Hacker News new | past | comments | ask | show | jobs | submit login
How We Found All of Optimizely's Clients (nerdydatablog.com)
204 points by rexbee on Jan 11, 2017 | hide | past | favorite | 45 comments

Always liked this stealthy prospect finding method. Here's another few (best for B2B):

1. If some change happens at one of your competitor's product (got acquired, changed a beloved feature, etc.), go through threads on public forums discussing it. You'll find a lot of current users (ie customers to poach) of that product. A recent example: Atlassian acquiring Trello.

1b. Also, look through the comment sections of news sites that reported this change. You'll also find tons of current users there.

2. Go on user submitted product review sites. Lots of them out there for different types of products: Chrome extension review page, Capterra, G2 Crowd, etc. Approach the users who left reviews expressing dissatisfaction with the competitor product. Or users who use a complimentary product.

3. If a competitor product hosts customer sites on their servers (Shopify, etc.), you can reverse lookup their IP to find all of them.

Then it's a matter of introducing your product with a semi-personalized cold email, reassuring the prospect you have the competitor product's most vital features and you also do XYZ better than them. Here's a template: http://www.artofemails.com/cold-emails#competitor

Also, if you want to find leads on Twitter, find who follows the your competitors support Twitter accounts.

> Approach the users who left reviews expressing dissatisfaction with the competitor product

Though obviously be careful to filter those. There are some customers who are never happy with just one moon-ona-stick, who you would rather leave for your competitors to waste time on!

Interesting. The biggest source of annoying search traffic at Blekko was the people who were trying to find exploitable shopping carts, wordpress blogs, and forums. It was annoying because one could be 100% certain these folks are not clicking on ads ever and so it cost to serve and generated no revenue.

On the plus side it was a great leading indicator of a vulnerability in various bits of code because we would see searches that tried to match a particular package and version increase and then shortly thereafter a story would break about some data breach and personal data being stolen.

I was always a bit conflicted by it. I developed a number of tools which could identify this traffic and automatically ban it on our search engine which was the right thing to do, but you can probably sell that information to these organizations. So as a startup it was leaving money "on the table" as it were. I expect the way extract money out of that stream would be to have a site that accepted bitcoin and would return URLs of pages that matched a particular software package pattern.

Its also a great tool for sales people who are trying to sell wordpress themes for example (or identifying you is using your non-free theme without paying).

Finding unlicensed Wordpress themes is a great usecase! You can identify them as the theme name is in the stylesheet URL by default.


I'm always amazed about how there isn't more competition in this space.

As my Master's thesis [0], I built a crawler that did similar fingerprinting (although less generic). It wasn't something breathtakingly novel, but all in all a somewhat successful project.

It detected > 100 CMS, additional features like ad networks, social embeds, CDN, industry detection, company size etc. In the end, you could run a search and get the result as an excel sheet (because apparently that's what people like.)

The whole thing took about 6 months and ended up with > 100 million domains on a single (mediocre) machine humming away at around 100 domains/s. The sales/marketing folks loved it.

Since I was just finishing university, my skills were still pretty raw, so I'd assume that an experienced engineer would be able to do this a lot faster. From what I can tell, there was a lot of demand out there and sites like builtwith sold their somewhat limited reports (at least at the time) for a good amount of money.

[0] http://blog.marc-seeger.de/2010/12/09/my-thesis-building-blo... Previous discussion: https://news.ycombinator.com/item?id=2022192

That was 2010. In 2017 this space is flooded. We all know how to write web crawlers now and this data is sold by hundreds of companies.

Just in France we have 3 or 4 main actors in this space. Can't even imagine how many US-based companies are doing this.

There's plenty of competition in this space—there's just no reason for them to advertise it.

Ha, maybe the market has indeed adjusted without me hearing about it :)

This is just recycled content from them....


Interesting; I had never considered scraping sites looking for specific embeds as a way of sourcing potential leads.

