Hacker News new | past | comments | ask | show | jobs | submit login
How to sell to 20M software devs with amazing onboarding (garrytan.com)
195 points by garry 11 months ago | hide | past | favorite | 82 comments

I tend to feel a little intimidated by companies that offer fantastic experiences like this, but then I keep trying to remember this:

Algolia looked like shit when it started. There was none of this shine and polish. IIRC it was some text styled with Bootstrap 2. Or might have been 1.

Stripe also looked like shit when it started. No shine or polish, no developer website based developer experience to speak of, no world leading 60fps animations, no fantastically versioned docs, nothing.

Build something people want / something that solves a hard problem and makes people powerful and more enabled. Companies that do this can afford to build fantastic developer experiences after 10 years of doing it, but companies that try hard to build excellent developer experiences will not necessarily be successful.

It was absolutely magical even at the start

I was a very inexperienced developer with tutorial-level experience of Rails, and I learned javascript from Algolia client tutorial. I have a 50-emails support thread with their CEO and CTO when they were doing YC.

This level of patience and care at onboarding a new dev was incredible and I always rooted for them. I came to know them better a few years later (small world) and they are absolutely adorable, very down to earth and modest despite their tremendous growth

But one thing has not changed : when you implement Algolia in a project, it feels like magic. The speed + convenience combo is unbeatable

The landing page was, in fact, Bootstrap! https://web.archive.org/web/20130613201137/http://www.algoli...

> Build something people want

Selling to developers is a bit harder then that. You need to offer something they already know they want. (or you will need a good sales team, and or sell to their boss). And something that can't easily build themselves. And something they can't already get for free (or you will have to also give it away for free).

Yes, keep in mind this is seven years of working hardcore on one problem.

If someone starts a Algolia or a Stripe competitor today, they have to have good user experience to even compete, isn't it? Because Stripe and Algolia have already shown what is possible.

If you plan to compete using their feature set, then yes. But that's a terrible idea.

Find something that Algolia or Stripe won't or cannot do (for deep reasons) and offer that. Stripe has a list of "Restricted Businesses" each of which is a market that would be delighted to be served at all, let alone with some polish. Of course there are complicating factors... but you don't expect this to be easy, do you?

Algolia sadly cannot yet do things like "show me the items around/about $200", i don't care if the are +/- a few dollars

Hi there! I work at Algolia and I may have a solution for you: you can filter on ranges to offer that kind of experience.

For instance, if you decide to accept a +/- $20 margin, instead of filtering on exact price, you'd create a filter like `price < 180 AND price > 220` (replacing the numbers with your formula to calculate the range).

This gives you full power on what is the "error margin" you accept. Hope this helps!

Even if you match their level of polish (which is nearly impossible due to manpower), why would someone use an unknown service over theirs?

I think this follows the other reply: you should offer something different that's attractive enough for a client to pick your service over theirs. Something you enable or focus on that Stripe can't.

Usually this happens because they'd have to sacrifice the experience of a much larger group of clients or your niche is too small to be worth it for a team at a bigger company.

The bet is that serving this new group/need will help you a)end up with a profitable group of customers that grows and/or b)end up developing a user-base/feature-set that makes it easy for you to expand to other users/products and steal market from the bigger competitor.

Except its right there in the transcript. It used to be a self-service SaaS application, to enterprise sales. Nearly all features are gated behind an enterprise-sales "contact us" box with no transparency in pricing.

They're not selling to developers, they're selling to business executives, they give a nod to the fact developers are the end-user of this product, but all the material on the site is as executive-oriented as it gets. Listing other brands they've sold the product to, big stats about ROI and increased sales and revenue and "white-papers" which are really just fluff-pieces to hype up the product.

As a developer, I need to see transparent pricing up-front. It's no good telling me about features, APIs without a way to figure out what it'll cost because that cost might easily be orders-of-magnitude out of range. Look at the pricing for Algolia - $499/mo and you don't actually get any of the features that separate it from products like ElasticSearch which are available as open-source downloads or as a much cheaper hosted service. You get tooling to make integrating search easier than the open-source alternatives, but now you're bound to the tooling provided by that service and its capabilities, or writing your own tooling anyway.

Either give me a monthly cost I can use to determine if its worth suggesting a product, or give me usage-based pricing that lets me start out cheaply, and lets my bill grow as the value I gain from your product grows.

(Sylvain from Algolia speaking) Thanks for sharing, our pricing & its transparency is actually something we are working on right now to address this perception. We've been iterating a few times already over the past 7 years, but this is one more signal for us to act!

The reality is that based on the number of objects you want to search in, your search traffic, your searches' complexity and your index/relevance configuration; the underlying resources can vary soo much, it's hard to have a single pricing that fits all use-case.

This has always been a major deterrent to me wanting to use Algolia. It's basically impossible for me to know whether for x number of users doing y number of searches Algolia will cost $100/month or $1,000/month or $10,000/month. It amazes me that anyone commits to investing development effort in using Algolia (with lock-in, to boot) when you have no idea whether it's something you can even afford.

It's very easy to start with so from time investment it will always be a good choice. I looked at several options for doc search and since I didn't want to invest more than 2 hours I just setup a Drone CI job to scrape and update index every night and added that search bar :) quite happy so far although self hosted solution would be preferable

As with all usage-based product, when you start it is cheap (much cheaper than building on your own), then as you grow your resources will hopefully scale with the number of users.

If your resources don't scale with usage, yes, you have a problem, but I'd say not limited to algolia

What's different is that with other products, i.e., those that have clear and understandable pricing, you can forecast the future costs and make a decision about whether it's an affordable solution for the particular product/business you're building. With Algolia, you can't.

Why would I want to build a product/business that uses Algolia technology as a key component when I have no idea whether the eventual cost of Algolia will be acceptable? I wouldn't, which is why despite Algolia's extremely attractive features, I have time and time again been unable to recommend its use.

Btw, part of this pricing rework is also about removing some of the feature gating. It's an approach we've taken to really get the feature right, first releasing it to a smaller set of (larger) customers before opening it up to the whole mass.

So give people some sliders with all of those things, and let them drag it around.

What is my cost with a million objects, an average computational complexity of 3 compute units per search, etc.

If you can't do that, at least give some example scenarios (small, medium, big, huge, etc) with total pricing.

Just... something. Anything would be better than a "contact us" link.

I absolutely refuse to use or recommend anything that looks like a black box enterprise solution if it lacks transparency.

Hey, while you're on it, look into compatibility issues with older versions of Firefox on your web interface.

It's impossible to use due to bugs.

On the flip side I run our docsearch for stream on Algolia for $35 a month. https://getstream.io/chat/docs/

Search is fast, relevancy is awesome, it's hosted and I have analytics. So for that use case it's a clear and major win against installing, configuring and adding search analytics to Elastic.

I recommend Algolia to my friends who are working on side projects or want to implement search for small datasets (less than 100k documents to index). Algolia is indeed an amazing service / api!

But I don't use Algolia for my project (and my full-time job now) Listen Notes (a podcast search engine). I did some back of envelope calculation on pricing, and figured that I had to pay $50k ~ $100k/month (at least) to Algolia, which is like I have to raise a pre-seed round every month :) I run a podcast API myself. I understand that it's impossible for an api (and its pricing) to fit every use cases in the world.

There is some truth to this, API pricing is difficult. Hard to make it work for every use case.

How did you come to that conclusion?

My experience (target is small teams)

– Be up front about pricing and terms (depends on target business size though)

– Easy to understand but detailed feature lists, tier comparisons, screenshots, videos

– Public, well organized documentation that goes in depth and is easy to search/navigate, also acts as good SEO

– Plenty of samples in documentation so you can paste and go

– Quick support that can get technical enough

– Offer yearly and monthly subscriptions

Developers usually spend a few minutes scanning each solution they're comparing, so within very short time and clicks, they should be able to decide why you should be a pick and not the next one. There's usually a hard criteria. They need to solve a problem. Investigate what kinds of features your users picked you for.

Also, trials: If you don't offer trials, you'll get higher quality customers. (But still offer refunds just in case.) They'll be the type who can spend the first monthly fee to check out your site, and have realistic expectations.

And if you go to their home page and try and click on "Developers" in the navigation bar, nothing will happen...

You have to click on one of the other links that actually shows a correct hover icon and THEN click on the "Developers" link to get there.

Maybe their docs are great, but a bunch of links on the main site are broken...

Oh that's weird, a dropdown menu should appear and give you access to a list of developer resources :/

This is an advertisement of Algolia. Isn't it?

Full transparency, I worked with Algolia when I was a YC partner years ago. I also invested from my fund.

But I've been meaning to do a sit-down discussion with this team for a while. We've funded a lot of API-first developer-focused businesses in the past: Easypost, Lob, Apollo GraphQL, and developer focused stuff like LogDNA and so some of this is wanting to point to things that work well.

Interactivity in particular is really important to nail in first time experience.

I'm also a customer of Algolia— I use it in all my software projects including Posthaven and Bookface at YC. I had to implement Lucene at Palantir and I used both Lucene and Thinking Sphinx when I ran engineering (and devops too) for my top 200 site Posterous. (In retrospect I should have hired a devops team... but that's another story.) And the contrast between running your own search at scale and having something work out of box with code that's ready to rock... I genuinely believe it's pretty magical.

