Simplify or die. What should be the product for the Russian E-commerce?
Hello friends. A little less than a year ago, I started working in a new b2b project for online stores and wrote an article about this startup on Habr. Today I want to talk about our main mistake over the past year , which led us to change the principle of the service and launch the version 2.0 of the product in October of this year.
HucksterBot is a service that shows individual visitors to the site individual offers to buy goods at a discount. The main difference from the traditional sales mechanisms and standard modules of CMS-systems is that discounts are shown personally, depending on the user's behavior on the site. With this approach, the profit of the online store is not reduced, but increased.
Previously, in addition to installing the JS code on the site (standard step for SaaS services), in order to start the service, a representative of an online store in your account needed to create customer segments and set discounts for certain categories and products. To configure the segments, you could select regions, channels and traffic sources, devices (PC or mobile):
')
Users of each segment showed a special widget with an offer to buy previously viewed products at a discount. Thus, discounts could be offered to visitors from certain regions, mobile users (which increased conversion and sales in online stores, whose sites were not adapted for mobile devices), or, for example, to people who came through advertising channels.
This approach implied that marketers or owners of online stores are well versed in increasing conversion, can correctly interpret Google Analytics and Yandex.metrics data, select and configure several low conversion traffic segments and independently achieve the maximum effect in the form of increased turnover and gross profit. That was a big mistake.
In fact, not everyone was able (or wanted) to deal with the product and we had to do it with them, to help users customize the service. Yes, in the end, the service started to work efficiently, we and our customers made success stories with excellent results , but we only had to dream about scaling with this approach. We spent too much time analyzing and deploying.
It became obvious that we have 2 ways for development: the first is to raise prices, go to Enterprise sales, actually becoming implementers of an expensive product for large online stores with individual doping and accounting, and the second is to significantly simplify the product without reducing its effectiveness . We chose the second option, but for its implementation it was necessary to solve two big tasks:
1. Automation and the Pareto principle.
Now, to configure the widget, you need to specify only the size of discounts and, if necessary, limit the impressions in the regions, everything else is configured automatically. The system keeps track of events when the user puts the goods in the basket and analyzes the conversion for each product, taking into account traffic channels.
In the display of discounts, the Pareto principle is used: 20% of products that give 80% of applications are automatically selected, discounts are not shown for them. Thus, the service does not work with goods that the store sells successfully, but increases sales only for those items that are not sold at all or are sold much worse than the top ones.
2. Transparency of work.
The work of the service begins with the fact that it collects and analyzes data on the main part of the online store traffic, showing the widget for the indicated discounts to only a small fraction of visitors without any differentiation. After collecting and automatically analyzing data (from 1 to 2 weeks), the service starts operating in the continuous test mode: 20% of randomly selected visitors of the online store will not see the widget at a discount under any circumstances, the rest 80% will be distributed rules for displaying personal discounts. Thus, in your personal account you can always see how much the conversion of the net segment (20%) differs from the segment with the working HucksterBot service.
After the release of version 2.0, the number of users who independently configured and implemented the service increased by 100% . Only one question bothered us.
We decided to test the effectiveness of the new approach when using the service with the online store Snail.ru, which belongs to one of Russia's largest retail chains selling Garden Machines Garden Machines.
According to the Pareto rule, the system automatically selected 20% of the goods that are the most sold, the service did not apply to them. For the remaining 80% of products, personalized discounts worked, the size of which was set by the store employees.
Then began the A / B test, which reached its maximum statistical significance within 2 weeks. 20% of random visitors to the online store did not see the HucksterBot widget under any circumstances, the remaining 80% could see it with a certain behavior on the site.
Test results
The overall conversion of the site when implementing HucksterBot 2.0 (orders through the cart + widget) was 1.9% .
The conversion of the site into orders without a widget (according to the A / B test, as well as according to Google Analytics, when the service was not connected to the online store) is 0.8% .
Thus, the use of the discount personalization service allows you to increase the overall conversion of the site by 2.4 times .
Analysis of the store's sales showed that when using the HucksterBot service, the store's turnover increased exactly 2 times (by 100%) , and gross profit 1.4 times (by 40%) .
Thus, if you run a b2b startup and dream of a mass market, the product should be as simple and automated as possible so that its implementation does not require any special knowledge. In addition, it is important to prove its effectiveness with real numbers and graphs, calculating ROI, and not with promises and words.
In the next article, I will share the experience of promoting a SaaS product for online stores on the Russian market: I’ll tell you which advertising channels are paying off, which aren’t, how not to drain the budget and attract registrations for less than 200 rubles per item.
If you have questions, I will be glad to answer them in the comments.
