/gm/google-hit-hider/

You’re joking right? We’re on a forum where a good chunk of the people commenting have 6-figure salaries, some on the very high side of that. And some here have a lot more money than that.

$10/mo for something used many times every day is a drop in the bucket for many. It’s not true for all, but for many.

It’s not the amount. It’s the value for me.

$10 for something that is minimal value is the path to ruin, especially when done by many people.

There’s opportunity cost (ie, what if I donated to charity, etc) but mainly I want to support high value products with some link to costs. All these “just the price of coffee” 4-hour-work-week type things are an unhealthy way of looking at the world.

I like open source so I can stop worrying about stuff that has near zero marginal costs.

I hate to think of a future where everything I enjoy or use is “just $10 every month.”

To each their own. But to me, a search engine is a very high value tool.

Paid search means that the service is not incentivized to appeal to advertisers. Disclaimer: happy Kagi subscriber here :)

GHH just removes the results from the list, it doesn't add anything. I don't see how it could appeal to advertisers. EDIT: Oh, you're talking about Kagi. Gotcha.

About your PPS: I search from my own machine, from my work laptop, and from my phone. I also use different browsers. Maintaining a user script to “undo” crap results won’t work well in this scenario. Having those settings saved in the search engine itself is really nice!

I’m a paying Kagi user now, and this isn’t my first comment gushing over that product. :)

Also an avid user of GHH. It's excellent. Configurable, easy to export/import, and free.

I have yet to find any competitor which provides its own results and is even 50% as good as Google's results.

Features are nice, but for a search engine results are everything.

Kagi is even better than Google in my opinion. I’m still thinking whether I want to pay a subscription for it but it’s tempting since the results are so good. M

When I first switched from Google to DDG I found myself using g! all the time as I wasnt happy with the results, especially local such as finding a specific shop in my area etc. I dont recall using g! with Kagi, and when I was bored and compared the Kagi search result with Google to see if I was missing out, it turned to be the other way - I realised I was actually getting a much wider spectrum of results. I discovered many cool websites and blogs I never knew excited thanks to it or rather thanks to the fact that they show you what’s relevant to your search unlike Google that shows what they think is.

To me Kagi also feels way better than google. I am paying for the service.

> When I first switched from Google to DDG I found myself using g! all the time

lol, I did the exact same thing! After a month of that, I just went back to Google :(. Google's results are just so damn good.

Try !s for Startpage when DDG doesn't have what you want. Startpage uses Google's index and is quite good. DDG and StartPage together give me more (and higher quality results) without Google's obnoxiously deceptive ads.

That was me a couple of years ago, I've tried DDG every year since it was announced on HN as a project. Google got worse and DDG got better, I use DDG mostly now (also Kagi and Brave, Google, and very occasionally Bing).

Kagi is honestly not good at all about local results. I use !g the most often when trying to find information about stores or restaurants. For all other "encyclopedic" knowledge, like Wikipedia (pinned), Stack Exchange, blogs, etc., Kagi has way less SEO spam than Google.

We haven't rolled out our local results yet. It is work in progress, ETA 2 weeks to shipping first version.

By local I meant like when I’m searching for Adidas trainers or garden fence panels it will bring up shops that sell it in UK and a lot of them are local as they have physical shop in my location. With DDG I was getting shops from America even when I had UK setting switched on. For example tool shops like Screwfix or Toolstation never came up in DDG whereas Kagi shows them on first page.

I didnt mean like places to eat near me etc. sorry. Not something I really do to be honest.

I've had the same experience. Maybe it's my developer bubble, but Kagi has better search results than Google for me.

Ha! I'm the opposite. When I use someone else's computer, I get confused because the results are all crap, then I have to manually type duckduckgo.com into their URL bar.

At least Google Search has started blocking itself with a consent wall on new devices. It's the best feature from them in a while, at least for me. I wish their tracking stuff was opt in too.

A lot of people seem surprisingly happy with Kagi's results but I would like to put my hand up as someone who is, so far, underwhelmed and occasionally frustrated. I realise I've started automatically starting almost all of my searches with the prefix to search via Google instead. I recently reinstalled my desktop OS and I've been happily delaying configuring Kagi as my default browser search.

It's slower, sometimes painfully so (maybe due to downtime? Understandable but not fun). I really miss Google's cards; for example, finding opening hours for a local business is immediate in Google, but requires opening another link which may not even be correct in Kagi. When I'm searching errors or code examples, sometimes Kagi embeds a useless snippet with little relevance. Sometimes it has a bizarre 'memory effect' where one or two of the results will be ghosts from an earlier search but completely unrelated to the current search term.

It's not perfect. I'm suspicious of people pretending otherwise.

Kagi does not have local search results (yet).

Speed is on average faster than Google for most users, so if that is not the case for you please let us know via https://kagifeedback.org

And if there is any kind of bug, glitch or issue please also use kagifeedback.org

> It's not perfect. I'm suspicious of people pretending otherwise.

If is far from perfect. I would say we are 30% through what our vision for the product looks like. I can totally understand how it does not meet your expectation right now.

The beauty of our model is that people pay with their wallets, not their data, and the momemt the product sucks, we lose a customer (or don't get one like in your case). Incentives are perfectly aligned.

The fact that barely a week afer the public beta launch, over a thousand people already pay for Kagi, while still being in beta and (very) rough around the edges, is the greatest motivation we can have to serve our user community well and continuing improving the product in the future.

HN crowd for all their smartness are as ego-driven and as susceptible to human biases like everyone else.

Most opinions come from an irrational hate towards Google. (It's the nerd equivalent to be edgy/hip among peers by hating something popular).

