Evaluation of the store in real Internet methods

The online store is fairly easy to analyze in terms of visitor statistics and user actions. To see what the client is doing, there is always Yandex.Metrica and Google Analytics that help dig up the preconditioning, there are mousetracking and eytrecking services, there are own scripts that take into account the customer’s movement path through the site and generally “hold their hand”.

Shops in real life do not provide such luxury, but they need to be researched , because it increases profits. In the topic of the story about how we do it.


')

1. Counting customer flow

The first thing we started to count from the start of the first point is the number of customers, the average check and the ratio of visitors to those who are interested in shopping. It turned out to be quite simple: half of the purchases are counted by 1C, and the seller just behind the cash register counts just entering the store (or, if the store is large, an automatic counter at the entrance, such, by the way, are in many shopping centers).

Conversion sales varies by season. It is also very influenced by the actions of sellers, display and other controlled parameters. If the conversion comes out of its seasonal corridor by more than 10-15 percent, then something is wrong somewhere and you need to fix the jambs. If it suddenly takes off, you need to go and see what has been done in the store in order to deploy across the network.

2. Reason Why

The second source of information is internal notes to orders, or rather, “Reason Why” for each position. We sell board games, toys and souvenirs - and therefore we want to know who and what is more suitable. Most customers say who they are buying for (for example, a friend’s businessman, 12-year-old son and so on) - and this data is impersonal in the database for a particular game.

Analogue in the online store - viewing of traffic patterns on the site and the formation of new navigation based on the data, plus lists of "recommended products".

3. Questioning

In December and January there were huge queues in the stores, where it was boring and uninteresting to stand. We laid out the 52-second questionnaires and pens - and the clients shared their opinions with us, chose their favorite game, plus gave other information. We even met a resume in a couple of questionnaires, offers of cooperation even in a couple.

Questionnaires gave not only a lot of data that could not be obtained by other means, but also allowed us to tell customers about a number of store features that they did not know about. For example, the checkbox "I know that games can be ordered with delivery," said "people from the street" about the presence of an online store.

4. Analytics

With the segmentation of games and sales in real time in 1C, you can build graphics sales of games at different times of the day. For example, it turned out that children's games are mainly bought in the morning and party games at the lunch break and in the evening. The range of sales for the weekend is very different from the range on weekdays.

At the exit, we received data on the optimal time of mailings with offers on the client base, plus data on when it is better to show which offers and goods on the main (now this is not, but it is planned in the next version of the site).

5. Analysis of groups in social networks

We have a large community Vkontakte. Thanks to the analysis of the profiles of participants in our meetings and groups, we realized what kind of hobbies you should pay attention to: it turned out, for example, that among “ours” there are an unrealistic number of fans of art-house cinema (59%) and IT people (32%, half of them - from Moscow State University). This is a direct reference to the style of events held and the style of presenting information about goods. It is clear that the selection of VC captures only one segment of the audience, but this data turned out to be very valuable. From there, you can easily gather data on marital status, number of children and other things (which people for some reason write in their profile), which give precise targeting.

6. Aitracking and Routing

Finally, the last stage of observation is the passage from the beginning to the end of the client. This method is described in detail in Paco Underhill in books . In fact, you need to stand at the point of sale and, without attracting attention, monitor the activities of the client. This is similar to aytreking. A special person logs where the client stopped in the store after entering, where he went, what and in what order he looked, when he turned to the consultant, what games he opened before buying, how much time was spent on each operation, what caught the eye in the display, and so on.

We make such observations when opening points in the shopping center, laying out the games that most attract the attention of the passing stream, we track bugs (for example, children's games, to which young customers do not reach), “scaring” blocks (erotic games that look young mothers with child) and other possible glitches. As a result of such observations, as a rule, a number of improvements in the usability of the store appear at once - as well as after a good analysis of eytreking.

7. Macro level

In the west, there are even companies that build customer movement paths through shopping centers (based on tracking objects on video cameras) and make new projects based on this and determine actions in a particular building. For example, it is now known that the centers of the first generation simply drove the customer to the floors by a strange arrangement of stairs, the second one drove people to galleries with tasty and interesting points on the top floors, and the last ones were driving visitors to the ring with galleries. it is impossible to resist the heap of other temptations.

As you can see, it is possible to analyze stores in real life with the usual network methods: this is interesting and very useful for increasing sales. By the way, the analysis needs to be started from the very beginning: in our blog on Habré there is, for example, about choosing a place for a sales point .

Examples

UPD: at the request of the comments added case studies examples:

• The flow of customers. Every day we track point conversion. At the time of the beginning of experiments with the assortment, there are two control points (for evaluating normal changes) and three test points, where different new assumptions are made about how to put the product and which one. For example, we learned that laughing tigers do not go, but, for example, yo-yo - very positively attracts children from the whole area. Given that the yo-yo is positioned as a youth, it was also a surprise.
• Reason Why. A number of positions that we considered childish, went to the party list - and vice versa. We were surprised to learn that, for example, Monopoly is a gift to a 12-14 year old boy, Jenga is the best game for a girl, and the Dragonfly helicopter is bought by wives for IT-husbands. The result is more accurate segmentation of the target audience and more accurate case studies.
• Questioning. From the questionnaires, we learned how many games the person has at home (and compared them with their favorites). It turned out, for example, that there is a certain limit for different types of games that does not last very long: a person, buying an N-piece, becomes an expert and the soul of the company - and starts taking more and more. This is good data for the loyalty program, in particular, the thresholds for the cumulative discount. By the way, the best-selling game of the year by the results of the survey coincided with the most beloved.
• Hand Tracking. Gives knowledge about the best games in the showcase, corrects shoals calculations. For example, it turned out that the child's “Operation” needs to be laid down - the advertisements are strong, and the children are drawn to it when their parents do not even see it. But, for example, towns and birdhouses should lie strictly in baskets or at belt level, so that retirees can buy greetings from childhood to their grandchildren. Although I already know that the birdhouse is still young.
• Group analysis in social networks. We learned about MSU and went there with sponsorship and prizes. They found out about the arthouse - they collected several animators from this environment.
• Analytics. We compare the pair products - we know about the best offers for something. For example, after buying poker , Uno is often taken. In general, there is a real expanse for different data mining, so you can dig for years and constantly make something new.