Several cases of conversion rate

Previously in a Online travelling Agency International Hotel channel, I have been doing for 2 years, with the combined efforts of various team roles, the conversion rate increased by 100% in 2 years. For reasons of secrecy, some data may not be true, but it is generally true. Because the case itself is not systematic, it is not divided by any dimension.

These 2 years did a few things, useless things have the following several: Order page revision, details page revision; useful things: Support domestic credit card payment, 10% discount, search optimization; the effect of not knowing is landing page optimization.

Order and details

Among them, the order page, details page revision because there is no solution to any problem is useless. Revision is 4 months after I graduated from the International Hotel to do the first thing, may also be a lot of novice PM will make mistakes (in fact, I really thank the eldest brother to let me do-_-b). On the one hand, I did not recognize the duties of PM and optimize the key elements of the electronic business site, more to do some interactive, visual work. On the other hand, I can find some user concerns, such as domestic users are more inclined to select "Single room" on the details page, "Double room" booking, not on the Home/list page of the selection "one-occupancy", "two-person occupancy"; In the latter process, many users because they do not know the single room, double room simply do not subscribe to (Back in 12, outbound travel has just heated up, and many users are beginners). But this can't change, why? The General Electric Business website has its own backstage, but we use an overseas OTA inventory, direct call API, People's API does not support, we can not do.

Domestic credit card payment

In fact, in doing 2 pages before the revision, I have analysis of the site's conversion ratio funnel, found that the largest gap in the industry average, the result may be the biggest increase is the payment success rate. General Electric Business website will not have this problem, why do we have? Why not solve it first?

As mentioned above, we take the overseas OTA inventory, all by invoking the API. American Credit card usage is very high, so the overseas OTA's payment interface also only supports international credit card, but Chinese credit card penetration rate is low, can pay abroad is less, many users fill in the payment information results returned error. So why don't we localize the payments? This is because at that time International hotel is not strategic focus, resources are limited, technically call Api,roi best.

But with the first two failures, I realised that I wanted to solve the "API only" scenario, but the premise might be problematic. If we are difficult to do in this premise, why not break the premise? I had no concept of technical cost at that time, so I felt that we had to localize the payment process and connect the current payment methods with high user usage in China. This is not a plan that no one has mentioned, but it has not been implemented. This internal cause, external factors plus, including I am also very firm (in fact, just started to think less), is a project.

At that time, the company's strategy to push the credit card does not push Alipay (I actually did not think to accept this strategy, but also to bury the hidden dangers), we will first solve the domestic credit card payment problem. The plan is that we set up a fake international credit card, join the Overseas OTA white list, the user pays us, we pay to overseas OTA. The hole in the process does not elaborate, in fact, as an electrical business system, we do not take into account the different users. For example, this system not only needs to support users, customer service, but also to support the financial and overseas OTA settlement, so the smallest commodity units to be consistent, because at the outset did not consider the issue, and then the need to overturn rewrite.

Finally on the line, the conversion rate has indeed improved. Then many users feedback without credit card, why not support debit card, Alipay? I felt I should support it at first, but I found it again.

As you all know, prepaid hotel, the actual process is "credit card authorization, and so on after the deposit", reflected in the payment system, is "pre-authorized--> order confirmation--> deduction." If the system is reused on third party payment, it becomes the "Order confirmation--> deduction". Most of the time, you look at the electricity dealer's website orders, but is limited to lock the order, and other users to pay, if the user limited time to pay, release inventory. But because the overseas OTA API does not support locking, if we reuse the existing process means to first pay the user the following list, while some hotel inventory is not support refunds (special), we are sure to bear the loss. The project was postponed because of the development of a payment process coupled with various external causes.

10% changes brought about by the discount

After the payment localization project, outbound travel also more and more hot, the company also began to consider how to better do international hotel. At that time although there is no additional resources, but the boss saw that we have no advantage in price, we allow the whole network to reduce prices by 10%. The result has greatly improved the conversion rate! At that time, I began to think about the electrical business site, what is the most important factor, first or commodity. (And, of course, the product manager's values "collapse", laugh)

Although the price is reduced by 10%, in fact, we also have to consider a number of factors: is it directly in the price of the good or the use of Amazon to give discount code mode? Preferential use of direct reduction or return mode (later both use, the reason below)? In fact, this is a good ab testing opportunity, but one technology is immature, second, our base is too small (later to know, in fact, this is also the impact of AB testing test a long reason), so did not do, so in this respect I have no conclusion. But the decision is still to be made, and I have chosen the Amazon discount code (the small company's Non-core product line can make you try it quickly and wrongly). Because the direct subtraction, the user is difficult to have intuitive impression, OTA's list page home results are not the same, how can feel our price is lower (in fact, there are many users OTA search the same hotel to compare prices, so the best strategy can be refined)? Let's just tell the user the 10% discount.

