Hello! What's your background, and what are you working on?
Hey! I'm Kevin Ohashi, and I am the creator of Review Signal. My background is a mix of computer science and business education. I studied Economics and Computer Science at The University of Vermont I then went to graduate school in Lund, Sweden and got two Master's degrees, one in Entrepreneurship and the second in International Marketing and Brand Management.
Review Signal is all about honest web hosting reviews. It uses the massive number of opinions that people are sharing on Twitter as the source for reviews.
I was involved with the web hosting business for about a decade before I started Review Signal, and there still wasn't a single review site that had earned a modicum of trust from people who know the industry. So my goal became to create a consumer review site that built trust by bringing full transparency to the reviews and not simply ranking based on who paid the most.
It's hard to benchmark how well I am doing, but the rare glimpses I get from the companies themselves tell me my algorithm is pretty closely tracking Net Promoter Score (NPS) according to the one company who has publicly published their NPS.
Today, Review Signal is the largest web hosting review site in the world by at least an order of magnitude, possibly two orders, and earns roughly between $3,000-$4,000 per month.
What motivated you to get started with Review Signal?
My thesis for the International Marketing and Brand Management degree evolved into Review Signal. My thesis was about using Twitter to predict box office sales for movies, and it worked pretty well. I collected ~5 million tweets over a few months and found the predictive power of the data was very strong when combined with sentiment analysis. For those unfamiliar with sentiment analysis, it's all about figuring out the tone, generally good or bad, of text.
After I graduated, I moved back to the US, determined to start my own business and began working on Review Signal. I chose to focus on web hosting reviews for many reasons. I mentioned a big one earlier, that the competition is basically all pay-to-play affiliate junk which sells consumers to the highest bidder. It also made sense because I was intimately familiar with the space, brands were generally global, and nearly all had affiliate programs.
I honestly didn't do much validation of the idea. I felt I knew the space very well, and knew there was a need for an honest hosting review site.
What went into building the initial product?
Building Review Signal was a two-year-long process. The biggest challenge to me was the technical side. I tried for months to find a technical co-founder who would bring more skills to the development side and let me focus on product and marketing. I failed in the search, and eventually decided to just become the technical founder.
I had learned a decent amount from my Master's thesis, but it was nowhere near enough. My only expertise was in knowing the hosting industry and being an OK software developer. Everything else I had to teach myself, because I couldn't find anything off the shelf that was good enough for what I envisioned.
Nearly all of the tech powering Review Signal was built from scratch in PHP. I'm sure that many technical readers are asking, "Why PHP?" given that there are well-developed and -supported libraries for Natural Language Processing tasks in Python. My answer is: I didn't know Python. I had been writing PHP for around a decade at that point, and I was very comfortable with it.
So I wrote all the libraries I needed from scratch, including wrongly implementing a Bayes classifier, because I didn't really understand it at the time. (Yes, I re-made it later — correctly, I think.)
The only third-party libraries I used for anything significant were WordPress for the blog and a library to handle the streaming API for Twitter. The biggest challenge for me was that I had never built anything this large and complicated before. I made architecture decisions based on what made sense to me at the time with as much feedback as I could get from close friends.
I tried to design it as well, because "How bad could it be?" Tragically bad. Thankfully, I met Alexis Ohanian at a YCombinator event in NYC, and he went very far out of his way to sit down with me for an hour and give me feedback.
The most important thing I took away from it was how bad the design was and I really needed to spend some time and money fixing that. A great product with a terrible UI/UX isn't going to do anyone any good. Hearing it from someone I respected drove that point home. I went through three designers before I found JR Harrell who worked with me on the design you see today.
How have you attracted users and grown Review Signal?
A lot of people say, "Don't focus on the launch — it doesn't matter." But for Review Signal it mattered tremendously. I launched Review Signal on TechCrunch in September 2012. I can confidently say that that article single-handedly turned Review Signal into a business. TechCrunch has better SEO than I ever will. When the article was published, it shot to the top of the Google results for some of the most competitive keywords on the Internet. (We're talking $20-per-click territory!)
