Introduction
As companies and businesses grow, they often begin to segment into various profit centres where each division acts a kind of mini-company of its own. This can take many forms including legally independent corporate entities under an umbrella, geographically separated departments, or simply imposed divisions within a single company. The case I would like to explore could be any of these, but specifically I’m interested in situations where these separate divisions are expected to interact and transact with one another. In many cases, because each division acts independently we often see incentives between these divisions become dis-aligned with the group’s goals.
I will first provide a hypothetical case in order to better understand the problem. I will then explore inter-divisional collaboration with a focus on structure and the asset creation workflow, ignoring common profit-motives, looking at Disney as a case in point. I will then describe the common financial incentive models used in an attempt to realign incentives across divisions. Next I would like to explore how these traditional models are impacted when the asset being transacted is intellectual property as opposed to a physical produced good. Finally I will consider new structures within the group to better facilitate collaboration and therefore group profit maximization.
Note:
1. In much of the literature “transfer pricing” refers not only to incentives, but also has tax implications and certain rules are imposed by law and jurisdiction. In this article I am only interested in aligning incentives across divisions, I am not interested in what is accounted for tax purposes which is another animal entirely.
2. Throughout this article I may use the words “division”, “vertical”, “subsidiary” and “profit centre” interchangeably. Although different, for purposes of this article they are effectively the same and the goal of reaching common group incentives is also shared among them.
The Common Problem (Profit Centre Incentives)
If we consider a simple example of a company with two subsidiary profit centres where FoamCo produces and sells foam padding and SofaCo produces and sells furniture. SofaCo requires foam padding as an input material for the production of furniture and can purchase on the open market or from FoamCo. At the same time FoamCo can either sell on the open market or to SofaCo.
Each has the goal of maximizing its own profits. Generally we would assume that this in turn maximizes profit for the group. However, this may not always be the case. There are many reasons the company may want its divisions to trade with one another internally
not ceding profits to a third party when it has the components in-house
while the divisions themselves may feel they can be more profitable by trading externally
SofaCo can buy less expensive foam elsewhere
Although it is common that a parent would want its subsidiary profit centres to transact with one another, it is not always the case
FoamCo has a strong brand and therefore can command a higher margin from the market, while SofaCo is a budget brand that can produce higher profits by purchasing budget foam.
So we have a potential problem, the individual divisions are motivated to act in a way that may not be aligned with the goals of the group.
This is not a new problem, and there are many ways we can structure internal pricing to manipulate incentives amongst our divisions. Before we look at that, I’d like to consider how asset creation flows within a company and how that may alone produce different incentive structures.
The Disney Model
I was discussing the topic of inter-divisional incentives with a friend recently and while describing the issues they have within their own company they also commented on how Disney seems to be “winning at this”. They were frustrated that across the various divisions of their company there was so much potential for synergy that wasn’t being taken advantage of, so much profit was bleeding out of the group because third parties were being favoured as customers or suppliers over internal divisions. His comment about Disney was referring to how they are consistently able to successfully monetize across multiple verticals from music & movies to toys & computer games.
My view was that this occurs because they consolidate efforts around a single set of core assets which they develop initially and then drive down into the divisions. Each division does not operate independently first. In the case of Disney we are of course referring to the cartoon characters. Here it is important to note that the characters aren’t the initial creation, rather a concept for a movie is created and this leads to characters which are then fed to each division as their foundational asset to be monetized.
Directionality & Intentionality (The Core Asset)
This is important. The structure at Disney is “top-down”. One use of the asset occurs first — the movie — and this then drives the monetization across toys, games, soundtracks etc.
Each division or vertical knows from day-one that the movie being created will be sold as a sound track, that a game will be created, theme park rides will be developed, and the characters will be turned into toys.
A single core asset drives down into the various verticals which are designed to monetize components of the core asset in different ways.
This is the difficulty in many companies where verticals operate as profit centres independent of one another. The assets are there, but there is seldom a single core asset which can generate intentionality through each division, and even if a division has a valuable asset there is no clear path to monetization into the other divisions. Instead, each division is a disparate island focused on generating it’s own assets within it’s own vertical. They are not designed or organized around a shared goal.
