Implement regional pricing

C1231

Thank you for the feedback @Vlad

LvL

This discussion is now already two years old but how is current status?

in my country the monthly fee for Kagi is almost the value of two days of work for average workers. For students like I am its even more expensive. i recently tried Kagi using the free plan and i really loved it. but I think its too pricey and i cant effort it. I see there is no easy solution as you have also to pay your bills. but brave offers unlimited ad-free searches for just 3$ worldwide. how come that kagi is so much more expensive?

I dont need the summarizer, fastgpt, orion, etc. I just would like to subscribe to Kagi search and willing to pay for it. but right now i cannot effort it. maybe you could introduce an subscription without any extra features but with unlimited searches for a lower monthly fee. thats just an idea. maybe there are some more people like me, but i can only assume.

mullinggroove

LvL but brave offers unlimited ad-free searches for just 3$ worldwide. how come that kagi is so much more expensive?

Brave has received over 250 million in VC funding, uses affiliate links and has its own crypto tolken (BAT). Kagi has, to my knowledge, not talken any VC funding, has advanced AI features and uses several API calls to traditional search indexes like Google, Yandex, Mojeek, and Brave (all of which cost money) next to running its own two indexes.

Kagibeh

I think it would be fair to charge users in higher-income countries (a bit) more so users in lower-income countries can pay less. The question is if this could even be implemented. I'd guess that, if the fees for Brazil were lowered, there'd suddenly be a LOT of Brazilians using Kagi.

If there's hesitation to raise the fees for higher-income countries (which I'd understand) this could be done on a voluntary basis, but the question remains how to even identify those who deserve a lower fee.

Maybe this could be solved by taking into account the search region: I doubt someone from the US would set their search region to Brazil just to save a few bucks.

EDIT: The more I think about this, the more important I think it is. I understand the underlying issues with the cost of search, but that shouldn't be the dead-end argument. Making ad-free quality search a privilege of rich white people (globally speaking) can't be the answer to this.

Vlad

Kagibeh

Adding politics to debate is probably the worst thing you can do. This is all about unit economics and cost.

@LvL
If we add any discounts, somebody would have to pay for those searches as their cost will be the same regardless where they are made. Since Kagi has no ads or VC funding to subsidize the cost, that means that existing Kagi users would have to pay for other users to receive discounts. That mechanism already exists through the 'Gift Kagi' feature.

Other mechanisms would involve governments deciding that ad-free quality search is fundamentally important for their citizens and then choosing to partner with Kagi to subsidize the cost and provide it for their entire population. We would obviously love to see this and perhaps starting a discussion in your country about this through your local representatives is a first step.

Kagibeh

Vlad

Kagi Inc. is a company created with the mission to humanize the web. Our goal is to amplify the web of human knowledge, creativity, and self-expression.

We want to bring a friendly version of the internet, one that has users' best interest in mind, into homes worldwide through our portfolio of best in class products - search engine, web browser, and soon email.

Maybe you should read your own mission statement from time to time. The cost of searching with Google is the same everywhere - a click is a click, and a view is a view. That gives Google a HUGE advantage over Kagi. If you look at the buying power of money, you are charging some users multifold - that's not political, it's a monetary fact. This means you're failing your goals (unless you don't regard certain nationalities as human, or worldwide excludes entire countries/continents).

Sorry to point this out - I think it's a big deal. If you don't see a problem, fine, but denying the existence of a problem just because you can't imagine it could be solved cost-effectively is a missed chance imho.

mullinggroove

Kagibeh That gives Google a HUGE advantage over Kagi.

That is quite the ridiculous comparison. Google (Alphabet) isn't even a search company. Their search is mostly a data extraction operation for their core business - ads. Instead of traditional business model, like Kagi has, Google invented a deeply inhumane surveillance capitalist business model while conditioning users to expect to get everything for free on the web. While you use their services for "free", they actually generate impressive amounts of profit from you (~ 400$ year/user in the US). Google claims to not sell your data, which is true. It instead broadcasts your personal data for free to its 7000+ partners via the real time bidding (RTB) system. Further, being an monopolist, the company for the most part doesn't have to compete and innovate. Alphabet conspires with other big tech giants and rigs the market in its favor. It has cheaper access to money than all smaller entities and can thus just buy up every innovative or competing business.

