Four years ago, French spirits major Pernod Ricard launched a data and AI programme to help it scale multiple strands of the company.
Today, the group uses several data and AI tools to support operations across 28 of its markets.
Go deeper with GlobalData
These include D-Star, which helps Pernod Ricard’s sales teams decide which brands to promote in specific regions and channels, and Matrix, which it uses in more than 20 markets to help employees make wiser choices around marketing spend.
Maestria is another tool the Jameson whiskey owner uses to monitor consumer drinking occasions and demand while Vista Rev-Up looks to optimise its pricing and promotional moves.
In September, the group also launched Horizons, a career learning and developing platform for its employees. The tool maps the skills of its approximately 20,000 staff worldwide with the aim of providing workers more opportunities, either through the form of new roles, mentoring or short-term assignments.
Just Drinks sat down with Pernod Ricard’s chief digital officer Pierre-Yves Calloc’h at the company’s Paris headquarters to discuss the opportunities and challenges of implementing the technologies across its markets, and why he sees AI not replacing large parts of its workforce, but making them more efficient.
How well do you really know your competitors?
Access the most comprehensive Company Profiles on the market, powered by GlobalData. Save hours of research. Gain competitive edge.
Thank you!
Your download email will arrive shortly
Not ready to buy yet? Download a free sample
We are confident about the unique quality of our Company Profiles. However, we want you to make the most beneficial decision for your business, so we offer a free sample that you can download by submitting the below form
By GlobalDataFiona Holland (FH): Is Pernod’s AI technology all built in-house?
Pierre-Yves Calloc’h (PYC): We actually have a mix. We have decided to internalise a lot of the competency because we think it’s really important… as we want to have transparency on how the recommendations are coming out of the tool…
Now we use also a third-party tool. For example, there’s no way a company like us can train a large language model like open AI, so we use some of the interfaces from open AI externally. It’s a case-by-case decision depending on the type of usage, technology, benefit, cost. It’s a buy versus build… type of discussion. But we want all the strategic elements to be controlled internally.
FH: How do you expect Pernod Ricard’s use of AI to develop in the next few years? Are there areas of the business you would expect the company to be more focused on with this technology?
PYC: We have different topics, so we’ve started with marketing and sales because that’s part of the core business. We still have some more things to be developed on that. In that area, supply chain is one of the [areas] we want to expand… Agriculture and viticulture is another area of expansion of the use of AI.
FH: Can you provide more colour on how it’s being used in wine production?
PYC: We started almost 20 years ago. I was in Australia at the time, and we started to predict the maturation of the grapes using a lot of historical data that we had. When you do that, the good thing is that you can plan the much better the harvest. If you plan better the harvest time, you can plan better the crushing and then the usage of all the facilities… At the time, there was not the computing power there is today.
There’s this type of usage, but also to promote the usage of regenerative agriculture techniques. We’ve been testing a number of practices and the combination of them, and we want to optimise the best combinations to save on water, carbon footprint, use of pesticides… There’s a number of topics where we want to continue to improve. And the good thing about AI [is] it can process a lot of data, find the best combinations. This is where we work in close collaboration with our experts in the different regions.
FH: Does that help the company better prepare itself for future obstacles like climate change?
PYC: We can test different scenarios… even during a season we can see if the temperature is one or two degrees higher. We can calculate in fact or predict what would be the impact based on previous years. Obviously, you need to have had some similar situations in the past, but that’s a good tool for simulations.
FH: In manufacturing, is automation something the company is using now or thinking about adopting long-term?
PYC: Yes, we’re looking at some automation. If you look, in fact, on the production lines, you have already some quite good automation… you have already machines which are [detecting whether] there might be an issue on the label [such as] where there might be some issues on the label being well stuck to the bottle. We have already implemented this for a few years. The great thing is that now we have more possibilities (because the technology is getting faster, better, cheaper), to be more precise on those parts.
FH: Does Pernod use a combination of different types of technology? Does it also use other tech like Internet of Things or blockchain?
PYC: One of the best combinations [we] have is in marketing. The traditional AI, or what’s called predictive AI, is useful for marketing mix modelling. For example, for a marketeer who Germany has €1m ($1.1m) to invest behind a brand to decide between TV, out-of-home, print, activation in store, etc., we have a tool that optimises that allocation. Then we can use GenAI to optimise the content itself in terms of speed, quality, pre-testing. This is less mature because ChatGPT is only two years old. We started a little more than four years ago on the predictive AI, so we are more mature on [that] than the generative but it’s also coming…
We are using blockchain on specific topics. I think the most impressive one is on The Whisky Exchange, the fine whisk(e)y and spirits selling platform where… if you are a collector of spirits and whisky in particular, you can buy specific limited editions, and you might not want to receive the bottle at home because of your insurance not covering such high-level products, or simply because you want to resell it in a simple way without having all the burdens of the complexity of shipping wines worldwide. [Through The Whisky Cabinet web3 platform] you can buy an NFT, so a certificate of ownership of the bottle, but the bottle stays within our custody until either you decide to burn the NFT and then you receive the bottle, or you can sell the NFT to somebody else. It’s a really simple transaction without having to carry the really nice bottle yourself.
FH: Do you see NFTs as a new form of premiumisation in spirits that’s developing?
PYC: It can be. For me, it’s a new form that we’ve seen in art. The craze of NFTs is low… but there’s still a flow of exchanges, new projects coming in. I think it’s the new form of ownership, which is digital ownership, the same way that I don’t carry anymore bank a physical bank card.
FH: Are there any markets where implementing these different technologies has been most successful? Are there others where it’s been a bit more challenging?
