The List Efficiency

BY InVine Social

On one of previous posts we analyzed the influence of having cheaper wines on the list and whether that drives consumer behavior, particularly if there are people who will choose the Second Cheapest wine on a list without considering any other factors. One of the interesting questions that were raised was how the list size might affect the consumer’s choice.

We continue the analysis on our data, this time looking for the answer to the question: “What impact do large wine lists have?”

Here’s the report.

The question we are trying to answer is incredibly challenging so in this analysis we made a few simplifications:

  • We assume that if a customer takes less time to pick a wine, that’s better for the restaurant, not only because it’s less wasted time, but also because we assume that there are enough wines on the list and enough information about the wines for the customer to select the highest price he’s willing to pay;
  • We assume that if a customer takes less time to pick a wine, that’s better for the customer (i.e. better satisfaction since it was easier to pick, plus it took less time);
  • Last, and most important, it’s quite hard to compare the efficiency of different list sizes, due to the fact that lists vary a lot in their composition. We will try to infer conclusions regarding list size, by comparing customers who browse different number of wines, on large lists.


How many wines

We started by analyzing how deep each user browses through the wine list. We found that 50% of users see 10 wines or less on the list. Due to the very long ‘tail’ of usage (i.e. only a very small number of users are seeing a large number of wines) the average is a bit higher, still at 13 wines seen.

The chart below shows how many patrons are seeing a particular number of wines.

Seen Wines

These are very positive numbers in terms of the efficiency of the wine list. The data is originating from wine lists with at least 150 wines, some with more 1000 wines, which means that customers are not loosing a significant amount of time searching for wines.

To further analyze the efficiency of the wine list, we took a look at when customers see the wine they end up selecting. The results were once again very positive: 50% of users pick one of the first 5 wines they saw, which again seems to point to the fact that the wine list size/efficiency is working well.


Selecting versus Discovering

Before going into price dynamics, we decided to explore the usage better, specifically to break down two different browsing modes:

  • Selecting: When the user is actively searching for the wine he will pick. Price should be a major influence point here. Customer satisfaction will come mostly from the ability to pick a wine as efficiently as possible, and to be happy with that pick.
  • Discovering: After the user has selected his wine, he will wait for the waiter to come. During that time the user will continue to browse the list. Customer satisfaction will be about learning and getting interested about the list and the wines.

Since we can’t compare the total time spent in each mode, because the Discovering mode is limited by the arrival of the waiter, we looked into the time spent per wine.

The following chart shows, for each number of wines seen, the average time per wine, for users that fall in that category.

Time per wine

We can see that before selecting a wine, a user spends around 30 seconds browsing each wine. As we previously saw, most users pick one of the first 5 wines they see, which means that around 2.5 minutes into using the list, the user already has seen which wine they’ll pick. This is fast!

In fact, on average, users take 4.5 minutes to actually select a wine. This means that between 2.5 and 4.5 minute marks, the users are browsing the rest of the list, to see if they find a better pick. The interesting thing is that these values are approximately the same for people who browse a few or a lot of wines, which suggests that the InVine digital wine list allows for larger lists to be browsed the same way as smaller ones.

After selecting a wine, the user starts his discovery process. We can see from the chart above, that the time per wine increases significantly (more than double). However, we can see that both these metrics don’t change materially when we compare users who browse a few wines with those who browse a lot of them.



Our expectation here is that the issue of list size is not only about efficiency, but also increasing the odds that the customer will pick that ‘one’ wine that suits his taste, needs, budget, etc. Larger lists provide customers with more choice, which hopefully is better for that, and not just more confusing.

The following chart shows, for users categorized in the number of wines they saw, the price of the wines they saw and selected. The data has been smoothed, the original data is represented by the dashed lines.

Number Seen vs Price

The most straightforward conclusion is that browsing more wines increases the average price of the selected wine (blue line). For instance, people who browse through more than 10 wines select wines which are, on average, 10% pricier than people who browse 5 wines. This effect seems to show our expectation is valid.



Overall the list size has two clear consequences:

  • Time spent searching increases proportionally to the number of wines seen (people who see more wines, don’t do it faster).
  • More wines means a higher chance of providing best value to customer (but there will be a limit where the customer could “get lost”).

Sadly, at this point we can’t compare the effect of having different list sizes on one customer. However, there are a few related analysis we want to add to this later on, such as:

  • Cost of stock management: we didn’t consider cost of holding stock (i.e. unlimited storage space, or costless and fast delivery of stock replacement);
  • Customer satisfaction: is a customer less willing to come back if his/her meal was more expensive because of wine? This prompts the question of whether restaurants should maximize the revenue from their current meal, or from the aggregate of “come-backs” and “referrals”

Another insightful analysis from our LAB Team!

I’m still curious on whether we can actually try to compare the experience from customers who browse large and small wine lists. Maybe have customers tell us their satisfaction? Let’s come back to this subject soon!