Comparing such an entity to a small company with fewer than 30.000 users, with no VC money, ads, or data collection to subsidize its services, is a comparison of apples and oranges

carl

Racial hatred or envy aside – "rich white people" – there is nothing that would be fair in charging different prices for different countries. The people who would and are purchasing services such as Kagi in low-income countries are knowledge workers and they mostly come from the elite layer of the population. Meaning they can afford it if they want to. A low-income country is not a country where everybody is poor, it is a country where some people get very rich by keeping the majority very poor. Somebody earning a low income in these countries will not be a Kagi customer, just like they won't be a customer of most products and services. They need things that are a much higher priority than a premium search engine.

Apple doesn't sell their devices for a lower price in low income countries, and they would never sell them at a loss. Kagi shouldn't either.

IshanJalan

carl To this point, the physical iPhones might be more expensive in India but Apple One prices are extremely cheap in India. I'll post prices:
Source: https://www.apple.com/in/apple-one/

Individual plans:
USA: $19.95 per month
India: ₹195 per month (US$2.34 per month)

I understand they aren't equal comparisons but there is similar differences in pricing for Steam Games, Youtube Premium, Twitch Subscriptions, Spotify, etc.

Here's some PPP data: https://shiftnudge.notion.site/PPP-Discounts-86a674a50b7546369e7a0ff9cdb537d3

Vlad

Everyone, please keep on topic.

The way feature suggestions work is formulate not just the problem but also a solution to it. So for everyone involved, how does Kagi implement regional pricing - who pays to subsidize the cost for users receiving discount and what criteria Kagi should use to decide who gets a discount and who does not?

Kagibeh

Vlad To me, those questions are spot on, but if nobody sees the/a problem, then there will be nobody to brainstorm with. And finding a solution will be hard, I have no doubt about that.

The "replies" leave me kinda dumbfounded, because they have little to with what I posted.

#1 "rich white people" is a simple demographical remark - nothing political, let alone a matter of racial hatred or envy. If I was under the impression that Kagi would be used primarily by Kermit the frog, I would have pointed it out just the same. The bet is on - I bet anyone willing to hold against it a case of of beer that the average user of Kagi will have a higher income and a lighter skin tone than the average user of, let's say, Google. That's not an opinion - rather it's a claim that can be proven or falsified. If it's falsified, there's no problem. If it's proven, but everyone is happy with it, there's no problem either. But if that's not how Kagi is supposed to be (see the mission statement above), it might be worthwhile to discuss.

#2 I don't compare Kagi to Google for moral reasons, for ideology or for business concepts (and again, I have no idea where this is coming from). It's a simple fact that both are used for searching the internet, Google is dominating the market, and Kagi is fighting for a share of that market (in that sense, they are direct competitors). Whether you or me refrain from comparing apples and oranges doesn't matter, because that's what every internet user does when they choose how to search the internet. By focusing on Google's "surveillance capitalist business model", you are missing out on the fact that their "pricing" is actually quite socialist: While US users may generate USD 400 per year, a user in a lower-income country may only generate USD 4 per year but still get to use the same search features. If Kagi is planning to cap out at a couple million users, this is completely irrelevant, but if they intend to become a serious search contender (and again, their mission statement suggests just that), they need to wrap their heads around this, or else it'll become a huge growth impediment.

Vlad

I think at this moment Kagi's opportunities for growth are plentiful. There are 1.5 billion people with an iPhone, a not-so-cheap device that costs the equivalent of 10 years of Kagi subscription. And Spotify has 250 million subscribers, paying similar to what Kagi costs, and Kagi is arguably a better use of one's money as it makes people more productive and more competitive in the modern world.

Kagi is now at just 25,000 customers so there is plenty upside opportunity for growth (10,000-100,000x based on above numbers). So basically sky is the limit.

I do agree that everyone deserves to have access to a search engine that has their best interest in mind, not just those who can pay for it. This is a social and economic problem and one way to solve it is for goverments to create ad-free search engines for their citizens, with their best interest in mind, similar to the role public libraries held for centuries. We'd be happy to partner with such forward thinking governments around the world in providing this service.