Scraping sites looking for complementary or competitive products' customers sounds like a novel way to do market research.

BuiltWith[1] and Wappalyzer[2] offer this as a service.

For software with client-side exposure that can be discovered during scraping, BuiltWith has pretty solid coverage (at least when I used it ~1 year ago).

I don't know if those services do it, but you can also find some really useful intelligence from DNS records. From email and calendar provider data to third party services like analytics trackers and landing page service providers (such as Unbounce). If you have an app that integrates or competes with those services, it can be really useful. If you use a DNS lookup service that provides historical record changes, you can even time your outreach to coincide with their annual renewal period when they're most likely to be entertaining the idea of a switch. Or in the case of an integration, wait until after the renewal period to start outreach since you know they're locked in for another year at least.

You can also use DNS records to link together entity ownership relationships. Say a company has competing product lines and doesn't overtly market them as owned by the same company. If they happen to use Salesforce Communities, the CNAME for the Salesforce community subdomain will be specific to each site but will have the same Salesforce account id in it[3], Now you know that they're operating under the same entity, which itself is useful intelligence, but you also can combine the technology usage you sniffed from both sites together.

[1]https://builtwith.com/ [2]https://wappalyzer.com/ [3]https://help.salesforce.com/articleView?id=000205653&type=1

Are these sorts of scrapers able to grab any data from the post payment side of things in any way? I imagine there are a lot of interesting tags for ad tech, remarketing, etc. firing after checkout.

There's no reason these scrapers couldn't be coded to identify fields, insert plausible but synthetic data that'd validate, and submit forms. At least in the case of lead gen forms where payment details aren't required. It's a bit skeezy, but that's never really deterred the industry before.

Back when I used BuiltWith (1-2 years ago), they didn't appear to do that. But then, it's not really necessary since their use case is just binary identification of users. With the advent of universal tags that fire on every page (and you configure in the backend which page or funnel is considered a "conversion"), you can identify a lot of the ad tech in use without any form submission. Plus a lot of conversion and remarketing tags aren't hardcoded on the post-submission page, but wrapped in javascript functions. With minification and bundling, you can get a high success rate just parsing through the javascript files included on any page.

Where automated form submission would come in really handy would be a competitive intelligence tool that scrapes an entire site (and subdomains), identifies what actions on which page trigger which tags, and stitched together entire marketing funnels. Being able to monitor a competitor's likely marketing funnels (and seeing which ones they keep over time and which change) would be incredibly valuable, and would necessitate knowing precisely which tags fired on every page, including post-submission pages.

Interesting, thanks for sharing--these are great insights.

You're right that universal tags and the event naming that you could parse from the JS could be very valuable, although it would be hard to normalize.

And you're totally right about stitching together marketing funnels for lead gen conversions, but I'm not sure how you would get that from an ecommerce setup short of making a purchase and refunding it (which might be impossible to do in many cases).

Part of me has wondered if there are any partnerships BuiltWith or others have with popular browser addons. I imagine if they snooped this somehow from users, it would be valuable to them (setting aside whether it is ok to get this level of data).

I played around with someone's open source version of BuiltWith (forgetting what it was called) a while back and it was pretty cool to see how it works. I'm not a developer (although learning to code), but I've done similar research manually as part of my job, so this is really interesting to me to see what else can be learned.

For example, if there's a publicly traded ad tech company and you know a substantial customer of theirs just removed their tag, or many customers did in a certain time frame, you could short their stock (or vice versa if you see huge growth).

https://publicwww.com/ is better for searching the web for js or css.

I always wondered if someone could do this for households. Provide a way for people to find everyone with lawn needing mowing, or with a picket fence that would need regular painting, or with a tile roof, or a hedge to trim, or driveway needing repair.

When I was 13 I simply went around all of the houses in the area with overgrown gardens and popped a note through their door offering grass cutting.

Not a wealthy area at all, conversion rate well over 50%

Nice! "Do things that don't scale" applied by a teenager!