PYC: One of the big challenges of implementing AI is that it needs historical data when you do production optimisation. [In] the… data-rich markets which are more the Anglo-Saxon countries, so UK, US, Australia, New Zealand… you have availability of more granular data but, thanks to our local presence in the other markets, we’re also able to find ways to collect the data, even in emerging markets.
For example, we’ve developed an application that we give to our wholesalers in emerging markets and what we call fragmented markets, so that they can sell our products… the products of the competition… to the small ‘mom-and-pops’, the bars, etcetera. That way, we collect some really interesting transactional data that can help to make a better decision, get insights… so even in fragmented markets, we are able actually to find ways to [cull] data when we need to.
Usually, the data exists and sometimes you need to [have a platform] to store it somewhere, and sometimes we [team up] with our commercial partners to be able to have sales data or with the marketing agencies when we are doing activation… part of the creativity that we have when we’re implementing AI programmes is in finding ways to collect the data that you need for the different programmes to work.
FH: In which countries has it been more challenging to use the programmes?
PYC: I would say [regions] like Africa, South East Asia, Latin America, are a little more complex, because you have less [data] providers… which have already organised part of the ecosystem.
FH: Will Pernod Ricard invest in building up AI technology for those regions in the long term?
PYC: What we’re doing is we’re making the investment where it’s slightly easier. However, the good thing is that we’ve learnt it’s cheaper to actually replicate in the other markets. So, progressively, I would say most of the markets will get the benefits, but yes, it’s slower when the access to the data is more complex.
FH: Are you looking to implement the tech in any new regions?
PYC: We have a value approach to decide where we go and each time we are checking that the level of investment to implement a tool in the market is worth it. There [are some] fixed costs, so I would say our bigger markets are more favoured in that way than the smaller markets. However, we [try to see what [can work in] smaller markets and bigger markets, so that still we can get the insights and make the best decisions.
FH: Is AI changing the nature of any specific manufacturing roles at Pernod Ricard?
PYC: At this stage, [no]. We have a selection process to identify the best use-cases, we are looking at if we’re taking a risk or not. We are looking at what is the value, what is the cost of implementation. We have some cases which are planned in manufacturing, so one which has been already implemented in India is that in the area where [employees] enter the manufacturing facility, we check that they have the [appropriate] helmet, the gloves and secure shoes… and we have lowered the number of people which are not entering the proper security wearables.
FH: Is AI posing a risk to any roles at the company? Could it do so down the line?
PYC: That’s an interesting question. I get it quite a lot. My view is: do you remove people which are being more and more efficient? No, actually, you add to them… For example, our commercial people are getting more efficient because they [receive] the right recommendations. For me, we should add more because they have more value.
There’s probably the balance between that and some things which are not useful anymore but we have [few] things that can be fully replaced. For example… we don’t have customer service on the phone, we don’t have a huge plateau of 300 people doing something similar.
I haven’t identified yet use cases like this in Pernod Ricard where you have huge [staff numbers] doing quite mechanical work that could be enhanced or done much faster. At this stage, we haven’t identified it and we are more in the case of saying, ‘Okay, those people are being actually more efficient, we should have more of them’, which has been the discussion in some countries… [With] one of the use cases called D-Star, some of the CEOs have been saying ‘Okay, now my sales force is more efficient, so I should be putting more people in more regions because they can visit the right outlets and have a better impact on sales.’
FH: How is the company using AI to better understand consumer behaviour trends?
PYC: We have a programme which is called Maestria, which is segmenting the consumption in each country by what we call occasion, because the same person might want a really nice Lillet watching Emily in Paris on a Tuesday night and then the same person will go to disco and want Absolut Citron on the Friday or Saturday night. [These are] two different consumptions from the same person, so we are mapping this.
We are doing clusters; this is based on quite a lot of data. It’s starting with surveys that we complement with social listening. Social listening is all the public information which is on social media. People which have open profiles that are sharing with everybody, there’s possibilities to identify what people are talking about, whether they are positive, negative, etcetera. We are using this also to identify some trends.
Today, the good thing with GenAI is it’s easier to analyse a lot of text and get summaries, getting trends. We are enhancing the way we’re doing [that clustering] and the analysis of large amount of unstructured data, like text, images. We are getting to the next step thanks to the GenAI, which is now making it much easier.
FH: Is this happening across Pernod’s markets?
PYC: The good thing about the social listening part is that [it applies] in most markets. You have the specifics of the China ecosystem, the South Korea ecosystem etc., so we have slightly different sources of information. There are some using more Twitter than Facebook than Instagram, but we have found ways to be able to connect to the different environments.
FH: Is cybersecurity something you deal with in your role?
PYC: Cybersecurity [has been] a super important topic for every company. [It’s important] both to protect our own data, but also the data, for example, from consumers who have accepted to share their data with us. We have a full team in charge of the topic. Complexity is coming with the attacks being more and more clever, so it’s a race between the cat and mouse.
Our teams are getting better at cybersecurity, so that’s a constant topic that we are continuing to review. We invest[ed] quite a lot of money in the past, because today, all companies are super dependent on computers, on servers, on data, which is almost everywhere.
FH: Do you expect cybersecurity to become more challenging in the future? How is Pernod working to combat the problem?
PYC: One of the main risks in cybersecurity is me or my colleague getting an email which is personalised to me, thinking that it is coming from [CEO] Alexandre Ricard, clicking on it… there are more tools to generate personalised emails, being more precise on the topic, so I think that’s one of the risks which is increasing is [whether] every employee [is] able to identify when it’s not coming from the right person… The best [prevention] is upskilling, explaining to people what’s coming, being more careful, and it helps both in professional but also in personal life. We invest quite a lot in upskilling the different teams.
