Guest blog by Professor Youssef Boulaksil (United Arab Emirates University)

Many large Fast Moving Consumer Goods (FMCG) manufacturers and distributors, such as Unilever, P&G, Nestlé, and Danone, struggle with how to efficiently supply the thousands of nanostores that one typically finds in large cities in developing countries. Although each nanostore generates a negligible amount of sales, the cumulative sales of all nanostores in these markets is usually much more than the sales generated from modern channels such as hypermarkets. Hence, the nanostores form a serious sales channel that cannot be ignored. Distributing goods to them is a complex task due to several reasons. First, these stores do not keep track of inventory levels or historical sales, they do not actively order from their suppliers, they operate under very tight cash constraints, and there is a high turnover in stores with frequent closures.

Therefore, the distribution strategy that many FMCG distributors apply is the van-sales strategy. Under this strategy, a van or a small truck is filled with the products and visits the nanostores one after the other. When visiting the nanostore, the truck driver inspects the inventory of the item and tries to collect an order. If that happens, the delivery and payment occur on the spot. Under this strategy, the risk of not getting paid by the nanostore is eliminated, but it has some serious drawbacks. First, the truck driver is not a sales person by nature, and therefore, the generated sales may not be satisfying. Second, the truck driver may lose lots of time to find a parking bay, especially since the majority of the nanostores are located in densely populated neighborhoods with narrow streets.

An alternative distribution strategy is the pre-sales strategy. Under this strategy, the order collection and order fulfillment processes are decoupled. A pre-sales agent collects the orders from the nanostores by visiting them on a motorcycle, and the delivery and payment occur at a later moment. This strategy is obviously more expensive; as skilled pre-sales people need to be hired and some additional investments are required to implement this strategy. Distribution can be more efficient, since typically an additional helper serves on the truck, and only stores that actually have ordered products are visited.

In collaboration with Valencia, a successful Moroccan fruit juice manufacturer and distributor, we conducted a study in Casablanca, where the van-sales strategy was applied in one district and the pre-sales strategy in a similar district. Detailed data related to the distribution process (such as the number of stores visited, the traveling time, the collected orders, time spent in the store, etc.) were collected at a daily level during 3 months that allowed us to compare the performance of the two strategies. The result was that under the pre-sales strategy, the sales volumes were much higher, while the distribution cost was about 25% lower.

This result inspired us to develop a mathematical model that would provide insights about which distribution strategy to choose in general. Using the model, we show that the distance between the nanostores is the main parameter that determines which distribution strategy is optimal, given the difference in sales between the channels. We derive an expression for the critical distance between the nanostores, which is mainly dependent on the cost structure, hit rates, and the level of congestion in the city. If the actual average distance between nanostores is below the critical distance, then the pre-sales strategy is optimal. Otherwise, the van-sales strategy becomes optimal. This means that in densely populated neighborhoods where nanostores are close to each other, the pre-sales is the best strategy to apply. At country-sides or suburban areas where the distance between nanostore is larger, the van-sales will prevail.

The detailed model are available in a publication in Interfaces. For more information, please contact Professor Boulaksil.

A recent study, based on the data from the Swedish alcoholic beverages industry, shows that even in a very developed market like Sweden, gaining access to smaller retail formats increases sales at a much higher rate than when selling in large stores. The reasoning is that although small retail formats may provide only a small pie due to limited store size and smaller assortments, manufacturers can reach a larger market share exactly due to these very same reasons. Linking this to our research in the megacities of developing markets, this relationship obviously holds for nanostores, the small mom-and-pop neighborhood stores that flock the megacities of this world. Nanostores, which carry only a few brands of a product category due to limited cash and shelf space, allow a winner-takes-almost-all strategy.

However, serving the nanostore channel is costly. With tens of thousands of stores to be reached in a single megacity, the mere distribution costs can be very high. While many incumbent players serve stores directly, multinational manufacturers entering in these markets often adopt a strategy where wholesalers are used. The latter is driven by the lower operational cost and the lower headcount to such manufacturers when using this channel. Obviously, this goes at the expense of a lower market growth rate since wholesalers have less of an incentive to develop demand for a specific product; they are driven by their entire product portfolio. For example, as we have learnt from the data of a CPG company delivering in the city of Bogota, the wholesale channel is 9% more profitable per unit sold than the direct channel, while the direct channel grows the market 11% faster than the wholesale channel.