So not only was I getting this free, highly valuable traffic, the users were being told that my site was legitimate and everyone else wasn't. The reality of how it happened was that I just got really lucky that my email got forwarded from an editor I had met 6 months prior to a writer who was interested in the topic but probably wouldn't have opened my email unless it was forwarded by his editor. I even wrote about the experience later trying to figure out why it happened.
The most challenging part of running Review Signal is that if I do my job well, my visitors should never have to come back. They pick a great company and live happily ever after. If I don't do a good job, then the users shouldn't come back, because I failed them. So nearly every customer I acquire is brand new. That means relying on being present everywhere people search for advice on web hosting.
I have moderate Google rankings. I spend a lot of time in specific communities on Facebook, forums, Reddit, etc., trying to help people out and mentioning when my information might be relevant to helping them. It's a delicate balance between being spammy and helpful though, and I try not to cross that line.
The other big part of my marketing strategy is creating very high-quality content. The most successful series of content I have written is an annual analysis called WordPress Hosting Performance Benchmarks. It's a multi-month long series of benchmarks I run on basically every web hosting company that matters in the WordPress hosting space.
Today, after four years of running it, companies approach me and trust my benchmarks. I've tested enterprise-grade plans that I don't think anyone else has ever had the opportunity to benchmark. I've gotten this opportunity because I do things as transparently as possible so that everyone, companies and consumers, can see how all these options stack up against one another.
I've only ever had one complaint from a company about how I run my tests, and it was a misunderstanding because they were specific to one country and I test from a global perspective. They didn't come out as well as they believed they deserved to, considering their target audience. But misunderstandings create room for improvement, which is a nice silver lining.
One other nice part of focusing on certain niches is that my content receives a lot of attention within that niche and gets a fair amount of press when I publish because of the relationships I have built over time.
My general content strategy is I try write about things that matter and aren't the same old rehash people have read one thousand times. I'm not sure it's been the best financial strategy, but it's certainly built up my brand among the tiny group of people who care about this stuff. I write a lot about the bad things going on the industry, but very few people seem to care. Perhaps it's just cathartic for me, or the tiny number of people who care are important relationships. Probably a little bit of both.
The best-performing pieces of content for me have been articles aimed at helping consumers buy or achieve something. It's difficult to know which articles will do really well sometimes, but I just try to write about what matters to me, and convey why it's important as best as I can. I test ideas and early drafts out with friends, inside and outside of the industry, and see which ones seem to really drive interest, and with whom.
This has a great side benefit of getting feedback/ideas and getting buy-in from the feedback-giver. If that person happens to be influential, they are more likely to share something they feel like they contributed to. Just remember to say thanks to the people who help you out, and let them know when you publish.
Since Review Signal is a one-man operation, I try to automate as much as possible as well. I've built a lot of customized marketing automation tools and try to use existing tools like IFTTT to help me be as productive as possible with my time.
One of my favorite tools that I created is named Niche Lead Generator. It scans for high-value leads — in this case conversations about web hosting reviews — and helps me engage them. Instead of trying to check everything all the time, it does the heavy lifting and filtering for me, only giving me what it determines are the highest-value potential leads. I know it works because I've literally had articles written about Review Signal because of it and tracked sales directly attributed to it. It lets me spend more time focusing on creating content and building relationships with journalists/bloggers/influencers who help spread the content.
My biggest takeaway from growing Review Signal has been that I have no idea what I'm doing. I just keep trying different things and see what works. If it works, do it again, try to improve on it, try to scale it. If it doesn't work, try changing it, try to figure out why it didn't work. Sometimes it's ok to throw it out, sometimes it's worth testing a few times. I don't have a magic formula for any of this. I try to look at analytics and see where my traffic is coming from and which of it converts the best.
I also really like to go out of my way to talk to anyone talking about or linking to Review Signal. If I see someone write about Review Signal, more often than not, I try to contact that person and open up a dialogue. Why? Because they are already advocating for me and I want to thank them.
Sometimes they become friends, sometimes nothing happens. Sometimes they constantly send new people my way. Every relationship might make a huge impact on me, my business or my life; so I try to maximize my potential for serendipity.
What's your business model, and how have you grown your revenue?
The business model of Review Signal is quite simple. It makes money from affiliate links. Most web hosting companies have an affiliate program anyone can sign up for. An affiliate simply puts a special link on their website and earns commissions when someone clicks through and purchases something. Review Signal is signed up for nearly every affiliate program available, the exception being questionable terms of service language which might not allow negative things to be said about the company.