Financial Solutions
Later we will look at how we can try to simulate some of Disney’s behaviour without the same top-down structure. But first let’s consider common solutions built around financial incentives that try to solve the problem of internal collaboration.
Coming back to FoamCo and SofaCo, we can assume that FoamCo has some costs associated with producing their foam. Some of this will be costs that don’t change with small fluctuations in production, things like rent, salaries, and facilities, we call these fixed costs (FC) which for our purposes are flat at all levels of production.
We’re assuming they’re not going to expand their facility just for SofaCo’s needs and are therefore limited to their current production capacity.
FoamCo will also face marginal costs (MC). These are the costs that fluctuate with each unit manufactured which in our case would be something like the Polyurethane raw material used for making foam. For each piece of sofa-foam they produce, it requires an additional piece of raw polyurethane foam which costs some money.
For simplicity, we will assume that sofa-foam is a standardized unit. And that marginal costs are directly proportional to the quantity produced.
In order to explore how we can create incentives for FoamCo and SofaCo, let’s assume the following:
- FoamCo’s padding sells foam for $10 per piece
- Their marginal costs are $3 per piece
- Their fixed costs are $10,000 for the year, we’ll say approximately $2 per piece. Understanding that as production increases the allocation to each piece decreases.
Note: The total price that SofaCo pays for one piece of sofa-foam is itself a marginal cost from SofaCo’s perspective.
There are a number of common solutions or structures, each with its own strengths and weaknesses. Let’s now see what some of the most common structures look like.
S1: Market / List Pricing
A simple solution is for the company to just say that SofaCo should purchase foam from FoamCo at FoamCo’s market price of $10. This seems fair at first. However, it does mean that there is also no incentive for either company to interact with one another. If FoamCo can find customers willing to pay more they may choose not to sell to SofaCo. On the other hand if SofaCo can find a better deal elsewhere, or if FoamCo doesn’t have competitive pricing, or if FoamCo’s pricing is higher because of a strong brand, SofaCo may feel they are better off purchasing elsewhere.
S2: Marginal Cost Pricing
Under marginal cost pricing we set out that FoamCo will sell to SofaCo at a discounted rate equal to its marginal cost of $3.
The idea here is that SofaCo has an incentive to purchase internally and in theory it doesn’t make FoamCo worse off because any extra costs incurred in order to produce those units for SofaCo are being covered.
The trouble is that there is then no contribution toward FoamCo’s fixed costs and they will incur an overall loss especially if they aren’t in fact able to increase production, but rather have to divert some production to SofaCo that might otherwise have been sold on the open market at $10.
FoamCo’s profits will be understated relative to their total production, while SofaCo will have overstated profits due to their special deal on foam.
S3: Full Cost Pricing
Under this model, FoamCo would charge $5 to SofaCo, this is the sum of their marginal cost and allocated fixed cost per unit. This is somewhat more fair to FoamCo than the previous model because it gives SofaCo a better deal than the open market, but covers all of FoamCo’s costs.
This model theoretically also reflects the actual costs SofaCo would have if they were to build their own factory and produce their own foam. Economies of scale notwithstanding.
Full Cost Pricing still suffers from the problem of understating FoamCo’s overall profits and FoamCo is still better off selling onto the open market and would therefore only sell to SofaCo if their production capacity exceeds the external market demand.
S4: Markup Pricing
This model assumes a set markup dictated by the parent/board/group that is added to FoamCo’s total costs. This markup is therefore effectively the profit that FoamCo generates from the sale.
This at least ensures that FoamCo generates some profits from the sale of its goods to SofaCo. It, too, comes with a number of drawbacks.
If the margin is too small, FoamCo will still have a larger incentive to sell onto the market, or it will face understated profits while SofaCo profits will be overstated. If the margin is too high, SofaCo is better off buying foam on the open market first or have understated profits.
Interestingly, this precarious balance is just a mini version of the flaws experienced by any centralized-planning economy.
In all of the above cases, if FoamCo is selling a large proportion of its production to SofaCo, we can also imagine an outcome where FoamCo has little incentive to optimize operations or reduce production costs, because they’re always covered anyway.
S5: Negotiated Pricing
This is exactly what it sounds like. We simply let the two divisions negotiate amongst themselves. This can work, but often results in sub-optimal outcomes where the stronger negotiator wins. It it also very inefficient as it adds time to the process and can lead to disputes amongst managers. Depending on the agreed price, it can easily result in all the same failures we’ve seen above.