> Full transparency, I worked with Algolia when I was a YC partner years ago. I also invested from my fund. But I've been meaning to do a sit-down discussion with this team for a while. We've funded a lot of API-first developer-focused businesses in the past: Easypost, Lob, Apollo GraphQL, and developer focused stuff like LogDNA and so some of this is wanting to point to things that work well. Interactivity in particular is really important to nail in first time experience. I'm also a customer of Algolia— I use it in all my software projects including Posthaven and Bookface at YC. I had to implement Lucene at Palantir and I used both Lucene and Thinking Sphinx when I ran engineering (and devops too) for my top 200 site Posterous. (In retrospect I should have hired a devops team... but that's another story.) And the contrast between running your own search at scale and having something work out of box with code that's ready to rock... I genuinely believe it's pretty magical.

You didn't answer OP's question, which was:

> This is an advertisement of Algolia. Isn't it?

Reflecting on your answer, you appear to be advertising for Algolia as well. It's a nice product, but the impression I get is that OP is right -- this post is an advertisement.

It's certainly advertising, but he did seem to offer some extra role of: > wanting to point to things that work well

Which is more inline with normal HN content. Pointing to some technical thing which works well we can integrate into our coding/startups/etc.. In this case it happens to be a non-open source subscription business replacement for coding your own search, so it's very commercial, of course, but paid off the shelf solutions are technically an alternative to coding your own search with open source stuff.

You should advise the Algolia people to remove their branding from HN's search experience. It doesn't leave a good impression.

Why not ? Algolia is a YC company. YC is invested in Algolia’s success.

Sorry if I wasn't clear. It's not the sponsorship relationship that leaves a bad impression, it's the actual search experience that is bad and makes Algolia's product look bad.

Do you think about any improvement I could share with our team? I know they're always looking for feedback to improve the HN Search experience.

Yes, though it definitely raises some valid points. If I'm picking between a few services, I'm usually going to go for the one with the best docs and onboarding experience. That seems like a given, but a lot of companies are severely lacking in those departments.

Hosted by Garry Tan, Cofounder of Posthaven where the Masterclass is hosted and Managing Partner at Initialized Capital, who also just happens to have Algolia in their portfolio. Go figure.

As a VC, marketing is part of the job. Perhaps the most important part of the job.

Getting LPs to invest in your fund encourages strong personal brand building.

Getting startups to want _you_ to invest versus others—and for many "great looking" deals (the ones you want), the company has options—requires strong personal brand building.

In the medium term incentives are less towards being good at helping companies.

You have to do enough good marketing to get founders to sign on the dotted line. After that, if they don't like you or are "bleh" on you, there's nothing they can do about it. What's done is done.

By the time you've fully deployed the capital in your fund (2–3 years, excluding reserves for follow-on), you're raising your next fund. That's _far_ too short of a time span to know what the results will be for Fund I.

What does it take?

Good marketing.

It's not a coincidence, then, that successful funds revolve around good, or at the very least incessant, self-marketers.

Garry has appeared numerous times on @VCBrags (and has, funnily enough, blocked them.)

For me, all I'm really trying to do is lift up stories and experiences that I think are superlative and noteworthy, with the hope that it helps more of those things exist.

This is helpful feedback for me in that if this feels too much like advertising, I'm not doing a good enough job there. Sorry about that.

Trying to be helpful. If there are things you would rather see, happy to try to throw that on my YouTube channel too!

And an YC company, which maybe explains the #1 position out of the blue (but maybe I’m way too cynical)

No moderators touched this submission. We moderate HN less, not more, when YC-funded startups are involved. That's the first rule of HN moderation.


(Launch HN posts are an exception. Those are one of the things that HN formally gives back to YC in exchange for funding it: https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...)

Thanks for the explanation. It triggered my ‘internal’ red flag when I saw a post on #1 very quickly and with no comments at all. That is very rare. Maybe the authority of the submitter (and his network for upvotes) had something to do with it?

It's not as rare as you might think, but yes, I'm sure Garry's (well deserved) reputation was a big part of it.

I noticed one dubious upvote making it past HN's anti-voting-ring software, but that's not enough to push a story onto the front page, let alone to #1. It's also extremely common. I didn't notice evidence of any "network for upvotes".

If you mean that you don't signal boost YC posts then why not say that? Moderating less means to me reads as if they are less likely to be downgraded in ranking.

  if YC disable_moderation;
  Age all posts, push them down the ranking
No moderators touched this submission means nothing if they touched every other submission.