HucksterBot is a service that shows individual visitors to the site individual offers to buy goods at a discount. The main difference from the traditional sales mechanisms and standard modules of CMS-systems is that discounts are shown personally, depending on the user's behavior on the site. With this approach, the profit of the online store is not reduced, but increased.
Previously, in addition to installing the JS code on the site (standard step for SaaS services), in order to start the service, a representative of an online store in your account needed to create customer segments and set discounts for certain categories and products. To configure the segments, you could select regions, channels and traffic sources, devices (PC or mobile):
')
Users of each segment showed a special widget with an offer to buy previously viewed products at a discount. Thus, discounts could be offered to visitors from certain regions, mobile users (which increased conversion and sales in online stores, whose sites were not adapted for mobile devices), or, for example, to people who came through advertising channels.
This approach implied that marketers or owners of online stores are well versed in increasing conversion, can correctly interpret Google Analytics and Yandex.metrics data, select and configure several low conversion traffic segments and independently achieve the maximum effect in the form of increased turnover and gross profit. That was a big mistake.
In fact, not everyone was able (or wanted) to deal with the product and we had to do it with them, to help users customize the service. Yes, in the end, the service started to work efficiently, we and our customers made success stories with excellent results , but we only had to dream about scaling with this approach. We spent too much time analyzing and deploying.
It became obvious that we have 2 ways for development: the first is to raise prices, go to Enterprise sales, actually becoming implementers of an expensive product for large online stores with individual doping and accounting, and the second is to significantly simplify the product without reducing its effectiveness . We chose the second option, but for its implementation it was necessary to solve two big tasks:
- Automate complex settings (so that even the housewife can start the service).
- Make transparent and understandable the results of the service, so that each client can go to your personal account at any time and find the answer to the question “what will I lose if I turn off the HucksterBot widget?”.
How do we solve these problems?
1. Automation and the Pareto principle.
Now, to configure the widget, you need to specify only the size of discounts and, if necessary, limit the impressions in the regions, everything else is configured automatically. The system keeps track of events when the user puts the goods in the basket and analyzes the conversion for each product, taking into account traffic channels.
In the display of discounts, the Pareto principle is used: 20% of products that give 80% of applications are automatically selected, discounts are not shown for them. Thus, the service does not work with goods that the store sells successfully, but increases sales only for those items that are not sold at all or are sold much worse than the top ones.
2. Transparency of work.
The work of the service begins with the fact that it collects and analyzes data on the main part of the online store traffic, showing the widget for the indicated discounts to only a small fraction of visitors without any differentiation. After collecting and automatically analyzing data (from 1 to 2 weeks), the service starts operating in the continuous test mode: 20% of randomly selected visitors of the online store will not see the widget at a discount under any circumstances, the rest 80% will be distributed rules for displaying personal discounts. Thus, in your personal account you can always see how much the conversion of the net segment (20%) differs from the segment with the working HucksterBot service.
After the release of version 2.0, the number of users who independently configured and implemented the service increased by 100% . Only one question bothered us.
Is the service effective?
We decided to test the effectiveness of the new approach when using the service with the online store Snail.ru, which belongs to one of Russia's largest retail chains selling Garden Machines Garden Machines.
According to the Pareto rule, the system automatically selected 20% of the goods that are the most sold, the service did not apply to them. For the remaining 80% of products, personalized discounts worked, the size of which was set by the store employees.
Then began the A / B test, which reached its maximum statistical significance within 2 weeks. 20% of random visitors to the online store did not see the HucksterBot widget under any circumstances, the remaining 80% could see it with a certain behavior on the site.
Test results
The overall conversion of the site when implementing HucksterBot 2.0 (orders through the cart + widget) was 1.9% .
The conversion of the site into orders without a widget (according to the A / B test, as well as according to Google Analytics, when the service was not connected to the online store) is 0.8% .
Thus, the use of the discount personalization service allows you to increase the overall conversion of the site by 2.4 times .
Analysis of the store's sales showed that when using the HucksterBot service, the store's turnover increased exactly 2 times (by 100%) , and gross profit 1.4 times (by 40%) .
“We launched the new version of HucksterBot 2.0. If earlier we had to constantly experiment with different segments, adjust the settings in order to achieve the maximum effect, now the service does this automatically. It is very convenient"Stepan Maltsev, Snail.ru project manager.
Thus, if you run a b2b startup and dream of a mass market, the product should be as simple and automated as possible so that its implementation does not require any special knowledge. In addition, it is important to prove its effectiveness with real numbers and graphs, calculating ROI, and not with promises and words.
In the next article, I will share the experience of promoting a SaaS product for online stores on the Russian market: I’ll tell you which advertising channels are paying off, which aren’t, how not to drain the budget and attract registrations for less than 200 rubles per item.
If you have questions, I will be glad to answer them in the comments.
Source: https://habr.com/ru/post/314564/
All Articles