S6: Dual Rate Pricing
This method is an accounting work-around to the problem, but highly increases each teams motivation to collaborate while also more effectively and fairly allocating revenues and profits to both divisions. It is also not without it’s flaws and some might argue that for all its fairness, it also introduces all of the flaws of other models.
The way it works is that FoamCo gets allocated full market price for the foam they sell to SofaCo, while SofaCo gets allocated a cost only equal to the marginal total cost of production (or some other cost at the group’s discretion). The difference between the market price and cost is booked into a corporate account outside of either profit centre’s P&L in order to balance the bookkeeping.
This highly increases the likelihood of SofaCo purchasing internally. It also ensures that FoamCo gets the same revenue as if they sold on the open market.
Of course the main disadvantages are that it creates no incentive for SofaCo to seek out better pricing or try minimize its cost on foam. FoamCo also has less incentive to improve production efficiency or marketing efforts especially if SofaCo makes up a large portion of its sales… it can simply increase the price of foam and thereby increase profits.
The Digital Economy — Software
Which approach is chosen really comes down to the nuances of each business, the size of the market, the size of the division’s production, the goals of the corporate parent, and of course the motivations and culture of each team, not to mention each divisions cost structures and how they compare to the market. There is no single correct answer, but the above do provide us with a foundation from which we can assess and build.
The textbook solutions also only take us so far. In reality, we don’t live in a perfectly competitive market, our products don’t always have limited supply, clear marginal and fixed costs, or a known market value. In fact, when we look at the digital economy there are a number of differences.
Let’s consider the very common software business, whether a SAAS subscription model or a simple one-time paid download, this immediately introduces an essentially flat-cost structure. Each unit of sales generates negligible additional cost to the seller. They do have fixed costs of their offices, software maintenance and upgrades, and so on. But these can be budgeted and don’t fluctuate in direct relation to relatively small fluctuations in sales volumes.
We also don’t really have a maximum production capacity in the same way as with physical goods. In the physical economy a given production line has a limitation beyond which it cannot grow without capital injection to purchase more machinery or expand into a larger factory. In the digital economy, once the asset is produced, it can be reproduced ad infinitum without any change to the production line, or even if the production line were later removed.
I recognize that at certain points there will be a need for increased customer service and support, or pressure to keep growing and thus increase features or sales teams. But these to, in my view are more like tiers than directly proportional to individual sales units.
When marginal cost is zero*, and we cannot allocate fixed cost based on a per unit measure, almost all of our previous solutions fall away. We are left with market based pricing or negotiated pricing. We could also implement a form of dual-rate pricing where a discounted rate is agreed to at which costs will be allocated to the purchasing division, while the full price is awarded to the seller and the difference accounted to the parent. However, the key difference here is that the price is calculated downward from the market price, rather than upward from the actual costs.
*Marginal cost is seldom zero, but in this case and in cases I will discuss later, the costs are either negligible or are more closely related to sales value (% commissions or royalties) than sales volume (fixed $ per unit costs)
Intellectual Property — Copyright & Music
Moving further along the spectrum from our digital example above is the concept of intellectual property in the form of copyrights. An excellent example being music. Here again we have a fixed cost that is incurred ahead of the asset being produced, and once produced there are little to no further marginal costs other than perhaps commissions or commonly a percentage royalty to the creator.
We are all familiar with the common sources of revenue in the music industry — streaming on Spotify, downloads, CDs & Vinyl, ticket sales of live concerts — but another revenue source is licensing and synchronization.
Synchronization is the use of music in another medium. Whenever you hear music in a movie, tv show, or commercial, that’s a case of synchronization.
This can be a significant source of revenue for songwriters and artists alike especially if their music is used in a feature film or high-budget television show.
Where software generally has a published price or at least a small range of prices for various options, music does not have this feature. For each track, each commercial, each compilation, each movie, or each television episode there is a new negotiation where payments can range from free (because it provides “exposure” for the music) to hundreds of thousands of dollars for large productions and widely recognized tracks.