In chapter 4 of the PhD dissertation (now available online) of my student Jiwen Ge, we study this trade-off. Under which conditions should manufacturers prefer one channel over the other?

There are two simple metrics that drive this choice. The first metric is the gross margin (that we denote below by the symbol M), essentially measuring the additional profit margin contribution of selling one additional unit, taking into account the price (typically higher in a nanostore sale than when selling to a wholesaler) and the distribution costs per unit (typically higher in the direct channel). The initial thought would be to pick the channel where the gross margin is highest. For new entrants into the market, where the cost of setting up direct distribution to tens or hundreds of thousands of nanostores is high, this usually would imply selecting the wholesale channel.

However, this does not take into account the effects of better replenishment and faster sales growth in the direct channel. Hence, the gross margin effect needs to be adjusted for the anticipated sales growth differences. For that, we introduce the second metric of growth adjusted profitability (that we denote by the symbol P), which is the gross margin M divided by the anticipated sales growth rate.

Comparing the two channels, then three potential strategies emerge: going direct, using wholesale, or first going direct and eventually switching to wholesale. In the latter strategy, first the direct channel is used to introduce the product and to grow the market, and subsequently the wholesale channel is used to take advantage of its lower cost once the product has been established. The combination of the gross margin and growth adjusted profitability metrics defines whether one strategy is better than the other.

Using the above-mentioned metrics, the tables below show what to do for different lengths of decision making horizons. Note that M(D) denotes the gross margin of the direct channel, M(W) the gross margin of the wholesale channel, P(D) the growth-adjusted profitability of the direct channel and P(W) the growth adjusted profitability of the wholesale channel.

An interesting trait that impacts the decision significantly is the decision making horizon. In some companies, a return on investment (reflected for instance in the Net Present Value of a project) needs to be made in a short time. For instance, a company requires returns to be positive over a one of two-year decision making horizon. These companies will find it hard to start with a direct channel strategy, since the benefit of higher sales may not be realized within the decision making horizon, and the initial loss may be too large. In many case, this implies that these companies will not win in the market through using the direct channel. Entering into the nanostore channel directly requires longer breath, and a decision making horizon is typically a bit longer, which allows manufacturers to enter the market using the direct channel and then switch to the wholesale channel to take advantage of its cost efficiency. If the horizon is long enough, and the gross margin high enough, a direct strategy should be used without ever switching to wholesale.

Of course, our findings are based on stylized models, and details may differ in different markets and for specific ratios, but based on these insights, it will be possible for any neighborhood of any city to define the ratios mentioned above, and to make the trade-off whether to go direct or make use of the wholesale channel.

This blogpost has been published earlier on LinkedIn

On September 12, my student Jiwen Ge will defend his PhD Thesis “Traditional retail distribution in megacities”. As far as I know, this is the first set of analytical models that helps us better understand how manufacturers should operate in the megacities of developing markets. In these megacities, a significant portion of retail consists of nanostores, small mom-or-pop-operated retail stores that mainly sell consumer packaged goods. In other work, due to be completed soon, we are showing that this channel will prevail in these megacity environments for the foreseeable future.

Although the thesis is a PhD thesis, with hence as its main objective to develop theory, I do recommend reading the thesis to practiontioners, to develop a better understanding of this important channel. The introductory and concluding chapters have been written such that they accessible to a large audience and the thesis is open-access, so anyone can access the pdf version online from a few days after the defense will have taken place. If you are interested in receiving a (free) hardcopy, feel free to contact Jiwen by sending him an email, since the printing company mistakenly printed a few hundred copies too many ;-).

 In the thesis, using analytical mathematical modeling, four topics are addressed:

1.    How much sales effort should a supplying manufacturer execute, to optimize the tradeoff between the costs of extra effort (more pre-sales agents, less efficient routes) with the additional sales that this might bring?

2.    How should a supplying manufacturer secure shelf space and cash, given that competing companies are trying to get access to the very same scarce resources at the store?

3.    Should a manufacturer deliver nanostores directly, serving each store individually, or indirectly, making use of distributors or wholesalers?