The business model hasn't changed from the start. Review Signal started making money from the day it launched. It made a whopping $52.38 the first month (a very shortened month) and $535 the first full month of operation.
The revenue can fluctuate tremendously, because large sales can earn hundreds of dollars in commissions. But on an average month I expect to make around $3,000. The best month so far this year was around $5,000. Hopefully that becomes the new normal, but it immediately went back to around $3,000 the following month.
Because it's affiliate sales, I have very little control over seeing which visitors of mine convert. I have proxy information about visitors, like pages/visit, time on site, etc. But most affiliate programs won't let me actually track which exact visitor converted into a sale. They may also convert into a sale months after I send them to a website (affiliate cookies may last months or even longer).
I've tried to stop worrying about it, but in the back of my mind I'm always thinking about how many sales I made today, this week or this month. But the answer is generally random. The best month this year actually started out the first half on pace to being the worst month, and then blew up. Why? I have no idea and never will. Welcome to my life as an affiliate with little to no control over tracking my sales. It's a great way to lose sleep for no reason.
If you're building an affiliate site, realize the most important thing is going to be your ability to get traffic. So you better be very good at at least one of the following things:
- SEM (Search engine marketing)
- Content creation
- Getting press / marketing
- Conversion optimization
In reality, you probably need to be doing most/all of these things. Each of them can go pretty deep, and you can spend all your time on any single rabbit hole. I spend most of time focusing on SEO, Content creation, and press/marketing.
I don't think trying to push sales too hard is a great image for a neutral review site, so I don't focus too heavily on conversion optimization. SEM has been generally too expensive for me to compete. I do look for other paid advertising opportunities, e.g. services like Outbrain to push my best content out wider to targeted audiences.
Here's a look at my estimated monthly revenue this year:
What are your goals for the future, and how do you plan to accomplish them?
From the technical side, I'd like to redesign Review Signal and build more interactive tools for users to explore the data I've collected. I think it currently does a decent job presenting the data, but there is a lot more room to create more powerful and interesting tools.
I'd also love to expand it into other niches. But there are some extremely large challenges with that, because of the way I've built and trained my system. It would probably require bringing in some more developers to rebuild a lot of systems in place now.
From the business side, SEO, creating content, and word of mouth are my three primary methods of marketing Review Signal. I still don't rank on the front page for a lot of the biggest keywords like 'web hosting reviews.' I would like to rank first for that keyword, but the competition is extremely entrenched, and many use black hat techniques.
I'll continue to create content when I see the opportunity to create something unique and valuable. Word of mouth generally follows from awareness, so I'll continue to engage and help people in various online communities, and slowly convince more people about Review Signal's mission.
I don't have any illusions. It's unlikely that Review Signal is going to break out in a big way. It's slowly grown over four years since being launched. The people I'm targeting are generally one-off research-and-buy product consumers. So the only real way to grow sustainably is to create a wider marketing funnel and constantly explore new potential sources of customers. Every article, link, and search engine position will slowly be fought for, and over time, in aggregate, hopefully Review Signal continues to grow.
What are the biggest challenges you've faced? Would you do anything differently if you had to start over?
The absolute biggest challenge is dealing with myself. Review Signal has been a one-man operation since the start. There are times where I am super motivated to work on it, and I can make a lot of progress. There are often times where I don't want to look at it, and would rather be doing anything else.
After 6 years, burnout is something always lurking around a corner in my mind. I don't want to start a company alone again; it's too much for one person. (Sidenote: this doesn't mean I won't. And maybe that is simply masochism on my part.)
From a technical standpoint, building the whole system has been the largest learning experience and commitment I've ever made. I made decisions early on that I would love to redo, as I've learned better ways to do things. But refactoring code isn't something I see much value in, unless absolutely necessary. It also limits what can be done right now, like going after new niches.
If I had to start over, I'd probably build the system using Apache Storm, which I learned a couple years after launching. I also might consider using and testing some existing libraries versus building everything from scratch.
I would probably try to raise some money and hire another developer to work on all of this with me. I've been entirely self-taught in machine learning and natural language processing. It would be nice to have someone with more experience in the key fields I rely on for architecting the system.