But that would be utterly deceitful. If you're going to imagine that sort of interpretation, why believe anything I say in the first place?

When I say we moderate HN less, not more, when YC or YC startups are involved, the word "moderate" means all of the above: boosting submissions, downweighting submissions, intervening in comments, and so on. When I said we didn't touch this post, I meant we did nothing to affect its rank.

There's an important thing to add, though. "Less" does not mean zero. We still moderate—we just do it less. Mainly we try to be consistent with how we moderate other submissions, in order to be fair. In this case, for example, I just reduced the effect that flags are having in driving this post down the front page. But I reduced it by less than I would have if the post were non-YC-related. If that sounds scandalous, please re-read the previous sentence.

> When I said we didn't touch this post, I meant we did nothing to affect its rank.

Ok, that's clear and unambiguous.

> If that sounds scandalous

It sounds like you are following through on exactly what you said you were doing on the search you linked elsewhere. I read some of those and it clearly stated "less intervention". So for you to apply "less intervention" here is you being consistent.

> flags

For the record I didn't flag the submission. It didn't appear, to me, to violate any guidelines.

> If you're going to imagine that sort of interpretation

Sounds like I have violated the guidelines: Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize. Assume good faith. I appear to have escaped censure at this juncture, perhaps due to the lesser moderation applied in this thread.

Ha! Thank you.

> Moderating less

it means they are less likely to intervene in the discussions on them.

The accusation was the #1 ranking. Nobody mentioned any intervention in the discussion.

It's gamed but not by the mods. Not even sure in a bad way.

User posts story to blog and includes hn link in an email blast. Followers upvote quickly then the article slowly losses interest.

But catchy title gives more attention. Podcast during a workday means more bookmarking for the ride home.

I think that everything you've described here is "fair game" and mechanisms available to all of us.

Right. And from my interpretation dang wanted to clarify that not only they didn't touch that, they also keep away from such submissions in other ways.

It's not cynical, it's open and clear if you track HN story positions that a large part of rankings are not organic. In the past HN were more open about that fact than they are now.

I've posted tens of thousands of comments explaining how we moderate HN, and can tell you that we're more open than ever, so I'm not sure where you got that.

It is clear that some stories are boosted up, and some are boosted down, and it is not clear the process those get chosen.

It may be that "you find them interesting is enough", and it's your site so this is not a complaint, but it is not a transparent process because despite "thousands of comments" there is little documentation or explained policies.

We know there's an "anti flame war" trigger on the down side, but of more interest is what causes stories to jump hundreds of spots and that has never been well explained to my knowledge. (Or perhaps it has but I haven't been to read such an explanation).

My guess is that flags are responsible for a lot of weird movement in the front page ranking. You flag an article and it starts losing spots - a number of people do it and the article can die entirely. The flag count is hidden to prevent bandwagoning but is an important signal. Thus front page posts move in mysterious ways as if guided by an unseen hand.

Well they are and the hand is users flagging posts. Plenty of times I've seen

  Article A, 2 hours, 50 votes
  Article B, 1 hour, 100 votes
Where A is ranked above B despite being older and having fewer votes (the signals we can see). I think the missing link is flags. I bet if we knew the flag count the front page ranking would make more sense.

The part that sticks a bit is "In the past HN were more open about that fact than they are now". I'm pretty sure that's not true in any way, and we've worked (and work) hard on it.

Trying to write comprehensive policies would end up interesting only the sorts of users who like policies, who are also the sort who will raise objections no matter what we do. The more policy we produce, the more objections and meta concerns they will raise, so to go down that road would be to perform a DoS attack on ourselves. It would suck limited resources away from things we can do to improve HN for the community as a whole, instead of just a vocal, litigious few. Since we're going to get criticized from this angle no matter what we do, we may as well take our lumps up front. If that sounds harsh, I'm sorry—it's mostly because policy-writing is the last thing my soul cares to do, and the bureaucratic parts of this job make me grumpy.

HN has always been curated, and that involves human judgment and interpretation—there's no way around that, no way to spell it out, and certainly no way to formalize it. As far as I can tell, the community is somewhere between fine-with-that and prefers-it-that-way.

But we're fully open to answering questions. I spend hours each day doing that, in threads and by email. There's basically nothing people ask about HN that we don't answer. Before someone objects to "basically", I'm throwing that in because there are always corner cases (e.g., a question we can't answer because it would compromise some other user). You can't run a site this complicated without inconsistency. But the principles we practice are deeply about openness and satisfying curiosity.