Once again we cannot rely on marginal cost, full cost, markup, or traditional dual rate pricing models here. And in this case we can’t even use the “market rate” to guide us to a reasonable discount dual price model. We really only have negotiated pricing to rely on, or a fixed price model where the parent simply sets our fixed prices for various use cases.
Negotiated pricing in this scenario comes with all the problems of market pricing in traditional physical economies. The selling division has an incentive to focus their energy on high-value deals and/or high value tracks in their catalogue. This can result in the selling division missing lower revenue opportunities even though each sale is effectively pure profit, while the buying division is then forced to seek out music on the open market, resulting in that cost being captured as profit by a competing third party seller.
This is important to recognize. In a case where we have little to no marginal cost for a sale, every sale is pure profit. This means that every lost sale, is a direct lost profit opportunity for the parent company or group.
Each sale will often require approval of the songwriter / publisher / or other third party who will receive a royalty or % of the sale. But this is only a binary choice limitation on sales, not an actual cost limitation.
Fixed price models begin with the major hurdle of complexity. Firstly, in catalogue complexity — hundreds of thousands of songs spanning genres, popularity, and age — and secondly, usage complexity — major Hollywood feature films to small localized broadcast television commercials, or even compilation albums for corporate gifts.
Music and similar copyright products are also plagued by a knowledge problem. When companies have massive catalogues of music, it is impossible for a sales person to be knowledgeable about all the music in the catalogue and they therefore focus their efforts on key titles where they can make the most revenue per use. This is very different from both software sales models and traditional physical economies where the product catalogue is far more limited and easier for sales and customer service teams to learn and understand.
The above can also result in an unfortunate situation where revenue is perpetually generated to successful artists to the detriment of smaller, newer, independent artists, but this is again a topic for another day.
Back to Disney
Now that we have seen a number of approaches that can be used to manipulate incentives within a company and across profit centres, as well as considered why these may not apply in the case of digital goods and intellectual property, I’d like to consider a few adapted approaches that will help us achieve our goals. To reiterate, our goal is find reasonable solutions that provide the right incentives for profit centres to transact with one another when they have no reason to do so and would otherwise operate independently.
We cannot feasibly expect large corporations to restructure their entire operations, nor is it always realistic to think that every company can coordinate around a single core asset, in many cases there simply is no standardized or consistent workflow that generates core assets within just one division or vertical.
In many cases, the divisions already have entrenched structures of their own and generate significant revenues from other activities. There is little reason to restructure for a small proportion of revenue and risk breaking other aspects of the business that do work. But nonetheless we want to encourage a similar kind of exploration where a company sees what is being produced by its profit centres and acts to cross pollinate these assets to be leveraged for other verticals.
Let’s evolve our previous case of SofaCo and FoamCo, and instead let’s call them MusicCo and MovieCo. MusicCo is a subsidiary of the group and owns a massive catalogue of music and seeks to monetize that music by way of synchronization and other licensing opportunities. MovieCo is movie production company that requires music for its movies. It is also a subsidiary of the group parent.
To keep things simple, we’ll assume that MusicCo has complete rights to monetize the music and can unilaterally decide on any particular opportunity. To introduce concepts of composition vs recorded copyrights, publisher, songwriters, labels, and performers here would complicate this article far more than it will add value for the reader.
Let’s discuss a few concepts that I believe could be implemented in order to improve our inter-divisional collaboration problem. This will only serve as an introduction of the idea, and each would need to be developed further and in far greater detail to fit the particular situation and requirements of any user or use case. I’ve divided these into pricing solutions and structural solutions.
Pricing: Dual Rate Pricing 2.0
Although the traditional dual-rate pricing model uses actual costs and market pricing to generate the dual rates, we can apply the same foundational concept to different cost/price sources.
For example, perhaps MovieCo is allocated with zero cost, while MusicCo is allocated a price equal to the last placement of that kind for that track or similar tracks.
We want MovieCo to have an incentive to seek out potential tracks at MusicCo first, before looking to third-parties. As the transaction costs MusicCo very little, we don’t have to worry as much about them losing money, but we do want there to be sufficient incentive for them to facilitate MovieCo’s search.
This introduces some complexity when trying to allocate “similar tracks” or “similar placements”, but if the value of collaboration is great enough, the costs associated with categorization in this fashion could quickly be justified.
