Kagi Search Stats | Hacker News Reader

Kagi Search Stats

kagi.com

62 pointsdfee55 comments

> 7,942

Huh, I would never have guessed their subscriber count was so low and good on them for being open about that, I'm one of them. I've been happy with Kagi overall and once I got over the hump of "I need to ration my searches" it got a lot better. The 1000 searches a month seems to be enough for me, I'm still in my first month but it looks like I'm averaging about 30 searches a day (desktop and mobile) and I've only rarely re-tried a search on google (I use their bang feature so I can just add a "g" to the start of my query to send it to google).

On mobile especially the lack of "search as you type" and relying on google under the hood then redirecting to Kagi is a little weird and sometimes, again on mobile, the top UI will load and then you will up to 10 seconds (!!!!) for the results to pop in. I'll put up with that for a little while more but it's frustrating for sure. The domain ranking stuff is super cool though.

I'm a happily paying customer, and I'm a little ... pleasantly surprised? ... that their numbers are about 1/10 what I expected. I can't really explain it. I want them to succeed and wish them great success, but something about them being small feels oddly good.

I haven't experienced a 10s pop-in on mobile (iOS Safari with the extension installed). What OS/browser? You could probably file a bug report on their feedback site.

iOS Safari with extension as well. It's often instant, sometimes a few seconds, rarely up to 10 seconds.

I definitely do need to ration my searches.

Looking at the top domains, clearly this hasn't broken out of the programmer niche. Hope it eats some of Google's share there as well.

To be fair programmers are probably by far the most heavy users of search engines, and the cohort with the most need for niche search features. I'd expect to see a disproportionate representation in the usage statistics.

Yeah true. Would be nice if kagi could release stats that tell us the distribution of subjects.

Just FYI: if you’re a fan of uBlock Origin and can’t stand mobile ads, I highly recommend Kagi’s Orion browser. It’s great on iOS and overall better than safari + it has chrome and Firefox extension support.

I'm damn impressed that they seem to be the first to implement Chrome and Firefox plugins for WebKit.

its a damn shame that a third party implemented firefox extensions for webkit while firefox on android still doesn't support the same extensions on the same codebase, not to talk about missing all extensions on IOS which this browser does.

that's why, being a hardened firefox user, i see them doing directionless work just to kill time, sipping on google's ad revenue.

case in point. employees are paid by the work they put out and management is paid by "industry standards". that means, management has incentive to keep not doing groundbreaking work because that would jeopardize their paycheck from google.

Yeah, the whole "We need Google" schtick is not aging well. I wish they would partner with, I dunno, like Brave or even start their own search engine and follow a similar model, they're already partnered with Mullvad for the VPN so their getting the idea somewhat.

The moment it's available on linux/android, I will

I haven't liked it much on iOS, but the Mac version is my daily driver for tree view tabs. So hard to go back to anything else on my work machine.

I was excited to start using the 'pinned' and 'raised' domains feature, only to find that it breaks the way that I use the "I'm feeling lucky" bang (a single ! at the beginning or end of a query).

A separate "I'm feeling lucky" bang that directs to what would be the top ranked result, ignoring user-specific adjustments, would be a great feature.

You could post this to https://kagifeedback.org/. They seem to take a lot of feedback.

I really like kagi but the price point is just not worth it. Hope to see some economics of scale. These folks need to stop using bing and build their own crawler.

They would have to raise their price to build a crawler as good as what they’re currently getting.

I do hope they grow to where they can afford it at their current subscription rates.

Not sure, but won't argue that vehemently. I build my own crawler in a few days using pgvector, go-colly, and some embedding. It's a bit over-engineered because of go-python interop, but I'm happy with that. And with a subscription model like kagi you should be able to pay for the storage. I also think that a search engine doesn't need to crawl nor store the whole web. You don't need to crawl pinterest, facebook, instagram, tiktok, and known SEO farms or spam blogs.

I'm too busy to productize this just yet, but might do down the line.

A ton of businesses and events are only on facebook or only on instagram. Probably the same with tiktok. So they are essential if you want to provide a complete search engine.

Well, the business case would be a knowledge base for LLMs. I think that there's a lot of tinkering to be done to fine-tune such things anyway.

I'm not sure I understood you for reply. For example if you're looking for a specific sushi restaurant in your town, their Instagram page might be the best result.

I got you! My search engine would be for things like chatGPT in the first place. They look up knowledge.

But maybe you understood that. Anyway, your point is valid. As I said, it sure needs some fine-tuning ;)

That's a neat project, but it wouldn't satisfy a sizable percentage of searches as a Bing API replacement. It needs much more crawling, processing and a quality maintenance and improvement operation. I think you should keep going with it. Perhaps you could find a way to contribute to engines like Kagi; they already incorporate some smaller and/or weirder indexes.

I really like Kagi and have been a paying customer for a few months. The price is very fair. I find what I'm looking for faster than with any other search engine and I am better treated as a customer.

same here: 100% satisfied and am paying of course, online all day, and yet not overextending my search quota. very valuable in my experience that kagi exists.

In my case what stops me from using Kagi is not the price, but:

* No presence in Europe. Establishing a legal entity in Europe and using an European datacenter would at least add some pressure to handle data correctly. We can never know if they fulfil their part of the deal, but at least the consequences of a publicly visible privacy fail would at least be non-zero.

* Email is required. It would be way more appealing for me if this was handled like Mullvad handles their accounts. I understand Kagi's choice is mostly because they prefer a subscription model instead of a pre-paid one. Yes, I can use a disposable email address, but I'd rather not have to rely on workarounds like that.

