CommerceAI

Ian Jindal (00:02.21)
Well, hello, welcome to the studio where today you're joining us for another episode of the Commerce AI podcast. Now, in this world of AI, you're going to be very pleased to know that we are real humans, honestly. And just to prove, I'm going to invite my wonderful guest, Emanuel Gounod, just to introduce herself and tell us a little bit about what she personally

and commerce IQ are doing in this fast changing world. So Emmanuel, thank you very much for joining me on the podcast.

Emmanuelle GOUNOT (00:40.075)
Well, hello, Ian, and thank you so much for having me today. I'm thrilled to be here and discuss some exciting topics. So my name is Emmanuel Gounod. I'm VP of Customer Success here at Commerce IQ. And I work with our enterprise customers to help them make the most out of our technology to transform the way that they run their digital commerce operations.

We work across e-commerce sales management, the digital shelf and retail media, layer that with some agentic workflows to really help customers operate in this new world of, I would say, the AI shelf at this point. And so really excited to see how we can discuss some of these things further today.

Ian Jindal (01:27.532)
Wonderful. Well, firstly, Commerce IQ is such a good company name anyway. But let's just take a few moments to dig into what you actually do, because you ticked off a number of bingo buzzwords there, know, from retail media to digital shelf. So put simply, if I'm a commercial retailer looking to improve my online trading capability,

Where does Commerce IQ fit into the proposition I offer to the customer?

Emmanuelle GOUNOT (02:03.755)
Absolutely. So Commerce IQ will partner directly with brands to help them

optimize their business online, what we can do is that we have a series of solution that all come together into a unified platform. And I would say that's really where Commerce IQ distinguishes itself from point solutions, but really provides this operating system for digital commerce. The first one is going to be around e-commerce sales management, whereby we have a number of automations to help brands preserve the integrity of their product detail pages,

a revenue leakage and optimize.

profit recovery, as well as a reporting console that will provide them with a deep diagnostic of how they're currently performing from an e-commerce perspective. The second piece is around the digital shelf, so really helping them to manage their car's performance across retailers. And the last piece will be retail media, whereby we really distinguish ourselves by not only looking

at the attribution of sales, but ensuring that we're driving incremental sales for our customers.

Ian Jindal (03:19.074)
Fantastic work. You've talked about a number of areas there that are all at the forefront of where AI is helping develop the tools very quickly, but also it's where the customer is bringing her own tools and trying to sort of kick the doors down and take control herself. you know, especially in the digital shelf, for example, you know, we have now consumers coming.

directly to enhanced DPP, product pages, extra data, cutting out the search engine, ignoring our funnels. So you really are in a pretty dynamic, challenging area where everything seems to be up for grabs.

Emmanuelle GOUNOT (04:08.627)
Absolutely. And what we see is that everything is accelerating, right? So we always know that we've been moving at algorithmic speed because all the different factors are moving quickly, whether you're looking at search with SEO, whether you're looking at availability.

lost buy boxes, there are a number of variables that brands are constantly optimizing for. And so for instance, if you're looking at a brand, what we will always say a brand that would have, let's say 400 products, 30 variables that are changing, and we'll say only three times a day, it's probably a lot more than that. Already,

you're looking at about 25 decisions per minute. And that's something that's very difficult for brands to keep up with. What we see now is that things are only moving faster.

beyond the algorithmic world of retail, what we see is that with AI, there are also a number of dynamics that are changing those cars metrics. So we come, I would say the industry still calls it the digital shelf, really looking at optimizing the cars, content, availability, pricing and promotion of your assortment, as well as rating and reviews and search. But we're also in that across all your PEPs.

across retailers. But we're moving to something that's much more of an AI shelf. And there two components of that. The first one is really going to be around discovery of products and the customer journey. So what we see is that with agents on retailer platforms such as Rufus for Amazon, Sparky on Walmart, and many others that are now being developed, the customer journey is changing rather than, you know,

Emmanuelle GOUNOT (05:59.276)
searching, clicking, purchasing. Customers are now asking.

getting an answer from one of these AI assistants and then purchasing. And so it's a very different journey that needs to be accounted for. Likewise, on the retailer side, from a merchandising standpoint, retailers are now allowing a number of agent to agent flows that really change the way of doing business. And at that point, it really becomes impossible to do it all manually. And so what was already a compelling

case is now becoming an absolute necessity to move beyond the speed of these algorithms. The only thing that can help you to move at that speed is going to be some agentic workflows because otherwise we're stuck in, let's say, weekly business reviews that are no longer relevant by the time you review the data.