As for stories "jumping hundreds of spots", if you mean jumping up, that's the second-chance pool (described at https://news.ycombinator.com/item?id=11662380). If you mean jumping down, it's some combination of software penalties and/or user flags and/or moderation downweights. Most often it's user flags.

It may be a matter of perspective because from my point of view, the comment you have linked is the past to which I was referring: that is from 4 years ago.

That may have flown by for you but is much of the time I have been visiting HN.

(Sorry, I only now saw this reply.)

That's just the standard link I use to describe the second-chance pool. We've been just as open in the subsequent four years.



Seems like it ... my experience with Algolia was fairly lukewarm -- it seemed expensive for the capabilities offered.

I think this is one of the fundamental quandaries of many services that sell to developers. There's always a way to roll it yourself such that the cost in money is low but the cost in time is high.

For me, I just don't have enough time in the day (parenthood + day job as an investor) so when I do code, I am looking for a lot of leverage, and I'm willing to trade off a monthly fee for it since it's like licensing software that lets me get something done a lot faster.

Most cloud services are like this: You could do it yourself, but when you compare it to hiring more people on your team, you should use a cloud service.

The upside to this is that a lot of startups are going to be able to get a lot more leverage through building on top of services like Algolia. I mean, this is really the AWS strategy in a nutshell, isn't it?

I would agree on your point here, in some ways. Algolia is a limited search; there's a lot of things you can't do on the fly (that you can in, say, SOLR). However, the onboarding (as noted) is pretty good. It's really easy to get started using it. The learning curve is much lower and so, too, are the capabilities. If you don't need those extra capabilities, you just need to get search available, it might be the right thing.

Oh yes, low bar for sure. The challenge is to get it to be a high ceiling.

I see a lot of comments here deriding this as an ad for Algolia. With all the conflicts of interest, that may well be true.

However, automatically dismissing an entire piece of content because of a conflict of interest, which actually may not be that big (e.g. what percentage of their portfolio is Algolia, that it'd be worth pushing), seems closed-minded.

Investors are people, and like anyone, being excited about a company/product and using their platform to share it, is a pretty benign thing to do.

The over-the-top cynicism seems unnecessary.

Meta question: @garry - Do you have a post on the your video making process and what equipment you're using? Would be curious to know if you're mostly making these videos all yourself or if you hired a video production agency to help with the recording, editing, etc. Thanks!

Call this whatever you want, but it really feels like it's primary purpose is to sell me something, not to teach me something. And that's fine- that's something every company has to do eventually - but it doesn't feel like real HN content to me.

Maybe an over-enthusiastic customer here, but obviously I am also an investor. That's helpful feedback for me, so I appreciate it.

i would guess that a full %50-%70 of HN content is content marketing of some sort (either for a startup or for someone's personal brand). so many prescriptions on HN but no ban on this kind of content. makes you think.

The content has to meet certain quality standard to survive here. Most content marketing dones't, but some does, sometimes. And when it does, you may as well leave it up - per https://xkcd.com/810/.

Better that companies are putting dollars to creating tutorials, education, helping people directly instead of more banner ads.

Wow, that's fantastic, best xkcd ever!

Nice try :).

Oooh, oh, are you kidding me?! This is the meanest and most hilarious thing anyone has ever said to me on HN! Two words and I got accused of being a spam bot making a "not constructive" and/or "unhelpful" comment!


I tip my hat to you good sir. I have been roasted, but well.

dang? has explicitly stated that content marketing is allowed on HN; it just has to satisfy everything else required for a post (intellectually interesting, etc.).

HN occasionally has interesting submarine stories. But to paraphrase Freud, sometimes a submarine is only a submarine.



Right, here are two things I wrote about that on days when there was time to write about that.



From an HN point of view the question is whether it's interesting or not.

Trying to add search features in my personal blog. Its free plan looks good enough.

p.s. looks like Hacker News is using algolia for searching right now.

HN Search has been using Algolia for many years now.

Looking at the docs at 4:37 (https://youtu.be/idyfHs3DWmc?t=277), I'd suggest they add Elixir and Rust as options!

Disclaimer: I'm maintaining the Java/Scala/C#/Go API clients at Algolia.

Exilir and Rust would indeed be great additions to our integrations. However, we are really cautious when considering to support new languages. As said in the interview, we try to provide those API clients in the most idiomatic way for each language.

At the moment, we do not have a lot of demands for those languages nor the necessary workforce to onboard on those two languages (only very few people are proficient with those languages at Algolia and we don't have any production code using them IIRC).

Sure, I understand. It was a purely selfish request since 100% of my production code is currently Elixir and I'm also interested in Rust.

Can't blame a dev for being drawn to newer languages!

Applications are open for YC Summer 2021

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact