Reading time: 6 min

Who is the Analytics Translator? What is the perfect chemistry for this complex role? What skills and experience are needed to do this job effectively? Where is this role in the organizational structure?

Eszter Somos, Data Solution Advisor at Hiflylabs, answered in the interview.

Hiflylabs: What is the specific purpose of this role?

Eszter Somos: In fact, he/she is a mediator with expertise and business knowledge between people who speak “other languages” in two very different “worlds”. His/her goal is to maintain active communication between the two parties and to find out as soon as possible if something isn’t going to work because that way he/she can reduce frustration on both sides.

HL: What makes a good Analytics Translator?

ES: I see two typical options that work well. One, and also my way, when someone came from the technical side, did coding before, worked as a Data Scientist, so he/she fully understands what is possible, how long it takes to achieve the goal, and how exactly the implementation works. In this case, he/she knows the way of thinking of his/her peers and in the meantime, he/she gets to know and learns more and more about the peculiarities of the business side. The other, I think more common version, when business people go to a Data Science course or at home, delves into the subject in a self-taught way. This is not to say that they can code in practice, they may not (yet) have the statistical knowledge – which is otherwise needed to become a Data Scientist – but they can understand what can be achieved by data and what exactly it entails. So the common practice of placing a general project manager in this role is particularly risky and results in cumbersome operation. have more than once encountered the phenomenon of placing a general project manager in this role. I’ve only seen this work outstandingly once – the reason behind had to be the person’s special talent – but for the majority, it formed an extra layer of confusion in the process.

HL: From a business side approach, how deep your knowledge should be to fulfil this position?

ES: It is essential to deduce or code the problem mathematically, but it is important to know what data types, algorithms, technologies, model types there are and which ones to use. For example, it is good to know what a neural network is for and when to offer this option, or what a regression is for and when you can get good results by using it.

HL: What does this “Translation” job include? To what level should the business problem be “translated” into data language?

ES: If not entirely to the level of a database column, but I think you need to move the typical business questions (e.g. “How our users behave” or “We want to know our users better”) to a more specific level, e.g. Could dividing the marketing target group based on looking at and buying products in carts be an output that the business can use?”. That poses a clustering problem for the data team. 

HL: What is the most important skill for an Analytics Translator?

ES: Design thinking for data science is becoming more and more fashionable, which is a buzzword but covers meaningful things, emphasizing the need for empathy, for example. I think from a technical point of view, this is the key because the business often feels uncertain, as it doesn’t understand several things: what exactly is happening; why the process seems lengthy; what it means to be unreliable or probabilistic. It encounters a lot of uncertainties, from a data point of view, for example, it is natural that due to the statistical nature, it is not possible to say everything in advance or that a lot of data cleaning is required. From a business perspective, these things are not clear, so you need to be sympathetic about the business problem and focus on the solution.

HL: Do you also need sales skills in this role?

ES: Absolutely. He/she should help the client to believe that the solution he/she offers will solve the client’s problem. But often this kind of support is also needed within the company. Analytics Translator is also “expectation management” as you need to know what to expect from the two sides.

HL: What are the specific tasks for an Analytics Translator?

ES: There are business problems that are only articulated at a basic level: for example, our costs have risen too much, or we don’t have enough loyal customers and we want to have more, don’t want to make a process worse, or we just want to work more efficiently. In comparison, I would put the analyst questions at a different, specified level, because a problem with them, for example, is formulated as follows: from which variables can I predict how that other variable will change and what influences it the most. So it is a very typical task to turn a business question into an analyst question. You need to clarify unclear issues. You need to break down the influential factors further, find out what we can control and how we will measure it all, how the project can work, what input variables can be, what is at the data level and is available. 

HL: What are the goals of an organization with an Analytics Translator?