With physical goods we were concerned with FoamCo not optimizing production or not bothering to sell elsewhere due to guaranteed sales of a portion of their production volume. Now, because we are not limited by production capacity, any sale by MusicCo to MovieCo is purely additive and does not reduce the availability of product for MusicCo to sell. So we actually negate some of the pitfalls of dual-rate pricing that apply to physical goods.
Pricing: Internal Price Bands
I’ve already commented that it is likely unrealistic to assign a price to the assets, because of the shear size of the catalogue and variability of both the music itself and the use cases. However, for internal purposes we could perhaps create price bands in a similar way companies have salary bands.
We could group our catalogue by certain price bands (e.g. newcomer, local talent, sensation). This could be as simple or as complicated as the group wished. A very rudimentary approach would be to have each band associated with the number of streams a track or artist has generated. Although this is far from optimal, it provides a very simple and measurable starting point.
To add some granularity in case our internal divisions produce more than just blockbuster movies, we could create a kind of matrix as shown below separating the use cases as well. (e.g. budget commercial, low budget television, Hollywood blockbuster).
The goal here is to reduce the transaction costs for MusicCo because they are limited by sales capacity rather than production capacity, while also simplifying the negotiation and purchasing process for MovieCo.
There are a number of problems with this approach, especially in how to fairly and accurately allocate tracks to price bands or categories. This is not a simple task as one can imagine. The benefit however is that since there is little marginal or transactional cost to the transaction it could become an almost passive revenue source for MusicCo that reduces search cost for MovieCo, reduces actual cost for both MovieCo and the group, while also increasing and facilitating far more placements for the music being represented.
Pricing: Precedent Pricing
This does not require a lot of detail. It is a very simple concept that is being used already in one form or another. The concept here is to apply previous precedents as a benchmark for internal pricing. This sounds obvious, but I believe that these precedents could be used far more intentionally, independently or in combination with other approaches.
It could for example be used to refine Internal Price Bands by using the knowledge from external placements to guide pricing internally.
Alternatively, precedents externally can be used to continuously update and/or set the “market price” that is allocated to MusicCo when MovieCo uses one of their tracks.
Structural: Internal Sales Incentives
A relatively simple way to generate incentives within an existing structure is to directly compensate existing sales teams to sell internally. Creating a separate bonus/commission/assessment plan that includes evaluation of internal sales.
This either needs to be a fixed requirement, or very carefully structured so that the incentives are well balanced against their external sales. When a sales person is paid based on these metrics, they will lean toward the sales channel that will generate the maximum return. We have to be weary not to make these incentives too small so as to be ignored, but they should also not be so great that the person focuses entirely on internal sales.
I’m not a strong believer in this option due to the careful balance it requires although it could be useful in smaller teams where more dedicated interal sales are not feasible. We will expand on the dedicated sales model later in this article.
Structural: Internal Purchasing Incentives
What models work best are highly dependant on each situation, company, and asset. In some cases the bottleneck may not be on the sales side, but rather on the purchasing side. We could therefore create a similar incentive that rewards purchasing departments for using internal assets. This can be more difficult than with sales, because the rewards structure is even more sensitive to pricing and budgets.
Of course this is identical to the various price based approaches discussed before if it is only in the form of a lower cost accounted to the purchasing team and therefore only benefits those responsible at the P&L level. With this particular approach the implication would instead be to reward the individual or team that is responsible for purchasing, not the entire profit centre that is purchasing.
Structural: Internal Sales / Purchasing Position
Due to the balancing act required by either of the last two approaches another model which is only feasible in larger organizations would be to have a dedicated internal sales / purchasing person. This is very similar to the Core Asset Proxy concept, but instead of having an ex-P&L team working across multiple divisions, we could define a position within a profit centre whose task is similar — to take the available catalogue and generate internal sales for those assets.
This might be difficult depending on how roles and tasks are assigned as this person effectively needs to be involved in all projects. It is likely more effective as a sales role than a purchasing role, because purchasers are supposed to seek out competing and alternative options. And internal sales person on the other hand can be focused on selling from their catalogue and is competing for the purchaser’s business.
The key differences between this position and the personal incentive models or the proxy model is that this position is dedicated only to internal sales and does not seek to generate revenue from external sources, but is a position internal to the individual profit centre.