The silver lining of doing it all myself is that I understand everything. The mental model is completely contained in my head — there's no confusion about what any part of Review Signal is doing. This lets me change things fairly quickly without having to explain anything or seek explanations from others.
But it's a double-edged sword, because it means I have to do it. Which means I'm not doing something else.
What were your biggest advantages? Was anything particularly helpful?
One of the few advantages I had and still have is that almost nobody in the web hosting review business is honest. You can look at some really recognizable names in the space (including Drupal.org and WordPress.org), and they are either outright advertisements masquerading as recommendations or the process is so opaque and the recommendations are questionable given the tangled financial relationships between the recommended companies and the people recommending them.
Most of the people who know this space are aware of how bad the situation is, but there hasn't been anything they could recommend. I'm slowly converting influencers into believing and trusting my data. A few even go out of their way to write about the experience. Doing things the right way is my biggest advantage.
Working alone, you need to learn how to be productive, and tricking yourself into productivity is often required. My goal each day was accomplish one thing. No matter how big or small, if I accomplished something, I made forward progress that day and I could feel good about it.
Some days were incredibly productive. Other days, it could be something as trivial as sending an email. As long as I moved the needle any amount forward, I was making progress.
I also want to talk about the amazing coworking space that enabled me to build and launch Review Signal. It was possibly the oldest coworking space in the US, Affinity Lab, based in Washington, D.C. I joined Affinity Lab in 2011, and it became my home in DC and my connection to the local entrepreneurial community. It opened my eyes to how much of an impact real coworking could have. I built my strongest professional and personal relationships there. The coworkers there helped me in so many facets of building and launching Review Signal. I met my lawyer, PR person, and many technical friends who helped me solve problems to name just a few.
If anything was the biggest determining factor on my success, it was being in Affinity Lab. I was surrounded by people who were all on parallel entrepreneurial journeys but shared a willingness to cooperate and help each other out. Surrounding yourself with the right people and the right environment really made building Review Signal possible.
What's your advice for indie hackers who are just starting out?
The first thing I would say to anyone trying to start their own project is have money. If you're going to work on something full time, you need runway, which means money in the bank.
If you're part-timing it, having a day job that can pay the bills is necessary. But I don't like part-timing projects I really want to grow. I find my mind can't split focus that well. If I am working on a client project, because I sometimes do consulting, my mind is thinking about that project even when I'm not directly working on it. Those extra brain cycles aren't being used to further my project, which is where I sometimes get my best ideas.
Of course, having that choice is a luxury. But if at all possible, I want to work full time on my project and try to find a way to make that a reality. To launch Review Signal I moved back home for years to develop it and save money.
Another lesson I learned is that your technology might seem cool, but it really comes down to marketing (unless you're building something for other developers). If your product is good enough, the question is whether you can find customers. These days, I spend very little time on development and probably over 90% of my time on marketing, or something that will help my marketing. Most customers don't care about the underlying technology as long as you're solving their problem.
One of my favorite books I've read is Thinking, Fast and Slow by Daniel Kahneman. It's about Kahneman's research in the field of behavioral economics. It teaches you about how people's brains function and process information.
It might not feel directly applicable to running your startup, but if you're ever selling/presenting/pitching/marketing your product, then you're touching on some of the things addressed in the book. It's not a guide about how to do something specific, it's a guide into the human mind. The better you understand people, the better you'll be able to do a lot of things.
Where can we go to learn more?
I'm happy to answer any questions people might have about Review Signal or startups.
To start it off, I will answer one of the most common questions I get: don't people mostly tweet about a company when they are unhappy/angry/upset?
Yes and no. Every year I spend some time looking back at the data from that year and analyze it. In 2016 I added ~36,000 new reviews to Review Signal. 49.6% of the total number of reviews were positive. So yes, people do write more negative things than positive. However, 52.1% of unique reviews were positive.
What that tells us is more individual people said good things than bad. But the ones writing bad reviews were writing them more often. The other part people worry about is how these tendencies might bias my data. The answer to that is also fairly straightforward: my rankings are relative. No company's customers are inherently more or less likely to write good or bad things. So my system is treating everyone fairly under the same rules; it's simply adjusted to human nature.