Over the years, it has become increasingly frustrating to witness that most companies seeking to improve their supply chain performance are setting themselves up for failure through their own RFP (request for proposals) and RFQ (request for quotes) processes. The short story is that you should be careful what you wish for, especially for enterprise software. However, this state of affairs deserves a longer explanation.
At the core, there are two fundamental problems that, almost without exception, lead to the undoing of supply chain optimization initiatives: first, a misguided thirst for control; second, bureaucracy taking precedence over reality.
The most common form of misguided thirst for control, in a supply chain context, is the ambition to set up a grand plan which reflects the most minute details of everything that is about to happen, so that supply chain execution can be reduced to a simple matter of pure orchestration. The grand plan is conceived as a collection of fine-grained time-series1 forecasts, ideally per SKU per day. Operationally, the “forecasting” technicalities end up eating all the resources poured into the initiative, as the grand plan requires the very best forecast possible.
Unfortunately, the grand plan is terminally flawed when it comes to the irreducible uncertainty of the future. Production decisions, inventory allocations, price moves established based on the “grand plan” consistently turn fragile whenever the market forces end up deviating from the plan, no matter how accurate the underlying forecasts may be. The consequences are not subtle: stockouts, delays, write-offs.
While the grand plan is clearly at fault: by design, it is over-reliant on impossibly accurate forecasts that won’t ever be delivered. Yet, companies usually double down: if the grand plan failed, it’s because we didn’t plan enough. Thus, we should step up our planning and forecasting efforts. In particular, the lack of sophistication2 of the forecasting method will also, almost invariably, be pointed out as the root cause for the failings of the grand plan, letting, once again, the forecasting efforts consume all the resources.
The software vendor underpinning3 the execution of the grand plan should be getting their share of the blame. However, enterprise software vendors, over the decades, have become experts at deflecting culpability. “Bad data” will be pointed out - by the vendors - as the true root cause of the wild inaccuracies of the forecasts that came with the grand plan. In the end, the broader market context gets blamed, which is the loose modern equivalent of saying it was destined to be so.
A few years later, when the dust has settled4, the top management can’t help but notice that the problems haven’t gone away: stockouts, delays and write-offs. Thus, the company decides to move forward, tasking a committee to proceed through an RFP (or RFI, or RFQ). Unfortunately, committees cannot think outside the box: they are the box. The committee is the literal embodiment of the “grand plan” vision.
The underlying thirst for control will manifest itself through the production of very long5 RFP. Invariably, 100+ questions and requirements are conjured, ranging from impossibly vague concerns like Can the solution leverage weather data? to incredibly narrow capabilities like Can a planner manually override a safety stock for a given SKU?. The former type of concerns would warrant an entire book chapter; while the latter usually imply the wrong sort of solution. Either way, answers tend to err deep in the realm of irrelevance.
The RFP brings us to the second fundamental problem: the bureaucracy taking precedence over reality. Supply chain planning, at the scale of a large company, requires a sizeable bureaucracy. Indeed, there are significant benefits in having even an approximate alignment between marketing, sales, production, transport, etc. Without this bureaucracy, there wouldn’t even be a large company in the first place, there would only be a federation of small companies operating under the same banner.
Thus, planning, being the main activity of this specific flavor of bureaucracy, comes with its own long lists of roles, rules, processes, and workflows. Some of this complexity is necessary but a large part is just accidental. Bureaucratic elements tend to outlast their usefulness. Thus, while this supply chain bureaucracy is needed, it comes with a lot of dead weight which does not contribute in any meaningful way to the well-being of the company. The age old joke being that every CEO knows that half of their company does nothing of value, but they don’t know which half.
Now, when an initiative is adopted to improve a bureaucratic process like the planning, it frequently turns out that the people who are most readily available to contribute to the initiative - starting with the RFP - are precisely the ones that are the least involved with the “good parts” of the supply chain bureaucracy, the parts actually creating value for the company. Indeed, the “good parts” are typically under immense pressure to actually cope with everything the world throws at them: a pandemic, a flood, a new tariff, a union strike, a supplier bankruptcy, etc.
As a consequence of having the people most disconnected from the reality of the business steering the changes of the bureaucracy itself, is that you get even more accidental complexity and an even bigger disconnect from reality. Putting layers of tech sophistication on top of what is otherwise an unnecessary bureaucratic element only makes the situation worse overall. The element becomes a lot more opaque, and thus harder to root out later on.