4.    In the new On-demand retail services that are developing in countries like Peru, China, and Indonesia, what is the optimal network size of nanostores to contract to have an efficient network with fast response to consumers?

In an upcoming series of blogposts, Jiwen en I will provide you with the key insights from each of the Chapters. The blog posts should become available one-by-one over the next few weeks. Just follow me on LinkedIn to get alerts.

Interestingly, in all cases that we study and try to understand, we closely link commercial decisions and logistics decisions. Serving a nanostore network successfully and growing revenue and profit in this large channel requires this. While in developed markets the logistics and commercial (marketing, sales) functions have become siloed over time, and maybe for a good reason, our work suggests that if you operate in traditional retail markets in developing countries this is definitely not a good strategy. The functions need to be closely aligned, and key decisions need to be taken jointly.

On a side note to the above but definitely as relevant to logistics or commercial executives in such environments, our book Reaching 50 Milllion Nanostores – Retail distribution in emerging megacities is also ready to go into print and should appear some time in October 2017. This book, edited by Edgar Blanco, Christopher Mejia, and myself, brings you the relevant concepts to be successful in serving the nanostore market, and provides a series of case studies of leading companies in these challenging environments.

Contract manufacturing is now common in many industries. While this has been common in the electronics industry for a few decades now, we also see many instances of contract manufacturing in other industries, such as pharmaceutical and chemical. In order to pool demand, contract manufacturers try to serve multiple clients from the same factory. In that way, they can try to limit fluctuations in demand for the scarce capacity in the plant.

For the OEM clients of a contract manufacturer this implies that they need to provide advance demand information to the contract manufacturers. This information can be in the form of a formal order long in advance, in the form of a formal reservation, or in the form of a forecast. In our experience, often the exact character of this information may be an ambiguous combination of any of the above. In a recent article that I worked on with my former student (now professor) Youssef Boulaksil and my TU Eindhoven colleague Tarkan Tan, we try to understand how a manufacturer should act if it has contracted (part of) its manufacturing with a contract partner.

Despite the model being stylized, the results are insightful for (OEM) manufacturers and contract manufacturers alike:

In most cases, ordering each period a fixed order quantity with the contract manufacturer is best, and random demand fluctuations should be dealt with by safety stock at the OEM. This is driven by the fact that it is difficult for the contract manufacturer to respond at very short notice, since it has multiple clients to serve and typically operates at high utilization.
In case the dynamics in demand can be forecasted well, so are not subject to much uncertainty, it pays off to conduct the coordination with the contract manufacturer in more detail: make more specific reservations of the contract manufacturer’s capacity to enable the contract manufacturer to prepare. Relating this to (1): this only makes sense if there is little uncertainty in the demand and the demand pattern can be forecasted well.
The contract manufacturer needs to formalize the ordering process with its client by clearly separating between order reservations and actual orders, where a reservation is just focused on the total quantity (capacity) needed. The manufacturer does not instate a small change or cancellation fee in case the reservation is changed or cancelled, since otherwise their customers will not plan their safety stocks properly.
The two parties should agree and built a mechanism in the contract to ensure that reservations are not inflated by the manufacturer who clearly has an incentive to do so to secure future production capacity. This can be achieved by charging the manufacturer for unutilized capacity reservations, even if marginally.
Obviously, the latter is also dependent upon the “power balance” in the negotiations between the two parties, but we actually show that even for the customer it may be better, since in that case a proper tradeoff is being made between the small supplier charge and an appropriate safety stock level.

How are you managing this type of a relationship, and how is the order flow organized? Please leave your comments on the LinkedIn blog version of this post.

If you would like more information, or investigate how this stylized model can be engineered into a decision support tool for your company, feel free to contact us. We might be able to link you to an interested Master student.

Demurrage and detention costs are charged by ocean carriers to ensure that consignees send back their containers as soon as possible. However, sending back an emptied container at an import destination precludes the re-use of this container in the immediate vicinity to fill it with export products. Typical demurrage and detention (D&D) rates give a few days of free D&D; after the free period the consignee needs to pay a fixed rate per day which may increase over time.