Kagibeh

Vlad

Spotify has 250 million subscribers, paying similar to what Kagi costs, and Kagi is arguably a better use of one's money as it makes people more productive and more competitive in the modern world.

Here's what Kagi tells me about Spotify:

Based on the available information, here are the key points about Spotify's regional pricing:

Spotify prices can vary significantly around the world, ranging from as low as $0.66 per month to as high as $16 per month.

The cheapest places to get Spotify Premium are countries like Nigeria, Egypt, and Pakistan, where it can cost less than $1 per month.

In contrast, Spotify Premium is most expensive in developed countries like Denmark, where it can cost up to $15.84 per month.

The price differences are due to Spotify's regional pricing strategy, which takes into account factors like local purchasing power and competition.

Again, if nobody sees a problem, there's nothing to solve, but then there's also no need to bend the facts in the process. 😉

It seems to me like some of the posters in this thread have fairly limited understanding of the world economy and the concept of money, but luckily, that's not really relevant for this topic. Only two things seem relevant:

Kagi NEEDS to earn more money than they spend (or else they will fail financially).
Kagi WANTS to bring their version of the internet into homes worldwide (or else they will fail their mission statement).

Ironically, those goals are not so far from Google's (who aim for profits and world domination), so it's not like the problem hasn't been tackled before. As I've pointed out above, Google's solution is to have the users in the countries with the highest GDP subsidize the poorest. Doesn't mean that Kagi has to do it as well, but it's not like it hasn't been done before, and it somewhat disproves the claim that search cannot be discounted - the only prerequisite is that you make enough money overall.

As C1231 pointed out, another vector would be to find ways in which Kagi can save money (and I do think there's quite some potential). These savings could be used to either finance regional pricing, or to lower the nominal fee for all.

Logically, yet another approach would be to find additional sources of money. OldMan973471 hinted at this, although I find it frighteningly unimaginative to assume it can only be achieved by "selling out". Check out this thread for example - that's what it's all about.

C1231

@Vlad is there any idea to work around the regional pricing idea? Because what regional pricing wants to archive is cheaper subscription. If it manages to do so for everyone, the better.

For example, I just use search. Nothing more. No fastgpt, no summarize, no assistant, no quick answer. Nothing. Just search. I want a list of results and I want them fast and reliable.

So it seems Kagi has huge costs per search that is preventing Kagi from offering a subscription that is effectively cheap? And this is the issue to tackle. Bring the cost per search down and Kagi is going to be able to offer lower subscriptions plans.

In April I did ₁₀₀₀ searches. That's more than $0.01/search with tax, which is super expensive in this context!

I really enjoy Kagi but there's the constant feeling that I'm overpaying for it.

Vlad

C1231 95% of our cost is search. Also a single search costs us 1.5 cents to deliver, so at 1000 searches/mo, assuming you are on $10/mo subscription, Kagi is losing money on you.

Regional pricing is typically what companies with high margin products do, or in cases where the cost of goods to produce depends on the location. But searching the entire web costs the same whether done from a desert in the middle of nowhere or from New York, and the results are same high quality Kagi results in both cases so "regional pricing" is not possible.

If we are able to reduce the cost of search, we would just reduce the price for everyone, we would not discriminate users based on location or other criteria.

carl

As with any other product or service, the price is what it is, isn't it? And it is up to each and one of us to decide whether it is worth it or not. I personally think it's a great value at 10 dollars per month. Who said a high quality search engine should be a bargain product, peddled for cheap? For bargain hunters there are plenty of free options such as Google, with their own pros and cons. Because if we're being honest, if Kagi was 5 dollars or even 1 dollar, people would still complain that it costs too much. 10 dollars is accessible for those who really want to have it.

dreifach

Vlad What’s the current cost per search? I’m worried that my thousands of searches a month are compromising Kagi’s sustainability.

papajo

Vlad Well maybe you need to optimize this area first (or maybe review your cost calculation algorithms so that you et a better picture of your actual operational cost) I don't want to sound arrogant but 1,5 cents per search seems to be a huge over estimation.

e.g my ancient at this point intel 8750h (which is in no way shape or form an optimized cpu for this workload far from it actually) at 100% constand load at 3GHz (so after turbo goes off and thermals are at an equilibrium) can perform around 75× 10¹⁰ IPS (instructions per second) while consuming 35W

1 × 10¹⁰ instructions per second (according to my online searches) is what it costs to process a search querry fo r google

(^-- this also sees exceessive to me but I assume they include overhead breaking down the querry to their cluster, spell checking, ad indexing and whatever else they do behind the scenes, in orther words not just the instructions needed to run the code that retrieves the actual useful -to the user- payload aka the results)and that also represents a 100MB of bandwidth needed.

