Choosing Between Two Replacements
A mid-sized logistics firm must replace its aging fleet-dispatch software. The COO has asked for a recommendation between two finalists. Vendor A is the incumbent's successor product: familiar interface, $180,000 per year, migration estimated at four months, and 90% of dispatchers already know it. Vendor B is a modern cloud platform: $240,000 per year, migration estimated at seven months, requires retraining all 60 dispatchers, but offers real-time route optimization that Vendor A lacks. The COO's stated objective is 'reduce dispatch errors.' The operations manager adds a second objective: 'stop losing drivers to competitors who offer better tools.' Finance adds a third: 'hold software spend flat.' No one has defined what a dispatch error is, and the two vendors measure errors differently. A dispatcher survey from last year found the top complaint was the interface, not the routing.
1. The three objectives conflict. Explain how to structure them so the choice becomes tractable, and what must be fixed first.
Read the worked answer
The first fix is a common definition of 'dispatch error' - without it, the primary objective cannot be scored and the vendors' differing metrics will simply be compared as if they were the same. Define it operationally (e.g., wrong assignment, missed time window, manual rework) and require both vendors to report against that definition. The second fix is to separate objectives from constraints. 'Hold software spend flat' is best treated as a constraint (a ceiling, say $200,000/year) rather than an objective to be maximized, because maximizing it would trivially favor the cheapest option. 'Reduce dispatch errors' and 'retain drivers' are genuine objectives and should be weighted explicitly - for example 60/40 - with the weights agreed before scores are assigned. The third fix is to test whether the objectives are actually distinct: if the dispatcher survey shows the interface drives both errors and attrition, the two objectives may collapse into one, which changes the answer. Structure the decision as: constraints first (budget, migration window), then weighted objectives, then a sensitivity check on whether the ranking flips under plausible weight changes.
2. Generate at least one alternative that neither vendor represents, and explain why the option-generation step matters here.
Read the worked answer
A third alternative: keep Vendor A's successor and bolt on a routing-optimization module from a specialist provider, or run a hybrid in which B's optimization is piloted in one region while A remains the system of record. Another: negotiate a phased B rollout with training costs capitalized and a performance guarantee tied to the error definition. Generation matters because the framing has quietly assumed the choice is binary between two vendors, when the underlying objectives (fewer errors, better tools for retention, flat spend) do not require a single-vendor answer. A bolt-on addresses the routing gap at lower cost and migration risk; a phased rollout addresses retraining risk. Without an explicit option-generation step, the team will pick the 'least bad' of two options that may both be dominated. The test for any generated alternative is whether it satisfies the constraints and improves at least one objective without worsening another beyond its weight - if it does, the binary comparison was the wrong question.