As noted, this may make sense in companies where not all divisions need to be collaborating and so we focus resources within those profit centres we want incentivize. It is also unrealistic for smaller profit centres where a single dedicated position is too costly relative to the savings/benefits of transacting internally.
Structural: Core Asset Proxy Teams
We already discussed how Disney creates a core asset that is then leveraged and monetized by the various verticals. We want to synthesize a similar mentality in companies that don’t have this kind of structure. One way to do this is to quite literally create a corporate level team that operates outside of the cost centres of the individual divisions, but whose task it is to work with each division and be responsible for seeking out cross-division opportunities and facilitating the collaboration.
I call it a proxy because the goal is for this team to act as if it is the creator of the core asset, although it has no production of its own. It stands in for the Core Asset that doesn’t yet exist, but seeks out and finds the asset as well as how to leverage its value into the individual teams.
Structural: Ex-P&L Pollination Teams
Previous solutions seek a way to integrate existing assets into existing projects. (convince MovieCo to use music from MusicCo in their movie). A Pollination team is similar, but looks for new asset creation opportunities. (Convince the toys division to create a new toy based on the characters from the games division).
In a sense this is another proxy to core asset teams, where the main difference is the Proxy team leverages existing needs, whereas the Pollination team creates new needs.
There is no reason these even need to be separate teams, this is simply another perspective in how a company might structure and describe the teams, and how they are communicated internally. The decision to take one approach or another is entirely based on the unique aspects of the company itself.
Structural: Low-Revenue Divisions
This model is badly named, but the idea here is built on a common problem I touched on earlier; because intellectual property sales are limited by sales rather than production, sales teams tend to favour products they can generate the most revenue from the most quickly. In our case this means always looking to the most popular artists and tracks rather than newcomer or earlier stage bands, and potentially ignoring lower revenue placement requests like commercials or low-budget films, to focus on the big budget television shows and films.
Creating a person or team that is responsible only for opportunities below a certain threshold budget and/or only on a particular subset of the catalogue, could be avoid many lost opportunities.
This is not necessarily only related to internal collaboration as it often relates to external sales as well. If MovieCo is mainly producing films with smaller budgets and is therefore ignored by MusicCo because they always pitching high profile opportunities, a dedicated low-revenue person could not only solve the internal collaboration problem, but simultaneously open up the company to take advantage of a yet untapped revenue source from external sources as well.
Depending on the size and content of the catalogue, another approach could be to simply split the catalogue and assign to separate profit centres such that one deals with a smaller subset of higher profile content, while the other is responsible for the larger, lower profile subset.
Cross-divisional Staff:
We might consider another riff on the above where we take little bits of each. We could create a dedicated a new position which acts on behalf of the MovieCo, but has direct authority to facilitate and execute deals on behalf of MusicCo. Or vice versa. This may be complicated depending on the corporate structure, incentives, and copyrights involved, however, the benefit would be that MusicCo does not need to allocate resources for these uses while MovieCo has a direct channel to the catalogue so as to easily and quickly facilitate transactions.
An alternative is instead of having a salesperson from MusicCo, we could rather include someone intimately knowledgeable with MusicCo’s catalogue.
This model of course only makes sense if MovieCo is producing a high volume of content, need a large amount of music, and can safely be expected to find the right music at MusicCo.
Conclusion
As we’ve seen, there are a number of imperfections that emerge as companies grow, merge, and expand. We have a delicate balance between needing these divisions or subsidiaries to be independent and self-governing, while also ensuring they act in a way that is not detrimental to the group as a whole.
There is no single simple solution. Traditional models and text books provide us with only a basis for understanding and a starting point for what we face in reality. By keeping an open mind as we deal with competing business models, globalization, and digitalization, we can respond to these nuances with numerous potential levers and ideas that might help us improve, if not solve, the problems we face when profit centres need to operate both independently and in unison.
Note
This article is by no means exhaustive and I have been forced to make a number of assumptions in order to simplify the examples as well as a number of explanatory omissions (such as how music copyrights actually work) in my attempt to keep this post at a somewhat readable length. I welcome feedback, ideas and thoughts that may act as the catalyst for a future post and further exploration into any of the topics discussed.
Acknowledgements
Thanks also to my dad, Mauro Celotti for providing the images.