Which means my CPU can process a single such querry in about 100-110ms (100 ms being Google's average latency target btw)

but I do you a better one lets assume it will take it 250ms.

so lets assume my ancient 8750H can process 4 querries (1000ms/250ms) instead of about 10

4 per second *60 seconds = 240 instructions per minute

240 instructions * 60 minutes = 14400 instructions per hour.

so at full load in an hour my ancient CPU would have consumed 35Watt hours, with innefficiencies introduced via the PSU and various other stepping modultions don in the VRM lets say it consumes 45wh

Lets assume an outlansdishly expensive rate of 50 cents per kwh (usually the cost of a kwh in the western worl is under 20 cents and in many places under 10 or even 5 cents per kwh)

that means that the CPU in 1 hour of load would spend 2.25 cents of electricity (using the super expensive rate which means it will spend a lot less actually)

2.25 cents or 0.0225 dollars

so 0.025 / 14400 querries (because that's how much this cpu can do per hour) = 0.0000015625 dollars...

coincidentally the "1.5" part was right but it is not 1,5 cents it is a figure multiple orders of magnitute lower...

And sure there are other costs involved such as bandwidth (Which at big contracts though gets a flat rate and is not calculated per MB but even so 1 google querry needs about 100MB so its not that much) cost of hardware and paychecks etc

But again using rational assumptions 1,5 cents seems waaaaaay overexagerated figure.

So if it is right I have an ancient laptop CPU to sell you and help you reduce your cost.

If not then you probably need to review the ways you calculating cost which in turn will help price the product better which in turn will result in higher profit due to higher demand since more users would be able to affort it.

Or in other words aim for profits of scale not how to squize more profit per individual sale.

C1231

95% of our cost is search. Also a single search costs us 1.5 cents to deliver, so at 1000 searches/mo, assuming you are on $10/mo subscription, Kagi is losing money on you.

That is expensive. A lot. I really hope that you guys are able to tackle the challenges of folding down 10x or 100x that expense 🤞🍀

If we are able to reduce the cost of search, we would just reduce the price for everyone, we would not discriminate
people based on location.

agree

OldMan973471

The only way I see Kagi being able to reduce their pricing is if they sold out like every other company does - selling user data, doing ads, accepting VC money (which I assume comes with strings), trying to sell other products that may bring more profitability, etc.

The term RWP is a political term that really has come out in the last handful of years (especially recently). California is very diverse. To single out a specific skin color is....

Vlad has an excellent point - 1.5 billion iPhone users, 250 million Spotify subscribers, etc. I remember going to South America and seeing how expensive an iPod was (wow).

Biggest reason why I use Kagi is because it respects my privacy when it could easily do what every other company out there does - sells user data to make billions. I feel bad if I go over the 1.5 cents per search count per month with what I'm paying cuz I know I'm costing Kagi money and I want them to succeed because I really like their service.

On subject question: Even if Kagi were to have local servers to XYZ country, it wouldn't really change anything. Kagi still has to pay the 1.5 cents per search indexing costs regardless of location and employee/rent/hardware/etc. I imagine Google makes more money from affluent (note no skin color focus here) search users vs users from less affluent countries and that profit covers that.

Vlad

OldMan973471

Even if Kagi were to have local servers to XYZ country, it wouldn't really change anything. Kagi still has to pay the 1.5 cents per search indexing costs regardless of location and employee/rent/hardware/etc.

Correct. It is actually more expensive for us to serve searches regionally, beause in addition to the cost of search, we need to add additional infrastructure in the region to reduce latency.

We currently use 9 regional datacenters https://help.kagi.com/kagi/search-details/search-speed.html#data-center-locations

Vlad

IshanJalan That proves my point. Things that have hard cost such as iPhone or Kagi Search can not be discounted locally. High margin products can.