Yelp’s financials have been fairly strong lately, and with $550M projected revenue for this year and the aim of getting to $1B (82% jump) in 2017, the company will need to be open and aggressive about new revenue steams. Let’s explore how to generate an additional $100M (conservatively) in yearly subscription revenue that would directly monetize (and further diversify revenue streams) Yelp’s most loyal users, generate data that can improve the overall service, and avoid conflict with existing advertising partners. (note: my sister used to work at Yelp as an accountant, but provided no information or insight for this article.)

Yelp Power User Subscriptions

The basic premise is to provide additional functionality for a subset of Yelp’s users and charge them for it, but using the benefits from their usage to positively impact the service for all.

Personalized Recommendations – Replacing “Best Match”, Netflix Style

The problem with Yelp reviews, as with many review systems, is that actual results are skewed. On a 5 point scale, one might think that 2.5 / 5 would be an average venue. However, for Yelp, the average score is more like 3.8. And while Yelp may want people to be forced to dive deeper (Virtually identical ratings mean people have to dive into reviews to understand what’s different, said Vince Sollitto, who heads communications for the San Francisco-based company.) into reviews because these scores make it harder to differentiate among venues, this is not a very user-centric, empathy-driven approach. In fact, this is better for venues and Yelp – the more venues are better ranked, the more open venues will be to working with Yelp on advertising. The more venues are better ranked, the more people will visit them. This is a clear conflict of interest.

I would like to see smarter recommendations with the option of going deeper into reviews only when I want to. If you have ever used Netflix’s recommendations system, you understand how this could work. As a user creates more reviews, the system is able to predict which others users are similar to that user and provide predicted rankings for new venues. Admittedly, this only works if you and other reviewers have a common set of visited locations and would thus work best in places you live in. However, if you visit a new place, Yelp could use your demographic data to create a profile that may match other users in new locations – there are a number of different approaches to predicting responses without historical data, and this would be a very useful data experiment to create value across all users.

A simple story to explain the need for personalized recommendations comes from a friend. She is Vietnamese and went to Palo Alto in California for Vietnamese food. It was not only expensive ($50+ per person) but terrible. Yet, people in Palo Alto love it and review it accordingly. With a personalized recommendation, I would be steered away from this place despite its positive reviews and to a place that people with my tastes enjoy.

Personalized Recommendations – Incorporating External Data

In addition to predicting scores and using that to sort recommended places for each individual user, Yelp should incorporate external data. While user ratings are great, I also want to know what has been featured on TV (Bourdain) or has won awards from professional reviewers. In many ways, that data already exists on Yelp, created by users – for example, search for “Michelin” and you should find a good list of Michelin-places in that city. These metadata should be officially added into listings. Such locations would automatically receive a bonus in the personalized recommendation scoring or include special badges, and users who value (and visit) them would see more such venues in their recommendations.

Normalized Review Scores

Beyond ranking places for the individual user, scores should be normalized over the last one year of reviews using the full 1-5 spectrum. Ever see complaints about a restaurant that changed ownership recently? Or a place that lowered its quality standards after building a strong reputation? How good is this place right now? Current Yelp review scores don’t take currency into effect very well. I want to create clear separation in order to compare places more easily. How much better is this place than the other place?

While the math to normalize is pretty easy, the process (by area radius, venue category?) to do so is a little complicated and would need to be tested to finalize on format.

Getting More Data – Allowing Data Export and Pure Numerical Reviews

Although Yelp has the most user-review data of any source, it creates barriers preventing additional data that could be used for the product features mentioned above. For example, I am very uncomfortable with Yelp owning my data and making money off of it, thus I would rather write on this blog than for Yelp. I do not need Yelp to share money with me, but I would like to export my data (reviews, bookmarks, check-ins) for myself.

In addition, Yelp forces reviewers to write reviews. I prefer the IMDB method which allows both numerical-only ratings and detailed reviews for those who like to do so. To see the stark difference this can make in conversion and user data, my IMDB history has over 1,100 reviews (average of 70 per year) while my Yelp has 2 after eighteen months.

And More

It is unlikely I will ever be a Yelp Elite because I am not much a Yelp community driver. However, that does not make me a non-active user. I am joining a couple of official Yelp events below soon, but would like to see more, with exclusive slots set aside for paid Yelp users as an added benefit of subscription.