Ian Jindal (06:59.436)
Okay, so a couple of things in there. So it all sounds lovely. The slight problem is that we've heard it before. So if you go back 10 years, maybe 2012, 2014, you the growth of the algorithmic age where everybody became a Bayesian expert, a predictive algorithm, we had the machine learning, but the instantiation of that was

Here are some algorithms to run, review the output, make some choices or let it run automatically. So we've been sold that benefit and we can debate whether we had all of the value or not. But track forward to 2024 when people started talking more about AI, it was exactly the same sales pitch that the

Retails and brands are hearing here's some new magic. It was called an algorithm now It's called AI and you'll do all the same stuff faster What has actually changed? So if I'm a CMO thinking hang on Here's another proposition Asking me for budget when I haven't had the ROI from the last one and it sounds the same Help us understand what the real change is over and above

just version 3.0 of the algorithms.

Emmanuelle GOUNOT (08:31.115)
That's a great question because to your point, many of us feel that we've heard it before, right, with all those different automations. But there is, what's fundamentally different now is that we're no longer looking at something that's purely rule-based, but really looking to look at goals and layering some of these agents with what a human would do.

and ensure that these agents are calibrated and trained with context. So one of the things we were doing at Commerce IQ, and that's really key in terms of moving from, I would say, these rule-based to these goals, goal accomplishment instead, is ensuring that we're looking at, so what is the macro context? What is the industry context?

What is the retailer context, the brand context, and then what do I need to do about it? Different brands are going to have different tools, different constraints, and they will do things differently. So this is actually not something that's completely off the shelf and in that sense, a software that can be implemented once and for all to get the job done. In that sense, it's not about filing a ticket.

Right? This is not about saying Amazon, this is broken. Here's a ticket. This needs to be solved and automating that process. And that's still obviously something that needs to be done. And that's still part of our solution. However, what we're now helping brands to do and what brands need is really about understanding what's happening. All right. So my sales are down. In what category? What brand? What's Q? What's driving it?

Is it going to be availability? Is it going to be my traffic? Is it going to be from a deal perspective? Do I have my deal badge? What are all these different alerts? And really go through a checklist analytically the same way that a human would look at it.

Emmanuelle GOUNOT (10:44.203)
And that's something at some point as we continue to train and iterate and provide feedback on these decisions, the agent is able to come forward with a set of recommendations and sometimes might even be able to act on them independently to drive tangible business outcomes.

Ian Jindal (11:02.392)
Hmm.

Emmanuelle GOUNOT (11:03.485)
and what's going to be good for one brand may be very different for another one. For instance, if I know that I have logistics issues, supply chain challenges on one given ASIN, I may not want to necessarily push that one from a media perspective beyond a certain point. And so when we tie all of this together, same thing when we look at retail media.

We want to make sure that we're not just optimizing for ROAs, right? Which is really looking at how do I get a good returns and sales attributed to a particular ad, but much more how am I going to drive incremental sales from my media spend? We do not want, for instance, to be spending on products that would already be converting.

Right? you have a... Yes, go ahead.

Ian Jindal (12:01.772)
So I'm agreeing, everything you're saying sounds amazing. And I've got two thoughts then from the perspective of a trader. So firstly, it seems you mentioned the weekly trading meeting. So typically you'd have a hundred pages of Excel to read and think and pick out data. And then you'd have your own excuses and triumphs about last week's trading.