Datanyze (https://www.datanyze.com/) has built a nice business doing this. You can learn a lot from understanding what software a company is using on their website. It's especially useful for generating sales leads.

True, but datazyne comes out to be crazy expensive.

Some time ago Rob Walling, on Startups for the Rest of Us, mentioned that Datanyze is closer to real time than BuiltWith. So real time, in fact, that you can get a list of sites that have just (yesterday?) started a trial of your competitor's product, and approach them during the buying cycle.

Ghostery does this as a service: https://sitescan.ghostery.com/

Pretty much everyone in a relevant space is already doing it.

> Our search engine is different from search engines you’ve used before. Traditional search engines are geared towards providing answers, whereas our goal is to give you the best list of results for a query.

I kinda get what you're going for here, but I think there's probably a better way to describe it, "best list of results for a query" sounds a lot like a standard search engine to me.

There is a current service - builtwith.com - that does a nice job of productizing this type of research - eg:


It looks like Builtwith only has a few predefined technologies. NerdyData can search for any string in HTML or JS files on millions of websites and do regular expression searches as well.

Here is an example search that can extract the IDs from Google Analytics code on websites into a downloadable list https://nerdydata.com/search?query=UA-%5Cd%2B-%5Cd%2B&regex=...

Slightly OT - Anyone else find it hard to (other than the single link to the search) navigate to their main site from the blog? Strange to have its own domain without core/nav links back to the main site.

When you do get to their main domain, it has these weird links (sitemap a b c...) to another domain in the footer of an extraordinary number of other domains (SEO?)

I thought they were gonna use those client ids and get more info about them using Optimizely API or some hack.

Kind of like using the private keys found in GitHub

I want to know how to do this. There are a number of apps companies use that fit into my demographic. I want to know how to find them. Do you provide this as a service? I couldnt find a link to a homepage or anything.

Cool but they need better coverage of the internet. Looks like they are only scraping some sites. I couldn't find any sites in my niche. Top alexa sites only? ¯\_(ツ)_/¯

We crawl 200+ million websites :) Click the "Deep Web" button under the search box on the homepage.

Thanks for the tip but I tried the "Deep Web" option and still can't find any of the sites. They are small sites that get about 100-500 visitors per month. Also, there are a lot more than 200+ million websites on the web.

What is your niche?

On a interesting side note Optimizely was a very early customer of ours (BuiltWith) starting in 2010 which helped them find customers for their own tool based on sites using their competitors (which at the time wasn't very many), I don't think it will bother them that other businesses can do the same thing.

Glad to see all these other tools in the market now there was and is clearly a need for it.

I wonder if you could use nmap or something like masscan(https://github.com/robertdavidgraham/masscan) to figure the IP addresses of people using a certain software(say mongodb on port 27017). And then reverse looking up those IPs to figure out which companies they belong to and then you contact the said companies to sell something.

Cool stuff. Back in the day, I worked for a website hosting company and was tasked with finding several of our major competitor's customers.

I chose a more rudimentary route than this - which was to convert the huge lists of hostnames to their IP addresses.

Everything in our world then utilized shared hosting, so a relatively small list of IP addresses would map.

We then would proceed to scrape the sites for emails, phone numbers. Which was easy because our competitors had standard templates for their clients.

This can be achieved just by using the publicly available HTTP Archive data set and Google's Big Query to search it.

How do you do market research for SaaS that don't leave an artifact in the client website? Consider project management - what ways can we find people who use Trello/Basecamp?

Nice work. Can imagine tools like this would be hella useful for identifying potential new employers or users of an open-source library you're developing.

Handy for finding employees too:


love nerdydata, always wanted to build something similar myself., but would any sane business person enter the space now?

Love Nerdydata.com! I used their service a while back to find optimizely clients for my sales team to go after.

Looks like they added regular expression searches and a few new data sets. sweeeeeet.

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