For a while now, I have wondered why this strange mechanism is in place. The most likely answer is that this has probably grown “historically”, as it does not make sense from any perspective: ocean carriers encounter extra costs by sending empty container for loading to export destinations, and consignees encounter paying additional fees to the carrier, even if they are trying to optimize hinterland operations. My former PhD student Stefano Fazi, now at the University of Groningen, has shown in his PhD thesis that the current D&D fee structures in the port of Rotterdam lead to higher cost for the supply chain as a whole, and moreover lead to favoring road transportation over barge, thus leading to higher emissions and congestion.

Recently, Benjamin Legros, Yann Bouchery, and I completed a very interesting theoretical study, in which we show – using a fairly simple mathematical line of reasoning – that all of this D&D does not make much sense: the supply chain cost go up and unnecessary container movements increase. It would make sense to keep a small number of empty containers in each hinterland area, void of any D&D payment structure. In our paper (warning: the paper is mathematical!) we show that with a simple threshold policy (essentially agreeing how long to keep a container in an hinterland area – or alternatively how many containers to keep in the hinterland – , depending on the total flow of containers to and from that area) we can reduce costs by 20-80 (!)% compared to a common policy where a container is returned immediately upon unloading – see the figure above taken from our paper. Interestingly, I recently heard that a number of deepsea terminals now have contracts with carriers where free demurrage is no longer expressed in the number of days, but in the total number of containers kept. This idea is very similar to ours, but as far as a I know this has yet to be deployed in the hinterland.

Of course, the question is then in the end whether the ocean carrier’s real objective is to have the supply chain work best, to have the containers back in their yard as quickly as possible, or to maximize its revenue from D&D fees. I leave it up to your imaginination to guess the answer. In the paper, we however show that it is possible for the carrier to devise a pricing structure such that its own profit margin remains unchanged. Hence, for a carrier there is no reason not to switch to such a new pricing structure. Curious to learn how carriers would be prepared to take next steps.

While the research that Dr Christopher Mejia and I are conducting has not yet fully concluded, we believe that the family-owned and operated nanostores that are so common across most of the developing world are here to stay over the next decade. In megacities such as Mexico City and Bogota, data we have obtained and analyzed show that over the past decade the number of nanostores has increased rather than declined. In China, the percent of retail sales that goes through the independent channel has increased. And in many countries (not only developing), supermarkets and hypermarkets are having difficulty retaining market share or even surviving.

However, many opportunities abound for new technology to enter into this market, leveraging the high density of stores that allows for convenient proximity shopping with the efficiency, resources and assortment that is present online. Examples of such platforms have been started by companies like BeeQuick in Beijing, using a virtual network of thousands of nanostores to facilitate rapid home delivery of convenience products. Another, in this case already very successful company in Go-Jekin Indonesia, which links thousands of nanostores and other nano-services such as massage and food with a platform of motorcycles to do delivery in Jakarta and other cities.

It is then sad to see that the reputed magazine El Financero reports (“Corner stores could disappear”, 25 September 2016) so poorly about the actual developments. They interview the intiator behind the Tecno Tienda: a government supported initiative to bring technology into nanostores. As mentioned above: huge opportunities abound. However, the magazine brings this as a means to save the channel. They cite a repeating myth (which I have heard on many occasions in Mexico but nobody has been able to show me the data for) that for every Oxxo opened, 15 nanostores close. Let’s call this the (presumed) Oxxo effect. Oxxo is a successful convenience store chain in Mexico that has grown rapidly operating on a franchise model. I am not sure of the exact number of Oxxos opened in Mexico City, but it well over 1000 over the past years. If the Oxxo-effect is true, this would imply that in Mexico City, in the past few years, at least 15,000 stores would have been closed. However, the data from the national statistics agency INEGI suggests that the number has actually grown.

There is a strong misperception amongst many business people in Latin America about the interaction between the channels. The success of the Oxxos barely goes at the expense of the nanostores. It is at the expenses of other channels, in particular the supermarkets. The single exception might be the rich and developed areas of Mexico City, such as the Polanco neighbourhood. Further, I am truely convinced that technology could bring a revolution to the nanostores. We should learn from a successful business model like Go-Jek and from struggling initiatives like Frogtek

When our work is completed, I will update this post with our graphs and results. However, if a reputed magazine such as El Financero brings forward apparent false insights, I need to raise this flag.

As firms become progressively more tightly coupled in global supply chains, rather than being large vertically integrated monoliths, risks and opportunities associated with activities upstream or downstream will increasingly impinge upon their own wellbeing. For a firm to thrive, it is increasingly imperative that it be aware of economic, environmental and social dimensions of the entire supply chain it belongs to, and that it proactively monitor and manage those. Finding efficient solutions towards a more sustainable supply chain is increasingly important for managers, but clearly this raises difficult questions, often without clear answers. Together with my co-editors Yann Bouchery, Charles Corbett and Tarkan Tan, we have been fortunate to include many leading researchers in our book. Each and everyone of them provide insights for students and practitioners, based on the latest academic research.

Managers and other practitioners will find a comprehensive yet concise overview of all recent academic work in this rapidly expanding field. While the book certainly is not easy bedtime reading (some chapter even contain a bit of math), all chapters are well accessible for the interested and committed manager or consultant. Each chapter provides a concise overview of a particular topic. We address a wide variety of supply chain decision fields, such as network design, inventory management, transportation planning, sourcing, and distribution. Similarly, for many of these fields we address a variety of sustainability concerns, from more ecologically oriented impacts such as carbon emissions, energy usage and water scarcity, to social responsibility topics like responsible sourcing and child labor.

University Professors and students will find this book the best suitable supporting material for their (graduate) course on sustainable supply chain management, whether this is in a business program, an industrial engineering program, or an industrial ecology program.

Together with the publisher, we have been able to keep the price of the book accessible. Furthermore, for teachers and students at universities at which the library has acquired the e-book, a cheap softcover version can be purchased from Springer.

Order Information

Hardcopy

Buy at Springer // Amazon // Barnes & Noble // bol.com (NL)

E-book

Buy at Springer // Google Play

Eerder vandaag stelden Wirken, Duisenberg Mohandis en Van den Boom in De Volkskrant dat studenten collegegeld per vak zouden moeten kunnen betalen als ze dat willen. Hun idee en het experiment daarmee in Amsterdam zijn interessante ontwikkelingen die ik met interesse zal gaan volgen. Toch kijk ik er met heel veel zorg naar. Een algemene ontwikkeling naar een systeem waarin studenten alleen maar losse vakken volgen en niet meer een samenhangende opleiding, is geen goed idee, om meerdere redenen.

Uit vele onderzoeken blijkt dat studenten die voltijds studeren, deel gaan uitmaken van de community van hun instelling. Dat is een sleutel voor studiesucces. Juist de binding met hun universiteit, met docenten en andere studenten, zorgt ervoor dat ze zich betrokken voelen en zichzelf volledig inzetten voor hun eigen opleiding en ontwikkeling. Maar het zorgt er ook voor dat ze in aanraking komen met studenten van andere opleidingen, dat ze breder kijken en kiezen, en dat ze zich ontwikkelen.

Een mooi voorbeeld van succes zijn de zelfstandige studententeams van Delft, Eindhoven en Twente die de eerste plaatsen bezetten op de afgelopen World Solar Challenge in Australië, met hun auto’s op zonne-energie. Die zijn juist ontstaan door betrokken voltijds studenten, vanuit verschillende studierichtingen, ondersteund door docenten en het netwerk van de universiteiten.

Het laten betalen per vak, zal ertoe leiden dat studenten zich in veel mindere mate onderdeel voelen van hun community. Dit zal hun leervermogen hinderen en hun ontwikkeling belemmeren. Juist het stimuleren van de leeromgeving door kleinschalige intensieve interactie tussen docenten en studenten en tussen studenten onderling heeft bij de (technische) universiteiten de afgelopen jaren erg veel aandacht gehad en heeft tot goede resultaten geleid.

De TU Eindhoven heeft enkele jaren terug haar bacheloronderwijs geheel vernieuwd, en aansluitend ook de masteropleidingen. Daarbij staat keuzevrijheid centraal, maar wel met intensieve coaching, onder meer om ervoor te zorgen dat er goede, coherente programma’s worden samengesteld. Die aanpak is op allerlei manieren een succes. Studenten kiezen niet alleen breder dan voorheen, maar nemen ook vaker deel aan multidisciplinaire projecten, en ze zijn succesvoller in het afronden van hun studie. Binnen het huidige systeem van financiering is flexibiliteit dus mogelijk, en hiervoor is geen complexe bureaucratie van betalingen per studiepunt nodig.