Then there is sunk cost fallacy, like I pay $10, it must be good. It's not Google, it must be good (never mind that the founders of all these companies are all cut from the same silicon valley cloth and are equally good/evil/shades of grey)

Finally, it's the illusion of "Feel Good factors" -- Privacy, David vs Goliath

From reading the site it seems like Kagi uses Google on the backend, so it's probably privacy, filtering and presentation you're paying for.

Yeah but I actually like the fact that Google knows what I search, because it adapts the results to what I care about. When I search for "python" I don't want to learn about snakes.

I've tried Kagi a few times, but the results were not better than qwant.com (which is quite comparable to DuckDuckGo and Brave).

The results have been fine for me, the date filtering and archive.org results in particular have been really helpful. And as an FYI, Kagi runs their own web crawler, but also sources from Google & Bing.

https://kagi.com/faq#Where-are-your-results-coming-from

Of course, I'm biased as I was a beta tester and now a subscriber, hoping they succeed.

Except Qwant and Brave are their own search engines, unlike DuckDuckGo who sources Bing. The first two add diversity to the search engine industry.

Just tried the first example search "best laptop" and results are not really very useful - usually old content and none of the results showed the publishing dates of the articles. https://kagi.com/search?q=best+laptop

What are the ideal results for “best laptop”?

This is for sure a really hard question to answer, but I think we can agree that a 2017 blog post does not constitute the “idea results” for this query, no?

FWIW I was a beta tester, and (mostly) enjoyed my experience, but having this as a demo of Kagi’s abilities on your splash page does not inspire confidence in the product IMO.

There are two schools of thought on how to answer this query:

Here is #1 result on Google: https://www.techradar.com/news/mobile-computing/laptops/best...

Here is #1 result on Kagi: https://marco.org/2017/11/14/best-laptop-ever

Kagi's result showcases our philosophy of preferring non-commercial content to commercial/SEO/affiliate content. It may not be for everyone. If the users want that they can find the same link as Google's #1 in a section called listicles.

We show this example on purpose so that we manage expectations on what you will get Kagi from the very first interaction.

Being using it through the beta and moved toward their annual subscription :)

Also you can give more importance to some website in the search result, create your own search template.

It's so feature rich, and to make things easier they provide and extension to set your default search engine for you.

I was considering subscribing to Kagi but $10/month is almost as expensive as my fiber broadband connection and I don't think I can justify this expense. I would likely subscribe if it was $5/month.

I thought the same at first. Then I thought on an annual budget and price.

Does Kagi worth $120 per year for me ? Yes, yes it does, it's barely nothing compared to the value of having proper search result. I checked ... I make more than 50 searches per day...

$120 yearly is nothing compared to other expenses that I have like my broadband internet around $70 per month for it...

My thought process was like that: i do 50 searches per day, i lose on average 3 seconds per search filtering ads, pinterest,quota,geekforgeek,etc. results. This means 150 second or 3 minutes a day. Which means 1.5 hours a month. My hourly rate is about $70 USD, so paying $10 to give me back $100 of my time seemed like a good deal.

Thanks for the recommendation. It looks like a great service and I'd love to support them, but as a student I just can't justify $10/month as long as Google with an adblocker is still an okay experience. Credit card being the only payment option is another problem, since I (like most people where I live) only have a debit card, with credit cards usually costing extra. I'll definitely keep an eye on kagi though, hoping for more payment options and a <$5 subscription.

Kagi accepts debit cards.

It depends on the kind of debit card apparently. What I have is a girocard[1], which doesn't have the required 16 digit card number and CVC. I didn't even know that debit cards can have those. My online payments are usually made via SEPA bank transfers, Giropay[2] or Paypal.

[1] https://en.wikipedia.org/wiki/Girocard

[2] https://stripe.com/docs/payments/giropay

Paypal support coming soon!

I signed up for Kagi during the closed beta, and used it for a bit. But I kept going back to Google for some searches, the proportion of which was steadily increasing over time, until almost unconsciously I was tabbing down to "search with Google" every time I searched. At that point I figured it was a revealed preference and went back to Google.

I've been using Kagi for almost two months and I absolutely love it. Well worth the $10/mo they started charging a couple weeks ago.

Landing page : signup or login

No thank you

It is great, I used it during their early-access, but the 10/month is a bit steep tbh. I would be more inclined to pay up at <5/month. One of my favourite features was the toggle for showing results only from discussion boards.

$1-2 per month would be worth it for me. But $10 per month just for search - no, it is not, sorry.

I started using it a couple of weeks ago, absolutely love it. I haven't gone back to google a single time yet.

I see people complain about a price, but I suspect also they complain about being tracked by google. I guess you can pick your poison. I think I'd rather pay a few bucks a month at this point, but to each his own.

There are other, better, free options. See my comment above. Startpage + GHHBD.

Wow it costs them 1c per search to service your request!

I tried Kagi a month ago, compared about 20 queries to Google, and all of them were either equal or worse on Kagi compared to Google.

Also showing the date on reddit threads was either broken or not there.

uBlacklist is actually better than Kagi for a multitude of reasons.

http://len.falken.directory/web/overall-disappointed-in-kagi...

Expresses it a bit better...

Does it support IE6? (Google does, and it's a browser I enjoy using.)

Is this the next iteration of the "I browse with javascript disabled" HN comment?

Why would anyone support a deprecated browser?

Because customers like it and they want to retain or attract those customers. (Basically the same reason any company does anything)

We didn’t try yet but Kagi works the same with JS disabled. JS is there to enhance UX, not create it.

Let us know.

That is a good start!

for those who need a double scoop of agencies to go with their agency.