ES: Last year, Gartner released very frightening statistics and predictions, such as that by 2022, only 20% of analyst projects will be commercially successful. I believe it should be an organizational goal to be transparent about what value the data solution actually drives. This is important because these are often not clearly measurable things that are done even slowly. An organization that has a couple of Analytics Translators can make their projects more likely to materialize and bring business value. You can minimize cases where the project is completed, those who worked on it are paid, but the result obtained is useless to the business.

HL: Where is this role within the organization?

ES: Very few organizations have this role explicitly. Many times he/she is close to the project managers, as just like them, he/she is a bit around the project and yet not. In a larger organization, you definitely need to go very close to the business to understand what the problem is.

HL: In addition to improving organizational decision-making and business processes, what made this position exist?

ES: They have been involved in machine learning for a very long time, however, there was a point where the majority not only noticed that they had a lot of data, but also that it was easily accessible and could produce ever-increasing results. After that, many companies had a Data Science team, but it soon became apparent that it wasn’t bringing what was expected of it, in fact, it would even cost a lot, even though they said, “if you have that much data, the business will go better”. This has given rise to a great deal of frustration, for a variety of reasons. So as a response, we got to the Analytics Translator position, which is a bit like a step when we go to couples therapy and learn to communicate better together, to plan together.

HL: What dynamics does this role provide within a company?

ES: It’s not true that there won’t be problems, but they will surface much sooner. We first know what pitfalls may await us, where we should not go any further. In my opinion, these would definitely show up after a while, so there is no need to be afraid of that either. It’s important to note, though, that with an Analytics Translator, the process won’t be fast, and maybe even more frustrating at first, if someone keeps asking questions, but to shorten the process and later, while doing them, all the participants will be satisfied, is necessary. It saves a lot of energy in the long run.

HL: In a data environment, what is beyond the responsibility of an Analytics Translator?

ES: Areas where the problem is very well defined, such as the area of ​​IoT. The more technical the use, the more detailed the problem is defined, the less the role of a Translator is. For example, to stop a car at a stop sign or to build a controlling system for a company, you don’t really need that extra role because everyone understands what to do.

HL: Do you think this role will improve?

ES: Sure, but I don’t know where. If this growth continues in the data world, more and more people will likely specialize in this task.

Authors:
Eszter Somos – Data Solution Advisor
Patrícia Hanis – Marketing Assistant

 

 

Reading time: 4 min

Business problem

Retail pricing is one of those fields that can greatly be supported by data analysis, with measurable, high returns on investment. In many cases it’s simply not possible to set different prices on product level – due to missing data or technical limitations. In such situations pricing may be differentiated by location or by time period.

As an example, there are many several days and weeks in the summer whose sales are influenced by transit customers when the shops can expect less price elasticity and react with appropriate price changes.

Our client is a multinational company in the CEE region with partial retail focus, owning and operating a store network consisting of 1000 shops. With regards to convenience goods sold, the selection and definition of re-priced products and the affected store locations was done by Hiflylabs.

Let’s get to the data!

The basis of our solution is that price-changes will affect not all products and not all stores, as a general price increase can negatively impact sales.

Where to re-price?

The target stores were selected by checking which locations experience a growing ratio of transit customers in the summer season which indirectly decreases price elasticity.

Shops located near touristic, waterfront and high-traffic areas were selected in our first simulations which was in line with our assumptions, however many unexpected shop locations ended up being included in the project.  

What to re-price?

The included products were ones that experienced higher-than-average sales rates in the summer season, such as energy drinks, water or ice cream.

What to check our results against?

Using last year’s sales numbers we’ve come up with correction metrics (customer number growth, adjustments for inflation, etc…) that were used to clean the data from effects independent from pricing. 

Results were evaluated in 80 locations. Compared to base sales the gains are the following:

But what is the ROI?

Without any data analysis, the logical step would’ve been to apply seasonal pricing only in high-traffic areas, where the achievable margin increase’s base would’ve been 10-20 million in HUF. Carrying out data analysis however allowed us to find stores where we could increase margins by 54% more than in only high-traffic locations. This is the result of 15 days of work.