Samenhang is een ander punt van zorg bij vrije keuze. Curricula van universitaire opleidingen zijn weloverwogen samengesteld. Je kunt daar niet naar willekeur losse elementen uit halen. Een opleiding is meer dan een snackautomatiek waar je naar gelieve een gerecht uit kunt halen.

Overigens is er een alternatief pad mogelijk om studenten de tijd te geven om tijdelijk andere activiteiten te ontplooien: geef ze de mogelijkheid om per semester onderwijs te volgen. Dat is een veel eenvoudiger oplossing, met minder administratieve werklast, en met behoud van samenhang in de opleiding. Er moet dan een oplossing worden gevonden voor het feit dat studenten dan formeel de student-status verliezen als zij een semester niet ingeschreven zijn aan de universiteit, wat gevolgen heeft voor bijvoorbeeld hun ov-jaarkaart en hun ziektekostenverzekering. Ik zou het toejuichen als hierover initiatieven door de Kamer zouden worden genomen.

Bovenstaande opinie is in vrijwel ongewijzigde vorm geplaatst in De Volkskrant op 16 december 2015. Studenten reageren inmiddels ook in De Volkskrant dat dit studenten belemmert om juist iets extra’s te doen.

Earlier this week I had the pleasure, together with Dr Fabian Sting of RSM Erasmus University, to present the Netherlands Industrial Excellence Award 2015 to Philips Drachten.

When I was a student in the 1980s in Eindhoven, the city was still buzzing with Philips research, development, and manufacturing. By now, only research is remaining at a large scale, with much of the development and most of the manufacturing gone. Due to a cost imperative and insufficient attention on the strength and requirements for manufacturing in the Netherlands, Philips’ presence in the Netherlands has been reduced drastically.

In the Northern Netherlands (what we refer to as “Friesland”) , in the town of Drachten, however, the local management have taken their destiny into their own hands. Only recently this factory had been reduced to merely producing the shaver heads, which were considered a proprietary technology and too risky to move offshore. Management, product engineers, and process engineers have collaboratively developed a remarkable business case to re-shore an increasing amount of volume from China back to the Netherlands. This flagship re-shoring has drawn lots of attention in media worldwide, although unfortunately just the changing cost differential dominates that media attention.

Based on our assessment, Dr Sting and I are convinced the key value proposition of the Philips Drachten plant within the global Philips manufacturing base is not just in the cost differential, but in the continuous drive to improve and innovate, in both process and in products.

This is exemplified by three key characteristics of the Drachten plant:

1. Reaping the benefits of co-Location.

Product and Process development are located in one site and literally “sit in one room”. Intense collaboration has led to tremendous cost reductions (via better design-for-manufacturing and more efficient product ramp-ups) but also to product innovation driven by process innovations (via enhanced design freedom through better manufacturing processes).

2. A culture of continuous improvement that is shared down to the factory floor.

Via an effective Hoshin process strategy execution is followed through to the factory floor, with line managers fully aware of strategic priorities. Further, an amazing 90% of shop floor workers have participated in Kaizen process improvements to enhance safety, environmental, and productivity performance.

3. The factory as a network player.

Philips Drachten collaborates very intensely with its suppliers, particularly in steel, to develop new materials and processes. This in turn allows for designing better products. Philips invests heavily in these long-term supplier relations with mutual visits of not only the process engineers, but also of shop floor workers. Additionally, Philips as a major employer in the Friesland region takes responsibility for the local capability development, with heavy involvement in increasing enrolment and commitment with the regional vocational school and regional SME to enlarge the overall employment base.

Their strategy has enabled Philips Drachten to reshore production from China: Mid- and high-end production have been brought back from China, benefiting from co-location efficiencies as well as from higher degrees of automation and helped by increasing Chinese wages. This also enables more responsive production in the fashion driven shaver business.

In conclusion, it is not just the plant, but the entire Philips Drachten operation as a central part of its surrounding ecosystem that has convinced us to declare them the winner of the Netherlands’ Industrial Excellence Award 2015.

Are you interested in competing for next year’s award? Apply here

This blog was also published on 30 September 2015 on my LinkedIn page