After working as a VC for a long time, I started an online travel company on the premise that most people want to transact on the Internet. Not only it is cheaper than the brick and mortar business model, it also provided ease of use and do it on your own time.
With that premise, I started out to first learn complex details of travel booking technology components such as GDS, vendors XML or Web Services API, pros and cons of open source vs. Microsoft technologies, databases and a multitude of programming issues that crop up when you are building an enterprise-level system able to handle thousands of daily user searches on a website.
By the time we built and launched the portal, I was already familiar with marketing concepts such as SEO, PPC, Email and Affiliate marketing and various strategies to make each of these techniques to build large marketing campaigns. At one point, we were showing in SEO results on the first pages of Google and Yahoo’s search results for thousands of keywords above much bigger names such as Expedia, Priceline and Kayak’s of the travel world.
We spent days, weeks and months to tweak our website to make sure that the search results were returned under four seconds, the user flow was simple – 3 clicks to do a transaction, add filters and sorting tools to help user make a selection easily, allow user to search +/- 1/2/3 days of the travel date, change the search criterion from the result page, able to fly or stay in a nearby airport or city, used heat maps and collected data of users who left from the cart and make sure that the transaction NEVER fails as travel inventory is a very dynamic. Surprisingly to us, it took mountains of effort to improve by one-tenth of conversion.
We put in pop-up and pop-under banners and other ads to monetize the traffic. We started to advertise on meta-search engines (these websites showed results from multiple travel websites) assuming that the traffic, which will reach us, is already much more qualified and hopefully improve our overall conversion. We met with Google folks and put custom ads for users leaving the site to monetize each visit.
Whatever we did, it hardly made a difference in our conversion rate. The more we spent, more we lost. It seemed like moving mountains was easier than to improve the conversion rate by one whole point. Middleman always made money on our dime. Search engines, Metasearch providers, companies building tools to measure, companies helping to provide a global view on spend and marketing operations all seemed to thrive. From the outside, we looked big doing a lot of transactions. All vendors wanted to work with us as we bought a whole lot of traffic and it was easy to hire the best employees as we ran large operations. But more business we did, the more money we lost. In the first 18 months of large marketing spend, our losses added up to millions. Building a profitable business over the Internet seemed to be more myth than reality.
As we did more volume, it generated more service calls and we had to employ more and more customer service agents to handle customer service queries. We got to a point where we had filled all the open office spaces with the customer service agents and we could not hire any more customer support agents without renting a new space. As the cost of customer service agents became the largest part of our overall payroll, we contracted a small customer service company in an overseas location just to reduce our cost to handle customer service queries. To build a right call to customer agent ratio, I decided to spend a week listening to see if we could send some calls to voicemails so we can get back to them afterward rather then take agent time to answer them live. We asked our agents to create a list of issues customers were calling us about. Speaking with agents and listening to the customer calls, an interesting pattern emerged. As agents were trying to handle more customers to reduce the hold time per customer, agents asked the customer to go back to our website to book or call the vendors directly with any issues.
Agents gave us a list of issues where they were mostly not helpful to the customer. These issues ranged from confirming customer itinerary, provide a vendor phone number or a vendor policy, which customers could confirm by either going to our or vendor’s website and so on and on. As we made these changes by creating new prompts in our phone system, we realized that up to twenty percent of our calls were from customers who needed assistance to either make a reservation or help them with issues where they were willing to pay us lot more than if they booked from the website. An additional ten to twenty percent of customer calls were to cancel or rebook existing transactions.
We changed the whole phone system to bring new reservation calls to our best agents to make sure they provided extremely good service and we could justify charging a premium for these services. To our surprise, customers were willing to pay 7.5 to 10 times higher in service fees than what we were adding on our website for a transaction. Working through agents we further segregated the calls where certain reservations did not make us enough money and we asked those customers to reach out to our suppliers directly. Once we were done finalizing the customization of call segregation, we had found a winning solution. We were able to scale the formula to hundreds of agents to build a profitable company.
We further looked for tools to make the whole system extremely efficient. We integrated Google, Bing and Yahoo APIs directly into our CRM so we can see our marketing spend and visits in real-time from a campaign, ad group and even at a keyword level. We also added a unique phone number to a campaign or an ad group and most searched keywords. Now we could see which keyword, ad group or a campaign was generating the most calls and online transactions in real-time.
On the other side of the equation, we added the PBX API to the CRM. By doing that we could tell which calls generated from a Campaign, ad group or keyword were converting into a transaction. By doing that we could change the marketing spend to the converting keywords and minimize our marketing spend and maximize our conversion ratio and revenues in real-time.
Since we integrated our PBX API with our CRM, we could also see the agent performance over a day, week or over a month time frame. This allowed us to put our most converting agents to be available for the best converting calls by creating a priority system. The system automatically adjusted agent priority by looking at agents’ weekly conversion ratios on a daily basis. This made the overall system into a science. We could spend more on most converting campaigns and put the best agents to help these customers who in turn paid more to utilize our resources. We measured and created simple metrics for cost per call and revenue per agent to incentivize marketing and sales to cross these thresholds.
Today there are stand-alone tools available to do all these measurements and management. But at that time we had to create all these integrations and matrices as needed to solve each puzzle. But none of this was possible until we had our customer service agents guide us as to how to help them answer the calls, which used their time best to help our customers.
Welcome to Utkarsh Kansal's Business Blog 