eHotel: Some Lessons on How to Pivot in Your Market
eHotel is one of the startups U+ works with: a website where you can book accommodations using benefit points, and it’s the only site in the Czech Republic where you can do so. In this piece, eHotel’s business director, Martin Jílek, discusses a bit of eHotel’s history, how they pivoted, and some helpful tips they learned along the way.
Why we changed
eHotel began in 2012 and until 2016 we simply outsourced all the hotels and only did customer service and marketing. In the beginning of 2016, a few customers asked us about paying for accomodation with benefit points. We didn’t know what this was all about, so we Googled it. This is a form of (tax almost free) benefits given to an employee by the employer that one can use for accommodation, fitness, or medical supplies. There are five different benefit companies, so we implemented one payment gate for each of them. However, our main concern was still in marketing activities where we tried to outdo booking.com in ppc. Eventually we decided to be honest with ourselves and admit that this fight was getting too expensive, and we weren’t really offering anything substantially different. If you have a startup of your own, this is the kind of self-assessment you have to make sometimes. So, in the fall of 2016, we pivoted.
Then the real startup adventure began. We did a thorough cost/revenue/profit analysis. This was a lot of work, but we broke eHotel.cz into individual departments and activities. We compared how many individual activities bring revenue in relation to costs, and these comparisons were measured in the short and long term. The results: benefit users are the best users — we needed our own network of contract hotels. So we hired an army of part-time, performance-paid workers and let them complete profiles of these hotels to sort out which ones were actually good accommodations. After that, we implemented all remaining benefit gateways. And then we made all processes maximally automated and customer-oriented. That left us with the time to make the best UX with a minimum amount of work — and it worked!
Some things we learned:
Don’t wait with automation until it is needed. When you need it, it’s already too late. That’s why we can sustain the same amount of expenditure in customer care despite the fact we doubled our profit within a year. For example: some hotels require payment before the host arrives. This model is ad hoc (on request by hotel) and doing it automatically for all hotels would be a cashflow nightmare. We added one button next to the reservation to compile the info into an email and send it to the account — a ten second job. In about one minute we can now do what previously took us an hour.
Don’t guess: always do market research, have focus groups among your repeat customers, new customers, and potential customers who never heard about you — then you can get closer to your customers and make your marketing better. Don’t do something unnecessary just because it might work; it will eliminate the ideas of assertive people in teams that are not data-related.
Don’t overestimate your abilities. Try to do every process as simply as possible and not bite off more than you can chew because there will always be some hidden flaws along the road that will double the time of actual development/research/acquisition of partners, etc. So don’t make promises/contracts, that you can’t keep
Question everything. The market can be huge, but is it really? We only take one percent of all the people that have benefit points and do transactions with them. Sound good? Definitely, but not all wanted to be accomodated, not all people use their points, and a huge portion can spend all their benefit points elsewhere and never give us a chance.That is significant for marketing campaigns. Never do huge brand campaigns, always customize and target audience.
Have integrity, long term. Use obvious truths that never mislead your customer. Now almost 80% of our customers come back for a second interaction within 6 months.
If data has low quality or quantity, test it. We are always analyzing data, but when the time frame is short or the number of users/interactions is low we need to test. Example: we don’t know which banner is best and which accompanying text: so for one product we launch 36 variations. The 3 best survived and then we boost them. Or we don’t know what customers want, so we A/B test with email questions about their preferences.