And then you come with a to-do list saying, pull back on these ads, we're out of stock there. So there'd be a lot of human activity that would be a combination of data, properly reported, your own knowledge, experience, intuition, but setting instructions for everybody else. So you'd have six hours of meetings, four hours of sending emails and delegating tasks. You'd go home exhausted and honest day's labor.

at a keyboard at the forefront of retail. It sounds from what you're saying as if this is now happening on a minute by minute basis, seven days a week. And so that's quite a major change for the activities of the retailers. It changes how people trade.

Emmanuelle GOUNOT (13:23.851)
Absolutely. And that's a great point. We know that e-commerce never sleeps. The customers are always shopping, retailers, websites. And to some extent, the agent will not take a break while you're watching during the football game. And so this is actually giving teams a lever.

to really continue to optimize their business 24-7 without having that operational constraint, which is capacity and hours. And so we are moving from, I would say from a unit economic perspective, usually working with agencies and with internal teams, we have a constraint.

on that operational bandwidth. And that's no longer the case with these agentic workflows. And so that's really what's changing with agentic commerce today.

Yes. And to your point, also having not only the ability to do, but so much faster, but the ability to prioritize. Of course, teams would spend countless hours reviewing all of the data, conducting the analysis. But imagine being able to walk into a meeting where you already know the three things that you have to work on that you need to fix and having that be automatically delegated to the right team and completed in some cases. And that changes the way teams work.

Ian Jindal (14:55.884)
No, it does, but it also changes the borders between teams because when we talk about this unified operating system, if you like, for retail, you have your media buys and the effectiveness there. You've got your digital shelf, as you call it, which for me is search and dicing, merchandising, promotion, but then also the supply chain aspect, which

You know, retail is all about availability. It's making the right product available to the customer for the right price at the right service level. So normally if I'm talking to retail teams, I'll say you need to bring all these three together. I'll sometimes get that that look where they go, bless him. Of course, that's theoretically a good idea. But do know how difficult it is to get, you know, real time?

supply chain or real-time retail media that doesn't go through the agency. So it's as if you've picked all of the really big hard systems, complicated micro amounts of data, real-time changes, and then said, let's all put them together live. So it sounds amazing, but how does this happen in real life when you work with retailers?

Do they have this data ready? Is it just an easy case of plugging in three cables and it works? With a challenge.

Emmanuelle GOUNOT (16:30.859)
So would say on the retailer side, different retailers work differently, right? So there are some retailers where with API integrations, you can have all the different data. It might come from different places, but you can operate on top of that data and get it directly. For some other retailers, at least on the shelf piece, this might be more in terms of what we would be able to crawl and

layer with some other pieces of data. Of course, for retail media, we would operate within the retail media network that they've built and be able to work through that directly. So it really depends on the retailer.

I think that what's very powerful to your point is being able to bring these different dimensions of a business in one place for a brand to manage. Often, and we've seen these silos, right, where let's say supply chain is going to manage one thing. You have your e-com manager who's working directly as well, let's say on pricing.

with let's say one of the retailers and then you have your media team that's doing something else but they may not be talking to each other or not talking to each other frequently enough and the ability to suddenly be able to bring all of this data in one place and ensure that your decision, the different

workflows or aware of each other changes everything. And I think the area where we see that the most, quite honestly, I mean, and they're all interconnected, but the most immediate application is going to be media. You do not want to be spending on an item that's out of stock or at risk of being out of stock or an item where your share of voice organically is already high, an item where your content isn't good, frankly, as well.

Ian Jindal (18:20.205)
Hmm.

Emmanuelle GOUNOT (18:34.189)
so being able to bring all of that together and then measure that lift in terms of sales is going to be what's really making a difference.

Ian Jindal (18:41.848)
Yeah.

Okay, now let's look at the people then, because everything you said, I'm loving it, but I'm also thinking that as more and more gets done by the agents, and you know, they're to be able to cover, you know, as a merchandiser or as a category manager, I might be managing my top 100 skews with personal love and attention, knowing it in depth, etc, etc.

