🔍 In Search Of...
A Better Search Engine
Welcome to Startup ROI, where we explore global technology trends and how they manifest themselves in France 🇫🇷 . Whether you're an entrepreneur, investor or tech enthusiast, I'm glad to have you here! 📩 Get in touch
📌 If you are a startup interested in being featured or a VC interested in collaborating, please reach out directly: bonjour@startup-roi.com
New Here? Sign up for a weekly dose of French Tech:
🎧 Audiophiles rejoice:
Intro: Boolean Bingo
Anyone else here learn how to conduct a boolean search from their middle school librarian? It would prove to be much less valuable than my elementary school typing class (my words per minute were off the charts!), but it did give me a basic sense of how a system might search a database for relevant information. Fast forward a decade and search became simple, all I needed were those typing skills, no logic. Jump ahead another decade and I can shout a command across my living room to a quasi-conscious AI bot and it gives me a comprehensive answer in less than a second. If you've been following along, you know that computing has come a long way. These days, we produce such vast amounts of data, it's a miracle we can find anything of use. Amazingly, search capabilities have kept up with (and crept into) our digital lives. So much so that you likely don't think about the fact that you're leveraging search algorithms constantly throughout your day. In this piece, we'll glimpse into an era pre-search (but not for long… it's a scary sight) and then examine how the search revolution unfolded to create the framework for much of our web experience today. But most importantly, we are going to go beyond the run of the mill web search and take a look under the hood of some of your favorite apps to better understand how they appear to read your mind.
Untangling the Web
If I were to ask you a question you didn't know the answer to off hand, what would your instinct be? Google it, of course. Google (now under the parent company, Alphabet) has become synonymous with internet search, so much so that like other exclusive market leaders, their name became a verb (others of note include Uber, Tinder, Instagram, Venmo/Lydia). Before they became the omnipresent internet behemoth, Google started as an idea, a noble one at that to help unlock the power of a nascent internet.
Stanford graduate students Larry Page and Sergey Brin had relatively early access to the web through their research at one of the premier higher education institutions in the United States. This novel technology that originally enabled academics to share and cite information at unprecedented speeds was expanding to attract ordinary people, now able to access info from specific destinations by inputting a web address into a web browser. If you had a URL handy, this was practically idiot-proof. You could then move from one location to another by clicking on hyperlinks nested in the page you were visiting. But that was pretty much the extent of your web surfing capability. If only there were a way to input parameters and have the relevant information come to you… This was no small feat considering that even back in 1995 there were roughly 100 million pages hosted on the internet. Larry Page recognized that each computer connected to the internet represented a node with every link signifying a connection between said nodes. It formed the largest graph in human history. Knowing this, all he had to do was:
Index every page on the internet
Quantify the backlinks on each page
Create an algorithm to rank said backlinks
Figure out how to present results based on user input
Easy peazy, right? Well, the math gets extremely complicated and at the time, the computing power required was said to crash Stanford's campus servers. But we all know how this story ends. What's the lesson here? Well, for one, the internet was tough to navigate, so in defining the way we search and find information inherently transformed the ways in which the internet itself developed. That's to say, it wasn't pre-determined that the PageRank algorithm (funnily enough named after Larry Page, not the term WebPage) would incorporate aspects of academia like "number of citations" and the "credibility of referring sources" as key criteria. For instance, it could have just counted the number of words on a web page associated with your search and produced results accordingly. They would have been shitty, but you could do that. Twenty years later, Google search is more powerful than ever. They dominate the category with 92% market share (#monopoly) and as such, they dictate the rules of the road. People produce content and structure their sites in accordance with Search Engine Optimization principles largely defined by Google's indexing process. It didn't have to be this way. But for better or worse, it is. Back in the day, however, the race for search (and the value of search engines itself) wasn't clear cut.
Information Oligopoly (alt title: Infowars)
I don't know why, but I have a vivid memory of my Grandfather telling me his preferred search engine was dogpile.com, presumably because of its accuracy. If you've never heard of it, congratulations, you were born after the year 2000. In those days, Google was still a noun and there were a shocking number of contenders in the race for search engine dominance.
So what happened to all of them? The majority went the way of the dinosaur, but a lucky few pivoted into other lines of business. Like I said, the approach to search wasn't necessarily predetermined. There were multiple theories around what methodology would lead to product market fit. Here they are:
The Open Internet
Spoiler alert! This approach became orthodoxy (aside from some modern exceptions architected by authoritarian regimes). An open internet meant there could be infinite, decentralized nodes leading to a diverse, dynamic and user-driven web experience. This framework clearly favors a search engine that can filter through the morass and quickly surface pertinent sites.
The Walled Garden
This concept didn't completely dismiss the idea of an "open internet," rather it favored what I'll call supernodes, or walled gardens, where a single entity served as an entry point to the internet as well as a curation service. AOL was a pioneer in this regard. Hindisght is 20/20.
Embedded Enterprise
Ironically, this was an early strategy for Google, desperate for a revenue stream after raising venture capital (Google Ads weren't in the picture yet). The idea was to power search on other sites, namely those of big internet companies, to make the user experience more seamless for internet explorers (pun intended). But the concept didn't quite click. Why would I want a user to search for a URL and leave my site? Fair point. At the time, this idea was a value detractor. You don't want to lose eyeballs on your webpage and the search functionality wasn't capable of performing search within the site. Slightly over a decade later, however, a more mature internet would invert this loss into a highly sought after gain.