An interesting fact is that stores that are less visited by transit customers experienced losses during our test, which further proves the necessity of data analysis and data-driven decision making in business environments. 

In the following seasons the same methodology can be applied with only 1 day of work, or a software can be created and integrated into existing pricing systems.

Any questions?

No matter how successful data projects may be, our experience is that business stakeholders treat them with caution most of the time. This case study shall serve as an example that even with relatively little invested time, great results can be achieved.

Author: Márton Biró – Data Scientist

Reading time: 5 min

Easy question, tough answer

When it comes to leveraging a company’s data assets, the first use-case that pops into your head is probably not the optimization of opening hours. However, if we consider an enterprise which operates a number of offline shops, the potential of such a project becomes clear.

Setting the goals of the optimization is easy as pie: if a shop operates too long then the operating costs are higher than optimal. If it does too short, additional revenues may be realized. In the first case, the open hours should be decreased while in the latter, they should be increased.

We might as well think of two things that question an optimization project like this:

1. is it indeed possible to “play” with the shop hours?
2. isn’t the shop manager adaptive enough based on business operation?

The answer to the former concern is the usual “it depends”. During our project, we have dealt with shops that had both legal (some shops’  working hours were regulated by law) and business constraints (employment is only possible with conventional hours/week for employees) on working hours. However, scenarios like the following may be accounted for:

• 24/7 vs closing at night
• 12 vs 16 hours of operation
• opening 1 hour earlier/later
• closing 1 hour earlier/later

A savvy manager might realize what customer volume to expect in a given hour and he might take this into account when deciding on the operation hours. Nevertheless, can he accurately guess the (additional) expected revenue when operating longer? And does he make an optimal decision on the network level?  By using machine learning, we can definitely predict more accurately and with more objectivity. It is quite sure that, by using only human intuition, we will not be able to find an optimal solution on company level due these two factors:

• one does not know whether, by operating longer, one actually steals revenue from another shop of ours (cannibalization)
• one (e.g. shop manager) might have personal incentives not to operate less hours or close a shop entirely

In the following, we present you the method that can tackle this problem in a data-driven way.

Modelling customer decisions in a simulation framework

Simulations are extensively used when optimizing complex, real-world processes (e.g. traffic design). In this case, the outputs such as a financial metric (e.g. EBITDA) of working hour combinations may be simulated. To make this efficient, we should model the decisions of the customers, thus we may see the effects of changing the operation hours on micro-level. The two most important effects are:

1. how many additional customers do we attract by operating longer and how many of them are taken away from another shop of ours (cannibalization)
2. how (where) do customers substitute when we close a shop

The animation above demonstrates how customers make decisions when we close two of our shops close to each other. We can see, that by closing shop No. 2, we lose 5 customers (they choose the blue competitor) but we also save operational costs. Our simulation framework models such scenarios in more iterations and larger scale. It assigns an output (e.g. EBITDA) to all inputs (operation hour combination of our shops). Having that, we can easily conclude the optimal set of working hours for our company.

The strength of our solution lies in its ability to account for various factors that influence customer choice. In the animation, the customers chose solely based on distance: after closing their shop, they choose the one that still operates and is the closest (it may also be a competitor). Such additional factors are:

• loyalty: customers of different loyalty have different willingness to travel
• competition: the heterogeneity of the competition may be quantified. Substitution with competitors of price level significantly higher or lower is less likely
• travel time: it may be more important than distance alone

Summary

Most companies operating a big network of shops base their decision of operation hours on instincts and conventions. By using sophisticated machine learning solutions, we can support these decisions. Such a project can be really beneficial but it also entails challenges. The biggest challenges is, by no means, the collection and processing of accurate data. If we can tackle this, we are on the right path to create vale with data.

Author: Gerold Csendes – Data Scientist