Whereas now an agentic helper could be managing my top 1000, 2000, whatever the number is, big numbers. So the question for me is how do you then develop your own skills so that you can add something else? So on the basis that, let's say we have a chat in three years time, Emmanuel, and everyone's adopted.

the AI, the AI is getting better and better and better, it's sort of obvious we'll end up with a high average sludge where we are, you know, racing like gerbils in a cage just to push our product towards the Amazons and other storefronts of the world, but it's all average. I'm looking, I'm sitting inside a retailer thinking,

Today, this is going to help me. Tomorrow, I need to master it. But next week, I need to be ahead and leading it. How are you seeing people making that transition from AI taking some things away to them getting ahead of it to use it better?

Emmanuelle GOUNOT (20:31.933)
It is tricky right from an organizational standpoint in terms of how some of these transitions are playing out. And we see that different CPGs, different consumer brands are in a different place. There is certainly a lot of appetite right now in experimenting with all things AI in lot of different fields. We see it on the...

I would say the actionable insights from shelf, right? We see it on the content piece as well. We see it on retail media. But at the end of the day, I think what the industry is still craving for and what we're moving towards now is the idea of orchestrating all these different workflows together so that people are not...

replicating silos in a different way. I think there are lots of solutions out there that are building fantastic agents and workflows, and that's great, and that's saving everyone some time. But at the end of the day, no matter what these agentic workflows are, we need to find a way to bring them all together and coordinate them, orchestrate them, and be that brain.

Ian Jindal (21:24.11)
Hmm.

Emmanuelle GOUNOT (21:45.101)
that will ensure that left hand is talking to the right hand. So I think that's where the industry is slowly moving towards to. And that's what will make a difference. To talk a little bit more about the appetite and how consumer brands are making that change. I think that today we're still in a phase

Ian Jindal (21:49.634)
Yes.

Emmanuelle GOUNOT (22:13.484)
at large of maybe experimentation. Let's try it. How do we, do we trust the agent?

Ian Jindal (22:18.243)
Yeah.

Emmanuelle GOUNOT (22:21.42)
Some, you know, I think a lot of organizations are still there. What can I delegate? What shouldn't I delegate? And so building those governance rules in terms of when needs to happen. Then how do I show, if I want to delegate, what is going to give me trust that this is really truly going to be a junior analyst, a member of my team that I can train.

and who can give me leverage over time. So there's a lot of onboarding and context that needs to be provided. And I think that once again, this is not a software solution anymore. Ways of working for us, we see that the way we work with our customers is very different today. We...

Ian Jindal (22:46.638)
Hmm.

Emmanuelle GOUNOT (23:04.372)
we started to lean a lot more into a forward deployed engineer type of engagement, where we will sit with customers, understand their workflows and acknowledging that they're all different, right? And their own data stack and what do we need to connect to? How do we train the agent better? How do we deliver those better results? And then we come to that layer of orchestration. And then...

What we see is that these teams, at first, they absolutely want a human in the loop. Most of them will. This is something that's quite important at this stage. They want to be able to audit, to review results. Over time, as they see, and we saw it, one of the first agents that we built around content last year. At first, we had recommendations that

Maybe we could, the customer could approve 30 % of the time the first week when we launched it. And within 45 days, we were 95 % approval. At that point, the team manager is saying, okay, all the long tail, just approve, let's approve that in bulk. We know that this is going to be good 95 % of the time and getting better every day. Let's do that. And that allows team to suddenly do a lot more.

Ian Jindal (24:08.526)
Hmm.

Emmanuelle GOUNOT (24:31.976)
and so you were talking about, know, the top hundred skews. get a lot of, a lot of love. we can do the entire catalog. We can do all retailers. the reality is that, you know, digital commerce is just getting more and more complex. There are more and more retailers that are part of the mix. and at the same time, teams are not getting any bigger. Teams are getting leaner. And so this is what's important. I think at first.

Ian Jindal (24:55.895)
Yes.

Emmanuelle GOUNOT (24:58.814)
even before saying work is changing to be able to get all that done.

Ian Jindal (25:03.47)
So it's interesting listening to you, it's painting a sort of Renaissance picture of skilled people who can use the tools but also bring that synthetic commercial view to the business. So I'm going to put you slightly on the spot now, Emmanuel, and just talk a little bit about your own history.