From the enterprise software vendor perspective, however, this disconnect from reality is a boon. The primary risk, for the vendor, is being caught as the obvious source of value destruction for the company. Eliminating the possibility of success is unfortunate, but largely inconsequential as the bulk of the revenues are attached to the initial setup. Then, a well-placed multi-year commitment can go a long way to secure a stream of revenues that is fully detached to any operational performance whatsoever.
Fixing those two problems - the misguided thirst for control and the bureaucracy taking precedence over reality - is easier said than done. It mostly takes fortitude from the top management and their teams. Unfortunately, fortitude cannot be bought or acquired, and it frequently happens to be in fairly short supply in many larger organizations.
Seeking control over its own supply chain is a reasonable proposition for a large company. However, seeking the sort of control that eliminates uncertainty is wishful thinking. Uncertainty is irreducible. Moderation is a virtue. Instead of dismissing uncertainty, it should be embraced. At a technical level, one of the best options we have to do so is probabilistic forecasts: assigning a probability for every possible future.
On a conceptual level, I have rarely met supply chain practitioners who disagreed with the superiority of probabilistic forecasts compared to average forecasts. There is little debate that dismissing uncertainty is a much greater risk than trying to cope with it, albeit imperfectly. However, as soon as we start delving into the immediate consequences of this proposition, many practitioners feel very insecure. All the roles and processes that rely on the assumption that the future is controlled get exposed for what they are: empty promises.
Thus, the main challenge with probabilistic forecasts is not the technique. It’s coming to terms with a future that won’t be controlled. There is no grand plan anymore, merely a strategy that steers all decisions toward the path of better outcomes, even if those outcomes remain ill-defined . Vanquishing this feeling of insecurity is precisely what requires fortitude. The crux of the challenge is to confront the “perception” of risk rather than the risk itself.
Speaking of risk, the larger the company, the more tempting it is to “play it safe” as opposed to “play it real”. The political forces at play in a large company punish failure more strongly than they reward success. Career-wise, in a large company, merely avoiding visible failures (and time) is all it takes to progress within the organization.
Thus, within a large company, it’s tempting to steer the evolution of the planning toward elements that are guaranteed not to fail6: collaboration, data visualization, workflows, storytelling, … It turns out that those sorts of elements are the ones that please the bureaucracies the most. Unfortunately, by avoiding chances of failure, chances of success are eliminated as well.
The litmus test “Can this element badly hurt the company?” goes a long way to establishing a connection with reality. If the worst that can happen is very little, then at best, success will be inconsequential as well.
Yet, it takes fortitude to constantly and routinely re-expose the initiative to danger. Naturally, endangering the initiative is only a by-product of ensuring that the stakes are real. Nevertheless, it’s an ongoing source of friction which goes against all the instincts of the various bureaucracies at play.
Time-series forecasts are just one type of forecast among many. It is usually not even the most appropriate form of forecasts as far as supply chain is concerned. However, many companies treat time-series forecasts as if it was the only option in existence. ↩︎
It also happens that enterprise software vendors may precisely be offering incredibly sophisticated planning and forecasting solutions. The vendors are both vocal and convincing when it comes to establishing that if their software solution had been used, the grand plan never would have failed the company the way it did. All it takes to fix the future grand plan is to adopt the software. ↩︎
As most of the supply chain initiatives are actually carried out through spreadsheets, Microsoft Excel usually gets blamed. Although spreadsheets have their shortcomings, the actual issue is that the undertaking itself is heading in the wrong direction. If you’re travelling in the wrong direction, travel speed is inconsequential. ↩︎
The institutional memory of a company is constantly eroded as people change jobs, even when employees happen to stay within the company. As a rule of thumb, in a company of 1000+ employees , few will remember anything from 5 years ago, about the struggles that were plaguing the position they now occupy. ↩︎
RFP aren’t just long, they also tend to be boring as well. The boredom factor is so intense that frequently, nobody manages to ever proofread the document. As a result, the document is not only badly written, but so full of typos that the RFP looks like the hasty homework produced by a lazy college student. ↩︎
I am not saying that such features don’t bring value to the company. My point is that these types of “soft” features are too often privileged as a “safe bet” precisely because it’s very hard to prove any of the downsides that may come along with the feature. For example, it’s near impossible to prove that a “chat” system that lets supply chain planners collaborate, is just a distraction and a net loss for the company. Thus, very few practitioners will go out of their way to resist the introduction of such a “chat” feature in a planning environment. ↩︎