Search Party
Once the "Open Internet" model was firmly in place the next question was who on Earth could challenge Google and win? As most of you know, Yahoo stayed in the game, but slowly lost favor to the superior search giant that is Google. When Google started raking in cash with their Adwords product (accelerated by their acquisition of DoubleClick), it was only natural other big players wanted a piece of the pie. Bing, Microsoft's proposed alternative is and probably always will be a dud. Google had cemented itself as the search portal to the internet and it appeared no one could catch-up, not even the other Goliaths in tech. Despite their head start, the past decade has uncovered chinks in the proverbial armor for Google's search monopoly. Here are the ways the tide is turning:
The first is privacy. As the public becomes better informed about data protection, a renewed emphasis on anonymous browsing has come to the forefront. DuckDuckGo offers an alternative that puts privacy front-and-center; Neeva, founded by ex-Googlers, offers an "ad-free, private and customizable experience" -- perhaps leading to a paid version thanks to modern consumers' ease and comfort with a subscription model.
The second is the Enterprise. Full circle, huh? What didn't make sense (and wasn't technically feasible) decades earlier is now an incredible category within Enterprise Software. This, too, was a relatively slow evolution. The first catalyst was the App Store. Apple built a home for millions of developers to create applications, but how would end users sift through them to find the ones they wanted? The answer: Search. So, we were back at square one, kind of. The entire world wide web is orders of magnitude larger than what needed to be indexed in the App Store. But the mechanics were similar (indexing, tagging, algorithmically ranking etc.). The Apple App Store and Google Play Store put in place the rails on which all future apps were to run. And run they did. A company like Facebook, for example, became a multi-billion dollar enterprise shortly thereafter, and they too needed search capabilities. This was originally reserved for the largest, powerhouse web applications but soon went from a nice to have to a need to have for any and all apps. If you need to re-invent the wheel (especially one this complex) over and over again, it's a pretty good indication there's a business opportunity. Which brings us to our French Tech feature of this week: Algolia.
Search has become so essential to and ubiquitous in our day-to-day workflow, you might assume it's a basic feature any app developer can turn on. Unfortunately, it's not that simple. You can't just copy & paste Google's source code and magically apply it to your own private data set, for example (idk, maybe you could, but legally, it ain't happenin'!). Every app experience is different, therefore, search configuration will be slightly different too. The good news is that most search functionality has been abstracted away by Algolia and made deliverable via API. This way, developers can leverage existing code for indexing or recommendations and easily integrate functionality into their own application. That means your product team focuses their energy on building cool new features for your app instead of toying around with a homegrown search algorithm that may (or may not) work.
This might sound underwhelming (tough crowd) but it's actually a huge deal. First there's a monopoly on web search (Google), then an oligopoly for app store search (Apple and once again Google), then only a limited number of tech companies with top tier talent and deep pockets could develop usable search and recommendations in-house to power their apps. These dynamics presented an exponential competitive edge -- near impossible to catch up to. Who is going to build an eCommerce store better than Amazon? Well, now a lot of people can by integrating Algolia search and recommendations into their Shopify eCommerce store. How can your app compete with the powers of Siri and Alexa (read: billions of dollars of R&D)? Leverage Algolia's Voice API to create a seamless voice experience, enabling your app to surface information based on the intent of your user's command. What I'm saying here is that a tool like Algolia is a great leveler. It's David's slingshot to the search Goliaths. And for those of you who don't care about how the sausage is made, it makes your experience online better.
Better apps. More competition. Greater innovation.
5/5 Would Recommend
Algolia is part of the early wave of French Tech successes. Launched in 2012, they initially planned to solve the problem of offline search on mobile devices. They ended up joining Y-Combinator's winter 2014 cohort and pivoted to the enterprise search platform they are today. Most recently, they raised a $150 million Series D valuing them at $2.25B. Unicorn status, baby!
To build and maintain relevance in this space, they executed on two primary levers. The first being SaaS, in their case Search-as-a-Service, which enabled them to scale quickly. They've got nearly 10K customers globally, counting among them some heavy-hitters like Lacoste (eCommerce/Fashion), Societé Generale (Finance), & Decathlon (Athletic Apparel) to name a few French brands. The second lever is without a doubt community. A product like this requires tight feedback loops, rapid deployment and enthusiastic power-users. They have a large and highly engaged community of developers who contribute to product direction, testing and building components.
Algolia isn't the only flavor of search in town. In fact, there are a number, each with slightly different specialties. I remember for example, when I worked at Salesforce, we had so many applications we used across CRM, Document Storage, Note Taking Apps, Productivity Tools, Chat etc. One solution, was Coveo, a Salesforce native "federated search" tool, enabling you to run one query across all platforms to find what you're looking for. Another great example is Yext, that also serves B2B clients but is less developer-focused and more business-oriented. Small businesses with online storefronts can leverage their AI Search capabilities with clicks not code and easily tie back to their other systems of record for marketing or support (or expand to other products within the Yext suite). I would be remiss if I didn't mention Elasticsearch, an open source project that's a fan favorite among dev teams.
Conclusion: Feeling Lucky?
By show of hands, how many of you have actually clicked the below button?
I truly never click it. Do you know what it does? It brings you to the top ranked web page based on your search without showing you the full results. Quelle arrogance!? Jokes aside, Google has certainly earned the right to take pride in their world-dominant search algorithm. But their days of absolute dominance seem to be waning. The market for search has widened to include areas outside of their purview. Furthermore, savvy web users are starting to take their data harvesting practices as seriously as their First Ammendment rights. Google's power is nowhere near diminished, but as data sets become bigger and more complex, new types of search will emerge. Algolia is but one of the challengers in this space. I guess we can check back in a few years and see who's feeling lucky...