And everyone who's listening is going to be clicking on your LinkedIn profile. And they're going to be saying the same thing, which is like, oh my goodness, how can she have done so much and in so many different areas? I'm just giving you a quick thumbnail and then see what we can derive from that, from your own AI journey. So you worked in B2B, is industrial B2B, then

BCG as a consultant, then in France as a working forum, as a homeware and living company. Then of course, Amazon, because everyone has to do Amazon. From Amazon, you then went to Alibaba, so yet more software enhanced retailing. Then you left the retail and just took the software to Uber.

and then into software now at Comms IQ. So you've ticked off. It's a bit like you've been on a sort of a learning journey to cover off all these different sectors. So the two questions that come from that are as you're now looking to help retailers make more of AI, what are the lessons that you draw? So when you look back and you go, that's a really good thing I learned when I was at

BCG or B2B. And then how are you maintaining your own disposition to AI? So as well as selling and helping people use it, how are you keeping yourself sharp and changing the way you work on that? So very long question. Put simply, we're just saying, what are the things you've brought forward from your experience and how are you taking your own learning forward?

Emmanuelle GOUNOT (27:26.636)
Absolutely. So would say, first of all, my conviction is that there is so much data out there. And so I was always looking for ways to get to insights faster. And of course, being a consultant at first of an hour, a while back, there was a lot of analysis and a of crunching data. Same thing at Amazon with a lot of tools to be able to do that. But then...

Ian Jindal (27:48.556)
Yes, yes, yes.

Emmanuelle GOUNOT (27:52.353)
When I joined Alibaba, for instance, I actually joined Lazada, which was later acquired by Alibaba. And that was a big startup. And there, there was a big effort to rebuild our tech stack as well and get to the same level of tooling and automate. And it was fascinating to see how that was being done. But I think what I've seen throughout, and then same thing going to Uber and then doing managed services with

with Intrepid that was later acquired by Flywheel, it was always like the, were constrained by human limitation and our capacity, right? How much we could do to react in, yeah, exactly, react in a world that's complex changing. And the only thing that can enable us to go faster is to move to that layer of automation and moving from just rules to goals.

That's the thing that brought me to Commerce IQ, to be honest, is the level of automation and having that, yeah, not only that vision, but that those automations and those AI workflows that could bring us to the next level, because I do see the need for it having been in some other environments.

Commerce IQ, I would say, in terms of how we work internally and with our brand customers is doing a lot of work as well to ensure that internally we are really doing what we're preaching. I would say we're.

or CEO on that and our entire leadership team from that perspective are certainly pushing AI education quite a bit. So we've all gone through a number of AI courses. We all have a number of tools that are working.

Emmanuelle GOUNOT (29:43.603)
working with, so whether that's Claude, for instance, that the entire company is using in addition to some of the other LLMs, encouraged to build workflows and share those on a regular basis to ensure that everything is constantly accessible and can be simplified. The idea is that if we have to do anything that's labor intensive more than once,

Ian Jindal (30:01.198)
Hmm.

Emmanuelle GOUNOT (30:10.71)
What's the process to make that tedious work go away and ensure that we can get to those insights and actions faster?

Ian Jindal (30:17.218)
That's a good point. mean, one thing I generally don't introduce into conversation is that I trained as a chartered accountant in the last century. And, you know, we we would benefit from inheriting spreadsheets and analysis structures that previous people had built up. That's a cumulative

knowledge and insight and experience. But we also had to do some things that were just painful. I remember spending five working days in the basement of a large international bank, adding up these computer printouts, the final column, because a part of the time didn't quite know whether he could trust the computer because it could just be printouts. So he made me add them up. I will never get those five days of my life back. But I got very quick.

on a large calculator. you know, that wasn't good work, but it did make me think about assurance. How do I know that the job is well done? How do I know this is a good question that's worth asking again? Or when everything seems fine, how do know it is fine rather than just the gloss of being fine? So I think I'm fishing for a way that, whenever I work with

a retailer that knows their product, their customer and their processes, the sheer depth of experience they have from fixing broken things on a Saturday night or from having the pain of how to do it themselves. How do we make sure that when an intelligent new employee starts with us, who has the brains but maybe lacks the domain expertise or the muscle memory,

or the critical ability to see the questions as well as the answers. How are we going to train people in our organisations to see AI as a help rather than as the only option?

Emmanuelle GOUNOT (32:27.828)
And that's a good question and one that everyone is asking in general, right? What sort of role, what value do we add and how do we train people? You talked a lot about the fact that saving time, that's giving us data. think it's also AI.

will pick up on signals that a new person may not always pick up on or may not have time or visibility into. So that's something I see a lot in my line of work. From an engineering perspective, there is a lot in terms of detecting anomalies and suggesting things. And so maybe guiding the work of those employees, but that's not a substitute for the human expertise that we bring.

Ian Jindal (32:51.224)
Yeah.

Emmanuelle GOUNOT (33:11.02)
and the training that needs to be done. I think that if anything, the bar is now a lot higher, right? For all of us in general is that we need to continue to ask the right question. Everyone can have access to the data. Everyone can have access to the insights and to, I would say, quick ways of getting things done. The question is, what would you do with it and how do you go further?

in asking the right questions, or picking up on signals and anomalies, and then feed that back almost into the agent. to some extent, I think that we need to work in tandem and change our relationship with agents in AI much more as like, do this for me, is almost like, do this with me. And you're my teammate. You're my analyst in my pocket. You're my helper.

And I'm going to continue to give you feedback, and we're going to have this discussion to take it further. And with all of us having this, think teams' ways of working are changing. What we see across the boards is that teams are getting a lot flatter, right? We no longer have, in many places, these organizations with huge spans of controls and layers of management. Everyone is a lot more closer to the work.

Ian Jindal (34:31.278)
Yeah.

Emmanuelle GOUNOT (34:31.417)
and that's changing quite a bit.

Ian Jindal (34:33.838)
Yeah, I think that's absolutely spot on and a very good place to pause our conversation in this world of chatting with not telling. So I think that's very important. But just as we as we leave, let's imagine we're chatting to a colleague, a client, a friend, and they're saying, look, you know, I'm using AI more and more and more.

but I want to develop my own skills or abilities, where would you suggest that our listener dedicates the next hour of personal development and learning time? Is there a topic, an area, an approach that you think is going to fit us well for the next phase of working with these agentic tools?

Emmanuelle GOUNOT (35:29.964)
So everyone's going to have their own problem that they're going to try to solve, But I think so obviously, I think everyone today is doing the basics. I would encourage people to, I think that's not yet the case, to look at building their own agents and going beyond the summary, rewriting, seeking information, and trying to build a more complex workflow.

Ian Jindal (35:33.314)
Diplomatic.

Emmanuelle GOUNOT (35:54.7)
I think that Claude, to some extent, has really removed a lot of barriers to do so. So I would say just pick one problem you're trying to solve. And that might require several steps. And there are lots of step-by-step trainings to do that. And literally you'd be surprised by what you can build within an hour or two. that becomes, once you get started, I think you get hooked.

Ian Jindal (36:17.826)
Wonderful.

Ian Jindal (36:23.256)
Good, well, let's hope so. So, Emmanuel, thank you so much for sharing your thoughts with us today. It's been wonderful to chat. And dear listener, if you're looking at your paper calendar on the wall and thinking, goodness me, is that the date? Well, may I say that if it isn't yet June the 3rd, 2026, then you are able to come and meet us in real life and join our think tank at Commerce AI.

We'll be in London spending the whole day chatting off the record, looking at how we can respond commercially to the opportunities and challenges of AI in our business. If however, the third is now in the rear view mirror, fear not. A, you've enjoyed this podcast, I hope. And B, we'll have on the website at commerceai.transistor.fm

there will be links to the event write-ups and further information so you haven't lost out at all. So on that positive note all that remains is for me Ian Jindal to say thank you very much for joining us and once again to Emmanuel for being such a wonderful guest. Thank you Emmanuel.

Emmanuelle GOUNOT (37:38.86)
Thank you again.