Paid subscribers should have the right to opt-out of ads (but can be on by default) and receive exclusive promotions (offers) for subscribers from businesses. Similar to social media ads on Facebook and Twitter, users would be allowed to vote for or share promotions in order to show interest. As on Google Adwords, advertisers who are just paying to spam users would have to pay more as a penalty for being less relevant. This would create a win-win scenario for both users and businesses who truly care.

I would like to see the ability to review individual plates or meals, not just the venue. Not everything a place serves is equal in quality, and I would like reviews to be broken down into smaller components such as service quality. Some styles of restaurants are affected by lower grades of service, but often I do not care about that. I just want to know what is the best food for a best price, and there is no way to quickly determine this. Yelp should be making this possible!

Revenue Forecast

This service would be provided at $5 per month or just $30 per year for annual subscriptions. Imagine the $5 per month as a perfect solution for travelers visiting a new location (ex. 4 day trip in Chicago) and needing to know the best places specifically for them. It’s not too much different from buying a travel guide. Perhaps only yearly subscription users would have certain features such as the history export, but I think that numerical-only reviews should be opened to all. I use $30 per year as a personal preference that seems reasonable to me but also as a stark contrast to paying the per month fee ($60). Fees would be due at the beginning of any subscription period, providing Yelp instant cash flow, but could be refunded at a pro-rated level. Yelp Elites could be given free subscriptions.

Yelp currently has approximately 150 Million Users (including international markets). To reach $100M in yearly subscription revenue, just 2.22% of these users would need to subscribe – I believe (based my own experience in social networks) that this number could reach 5%. Please note that I have simplified the calculation, not accounting for regional user / wealth populations, single month purchases, future growth, mobile vs. desktop, and new ad product growth for subscribers, etc.

If you are thinking you would never pay for such features, that is ok! You are one of the 98% who would not need to. However, I am one of the 2% who would. 2 out of 100 people is fairly low on the requirement side.

Stakeholder Impact

Since Yelp is trying to reach $1B in revenue in two years, they clearly are concerned about their existing sales, which has been slowing in growth the last few years. Although paid subscribers could turn off ads, by keeping them on by default, Yelp would reduce impact on the ad impressions removed. Normalized Reviews could impact businesses, but this would only be available for subscription users and would be a complimentary score to the existing system – most people could still remain confused (yay!) by the overly positive Yelp review system. Yelp’s current display of Google Display Network ads would be minimally affected.

A great benefit of reducing the review barrier and allowing numerical reviews is providing more data that can be used to promote businesses, which in turn helps businesses. In particular, this would help smaller businesses with less than 100 reviews because they have the most to gain from more reviews. (If you are concerned about fake reviews with the numerical-only system, there are different ways to filter and normalize that data as well) Offering advertising access to paid subscribers also creates new revenue opportunities for Yelp and focused opportunities to improve the perception of the business. Paid subscribers are more likely to review and create content for a business and Yelp helping businesses get subscribers in the door first is a more cost-effective method to seed business perception.

Recap and Conclusion

Here is a recap of my proposal:

For All Users:

1. Enable numerical-only reviews, with breakdowns for specific aspects, such as service quality and food quality (but not required)
2. Enable dish-specific reviews, numerical and tagged text reviews
3. Enable personalized recommendations, IMDB-style, based on past review history and incorporate external data such as Michelin and TV mentions – do not show predicted ratings

For Premium Users:

1. Normalized reviews for easy comparison, including recency data
2. Show predicted ratings for personalized recommendations
3. Op-out for ads
4. Exclusive targeting from advertisers for promotions, using Google Adwords and Facebook style feedback to penalize spam
5. Subscriber-exclusive invite slots for official Yelp events
6. User Data Export

Revenue:

1. Conservative estimate of $100M in revenue (150M users * 2.22% * $30 /year / user)
2. Not including monthly one-time payments for “tour guide” like service
3. Long term potential of $225M (even with no further growth of userbase)

Yelp is in competition with Facebook, Foursquare, Google and others for local advertising dollars. Despite Yelp’s data trove, it can do more to get more data as well as create more value to its users through that data. Over the long term, this would create more loyalty lock-in to the service, even without forcibly locking users in (as it does now). 

I welcome all comments and thoughts below!