(EDIT: I want to emphasize that this is only what is stopping me from using Kagi based on my own preferences; they (Kagi devs) are of course free to do what they consider best for the project from their point of view. I'm just not the target audience for this paid search engine, and that's fine.)

It's been worth it to me. I pay for NextDNS, Youtube Premium, and Kagi and the web feels so much better. From the threads here about taking your life minutes back I've also installed SponsorBlock for Youtube. Life's too short and brainpower too valuable to waste on all the crap. That is the value proposition. How much is your time worth? How much is avoiding wasted mental energy worth?

I get your point! But I don't like the subscription hell. 10 bucks here, 10 bucks there (and Kagi is what, 30?). No thank you. I don't even pay for video streaming.

I might change my mind for 10 bucks a month with unlimited search tho.

> the price point is just not worth it

They mention on the Discord that they're getting close to returning to unlimited searches for $10 per month, so this might improve soon.

> stop using bing

Interestingly, last month Bing was removed from the documentation on search sources[1] and replaced with Mojeek, Yandex, and Marginalia. Running their own crawler at this scale is probably not reasonable.

[1]: https://help.kagi.com/kagi/search-details/search-sources.htm...

> Running their own crawler at this scale is probably not reasonable.

You can interpret your comment in two ways, so maybe I misunderstood you, but they are running their own crawler, and they’re providing API access to their own web and news indices. These are geared towards niche content, and maybe that’s what you meant: they’re probably not running their crawler at a large scale.

Does Kagi really only have 8000 members? I'm a paying customer so I figured there'd be many others paying already.

Considering search engines are on the way out, it's hardly surprising.

I’d be interested in the subscriber breakdown between the Standard Plan and Professional Plan.

I’m in my first month on the standard plan, 20 days in, and I’m at 287 searches. I will have to start paying for additional searches later this week. The 300 limit feels just a little bit low for me.

> I’m in my first month on the standard plan, 20 days in, and I’m at 287 searches.

The 300 searches/month is typical for the general populace, who mostly stay in mobile mode(fb/tiktok/youtube/tumblr etc.) and occasionally open a browser to typein a website in search engine to visit it.

Sadly, from my best guess, these users will never pay for a search engine and they are not tech savvy enough to even find kagi. The professional one is targetted at tech savvy people who frequently uses searches on daily basis to find things. The basic $5/month one is sort of trial equivalent.

P.S. anyone claiming to be never crossing the 300/month quota is probably not searching enough not only on Kagi but in all other ones(excluding Google, where you need to seatch thrice to find anything these days). Entirely my opinion and I am sure, I am very wrong.

There are a lot of professions were google or Kagi isn't really that helpful for knowledge extraction so yes you are techy biased and totally wrong.

uhm cite a couple, maybe? off the bat i can't point at a profession that does not benefit of the internet knowledge

Same, I don't buy their metrics because sure, the average person might search 3/4 times per day, but the average person isn't gonna pay for Kagi. A quick glance at the most popular search queries[1] shows a very likely scenario where people type what website they want to go to and then just stay there.

I'm a heavy Google user, to the point that I can't switch to DuckDuckGo because the quality just isn't there, and by exporting and analyzing a month worth of web history I counted 4507 searches (not counting going after the first page, which Kagi counts as a separate search). I don't really see who the standard plan is aimed at.

Sadly the ultimate plan is way too expensive for me, 25$/mo is what an internet plan costs here!

1. Using Google Trends (https://trends.google.com/trends/explore?hl=en-US) gives me: "google" "youtube" "weather" "facebook" "translate" "whatsapp" "amazon" "instagram" ...

You are not banned from searching after exceeding the quota, this is a common misunderstanding. Instead, each search after the quota is billed at 1.5c

I'm also a very heavy searcher, and at most I've been at 1200 searches in a month. Do you let the browser auto-fill URLs you frequently visit, or do you use the search engine for that?

The blocking of Pinterest domains makes obvious sense (the bane of image search), but the blocking of news domains (foxnews, dailymail, nytimes) less so.

Are news outlets regular polluting search results for people for run-of-the-mill queries?

Fox and the daily heil are not news organisations

People who are of bipartisan ideology hate the other side enough that they ban their news outlets from their search results. I think people do search for a lot of news topics, and then get furious when a link to their enemies appear in the results.

I am not a fan of the text for the $5 plan, 300 searches per month, saying “Includes enough searches for 99% of internet users” when the average user who signs up for their service seems to be using 600 searches per month.

Yes, but it can be worth paying $5 dollars a month to get the highest quality results when you do need them.

Top 10 most-blocked domains:

1. pinterest.com

2. pinterest.co.uk

3. pinterest.ca

4. pinterest.de

5. pinterest.fr

6. pinterest.com.au

7. facebook.com

8. pinterest.es

9. foxnews.com

10. w3schools.com

One of the best features of Kagi. Getting rid of all the Pinterest spam instantly made searches so much better.

I am glad that I am not the only one who hates Pinterest, lol

I'm not aware of anybody who doesn't hate Pinterest. Surely those people must exist out there, but I can't imagine a person who wants Pinterest search results.

My wife somehow knows how to navigate that site and can extract knowledge from it that I would only dream was available.

I'm much closer to your side of the aisle. Burn it with fire.

What’s an example of this forbidden Pinterest knowledge?

Actually being able to get to the recipe for one.

Just generally being competent at navigating the site is the main thing, she uses it to find ideas for crafting often.

Previous discussion: https://news.ycombinator.com/item?id=37006082 (986 points | 25 days ago | 485 comments)

Yeah, except it’s not.

That’s why I appended the empty query string to the url.

This was apparently a re-used URL, announced today: https://kagifeedback.org/d/1899-aug-29th-2023/1

* also, am one of the family subs; come join us :)

Reply on news.ycombinator.com