After a 10% discount, the user was happy, but there was a problem I didn't anticipate. The brand hotel always maintains the unified price in each channel, this is for the brand image consideration. Although we do not contract with overseas hotels, but some brand hotels in our website to see their inventory incredibly discounted sales, said to close inventory. That's not good, brand hotels in our sales rankings are in the front. Before that, and because of the limited technical resources, I first considered the needs of users, many other stakeholder needs are not included in the scheduling period. But this incident made me more aware that we have a upstream and downstream, and that we need to balance the interests of both the supplier and the demand side to be sustainable. So how do we give users benefits, but also to help suppliers maintain brand image? We want to communicate with the hotel first, the statement is not a hotel price reduction, but our company "bleeding", but the hotel also does not agree. Therefore, the strategy of returning to the present. Some hotels are not available, there is no way, you can only do not discount to users n times integral. In addition, there is a class of users is the administration, they very much want the user company's money to book, but return to their own pockets, so also asked to support the return. Although I have experienced the early stage of the project, but I did not implement, because at that time I have left.

Search optimization

A lot of user feedback some destinations can not search, so we grasp the user search keywords, the system (in fact, is artificial ah, PM pull a large number of sampling data, in Excel CTRL keyword to compare data in the library, Ah, tears and blood history!) Analysis. No translation (we use overseas OTA API, so many data are in English), many aliases are small problems, the current large popular destination can be poor, even if the manual solution can. The problem of real pit dad is two, one is the particularity of this overseas OTA, another is the general problem of the industry.

Let's start with the first question. The history of Hong Kong is more complicated, the overseas OTA in Hong Kong as a country, but their API does not support the "country" for the search dimension. So, you search "Hong Kong" and return only a certain area of Hong Kong, and you have to search "Sha Tin" and "Kowloon" to return to a hotel in the corresponding area. Hong Kong is precisely the number one area, the geographical division of the strategy to the user's intuitive feeling is "you Hong Kong Hotel really small" (t^t). There are 2 solutions, one is, we pick out the areas under the big Hong Kong, add a "Hong Kong" in front of the 2 words, so in the keyword "Hong Kong" suggestion, we also listed other areas for users to choose. Another is that we associate hotels in other parts of Hong Kong under the "Hong Kong" keyword. Regardless of which solution, the current API does not support, we have to do a local database, the keyword to do a transformation and then to the API interface. Then we have a project, the program 10% Small, done first, and then on the program 2. Said earlier, our base is small not suitable for AB testing, so this time did not do. So the real utility is hard to say, but it is certain that the international hotel conversion rate has been on a flat upward trend for the past 2 years.

The above question triggers us to think about the general problem in the industry, "how to correspond to the relationship between geography and the hotel". For example, in the city level, you search for "Beijing", Daxing is not the Beijing? At the landmark level, return to the center of the landmark X-meter (and, X is a few suitable?) Hotel? or define a landmark as a polygon (who defines? What is the standard?) Return to all the hotels within its coverage? Since we have all been calling APIs before, we show what the OTA gives us, but often there are users who question the results of our presentation. The hotel is too far from the landmark! And so on), we start to think there is no room for optimization?

Through my own experience of going to Thailand, I feel booking.com doing better in this area. For example, Search Sukhumvit Avenue (specifically I can't remember, casually write a description of the hotel, covering the area is much larger than the real administrative divisions, because Booking.com know that there is a fast track through the hand-drawn area, the core of the residential price is expensive, but as long as there is a fast track, I live a little farther. But the problem is too complicated to be supported by a lot of data/algorithms, so the international hotel didn't do it until I left.

Landing page optimization

This is actually our business manager put forward (yes, he has a PM Sense, 2nd, 3 projects are inseparable from his support), because the electricity dealer a lot of money to Baidu, through the search for paid users, of course, to the greatest extent to retain. Then I follow the city page of Booking.com "speculation" the user intention, that if the user through the Baidu big search destination, the TA is probably not too aware of the OTA site of the Novice, TA may not want a variety of people to be confused search term, but tell TA: What's the best way to live in a strange overseas city? This problem, in fact, the site has the answer, but very low efficiency, and the booking process is fragmented, then I think, can we get through this link?

In fact, do not feel that the direction of the big do it, it is likely that the content of the Web site data is not fragmented, and the recommended algorithm is not mature? Raiders of the site are hoping to be able to book the site diversion to earn commissions, is currently looking at the poor travel network and booking.com cooperation, but the specific volume I do not know. But how did we solve it at the time? We selected 10 pilot cities, the Manual (All Tears) collection compiled a check-in guide, can be linked to direct link landmark screening, but the hotel sales list.

Whether the landing page is good or not is not conclusive. Because the conversion rate in some cities is higher than the normal list page, others are below; the rate of jump out is also mixed; As for the increase in the rankings, the impact on individual hotels and other hotels is not measured. And the biggest problem with this page is the difficulty of mass production (the content is all artificially compiled, and must be very accurate, practical to have a real help booking role), and then nothing.

Then my 2-year career ended. I want to do too much, too much, and too little.