Business Value Optimization in Multidimensional Target Worlds
Theory of Constraints Throughput Accounting
The typical S-curve - business value (y-axis) across all priority combinations (x-axis)
Business Value Optimization in Multidimensional Target Worlds
November 9, 2025
A practical playbook for project & portfolio leaders
Prioritizing a portfolio sounds straightforward - line up initiatives by “business value” and get to work. But in real organizations, portfolios are messy, objectives compete, and the team you need most is usually the one you have least of. What you need isn’t a perfect mathematical answer (it doesn’t exist at scale), but a clear, defensible way to get near-optimal results fast - and to make the inevitable trade-offs explicit.
Why “best order” is a trap
If you tried to find the absolute best sequence for even a small set of projects, you’d be forced into checking every possible order - an explosion of possibilities that grows factorially. Five projects become 120 sequences; ten become millions. Exhaustive search is a dead end for any living portfolio where projects start, stop, and change every quarter.
The following graph shows the typical resulting S-curve with markers for the commonly used priority schemes:
Resulting contribution margin across all possible priority combinations
#1 Evenly distributed - resources split equally among all projects, #2 First-come, first-served (FCFS) - more or less random, #3 Revenue - highest contribution margin first, #4 Smallest project first, #5 Best revenue / total project cost first, #6 TOC - see below
The right move is to stop hunting for perfection and adopt a heuristic that consistently lands near the top of what’s possible. That’s where systems thinking - and the constraint - comes in.
Optimize at the constraint
Every delivery system is limited by one capability more than the others. Sometimes it’s backend engineering; sometimes it’s UX, data, or a specific certification team. If you prioritize to maximize value at that bottleneck, you raise throughput for the whole portfolio.
A simple, powerful signal is what I call the octane ratio:
Octane = Contribution Margin ÷ Effort in the Bottleneck Team
Rank your initiatives by this ratio and you’ll usually deliver materially more value earlier than by common alternatives like “highest revenue first,” “smallest first,” equal slicing, or first-come, first-served. Why? Because octane aligns decisions with the scarcest resource - the one that actually sets the pace.
Evidence you can see: the S-curve
When you simulate all sequences for a small set (say, five projects) and plot cumulative value delivered over time, the strategies fall along an S-curve. Equal slicing and FCFS cluster low; “revenue only” improves things but still leaves money on the table. Octane sits near the top. It’s rarely the literal best, but it’s reliably better than the usual suspects.
There is one caveat: early in execution, you might see a short “startup effect” where the constraint isn’t yet fully loaded or temporarily shifts. Don’t overreact. As flow stabilizes, octane tends to dominate.
“Which team is the bottleneck?” - answer it fast
You don’t need to guess your constraint forever or calculate permutations. Most organizations have on the order of 10–100 teams or skills. Evaluate N octane candidate sequences - one per team assumed to be the constraint:
- For each team, sort projects by (value ÷ that team’s effort).
- Compute expected completion periods and value delivered.
- Compare the results.
Two insights drop out immediately: the best sequence overall and the team that is acting as the constraint right now (it’s the team whose assumption produced the best result). It’s a light computation and easy to rerun when signals change.
Real portfolios have more than one goal
Optimizing only for throughput is rarely sufficient. You might also need to increase the number of new customers, boost order intake, improve quality, or achieve other goals. Improving one can nudge another in the wrong direction. Extend the same approach across dimensions:
- Calculate Nteams × Ngoal-dimensions sequences (one per assumed constraint).
- For each sequence, schedule and calculate its outcomes across the goal dimensions that matter this quarter (e.g., throughput, quality, acquisition, cancellations).
- You now have a compact table: rows are sequences, columns are outcomes.
Computationally, the effort scales with (number of teams) × (number of dimensions) - still perfectly tractable for a PMO with scheduling software and a spreadsheet.
Make the trade-offs visible (and manageable)
Turn those outcomes into a ranked view per dimension. Visualize each dimension as a slider from worst to best.
Example: three dimensions - three sliders
Start at the throughput-maximizing sequence and “pull” another slider up - say, quality or acquisition. The view jumps to the sequence that best improves that dimension while showing the impact on the others. Suddenly, trade-offs are no longer implicit and political; they’re explicit and navigable.
New combinations when you start to move the slider on one dimension
Examples from live portfolio conversations:
- “If we raise acquisition one notch, we give up ~8% near-term value but hold quality constant.”
- “If we favor quality, throughput dips slightly for two sprints, then recovers as defects fall.”
This is where informed leadership replaces opinion wars.
Why common strategies underperform
- Equal slicing spreads scarce capacity across too many fronts, starving the constraint and delaying everything.
- First-come, first-served is operationally simple but strategically random.
- Revenue only prioritizes the right numerator but ignores the denominator that limits flow.
- Smallest first clears the board, but not the value bottleneck.
Octane wins because it optimizes against the actual limiter. Pair it with WIP limits, and you’ll cut multitasking - the silent killer of throughput.
How to put this into practice this quarter
- Define value consistently. Use contribution margin or a defensible proxy. Be explicit and stick to it for the current cycle.
- Estimate effort by team - perhaps using rough project plans. You don’t need precision - being directionally correct is enough. Update monthly; constraints move.
- Generate N sequences. Assume each team could be the constraint, compute octane, and simulate a delivery timeline. Pick the highest-value result and note the implied constraint.
- Add 2–4 other dimensions that matter this quarter. Score the same sequences on each dimension.
- Run the slider conversation with stakeholders. Start with the throughput-maximizing sequence, then explore one notch at a time in the dimension you most want to improve. Choose the compromise deliberately.
- Govern with timeboxes. Treat the choice as a quarterly policy, not a forever law. Rerun when signals change (queues, wait times, rejected work, rising rework).
- Stabilize the system. Use Critical Chain/flow practices: protect buffers, cap WIP, and shield the bottleneck from interruptions.
The takeaway
You don’t need heroic computation to run a complex portfolio well. Focus on the constraint, rank work by value per bottleneck effort, and make trade-offs visible across the few outcomes that matter. You’ll ship more value earlier, reduce the drama in prioritization, and - most importantly - turn heated opinion into a transparent, repeatable decision process.
All the tools you need, as well as the simulation shown, can be found in our Body of Knowledge for the DolphinUniverse Community - feel free to join for free today: