# Innovation Portfolio Concept Card: [Company or case name] > A concept card for managing a portfolio of innovation projects as it is taught in a second-year MBA technology and > innovation strategy elective: why firms underinvest in exploration, how to classify projects by ambition (core, > adjacent, transformational), how to value risky projects with expected commercial value and a productivity index, how > to select a portfolio under a budget, why staging and kill gates add value, and how to organise for exploration and > exploitation at once. It builds on the three horizons, the Ansoff matrix and the 70/20/10 discussion in > `mba/mba-growth-strategy-card` and on staging as a real option in `mba/mba-options-real-options-card`, and links to them > instead of repeating them. It is for MBA students preparing a case or exam and for managers who run an R&D, > product or venture portfolio. The card gives the idea first, then formulas with a notation table, a fully solved > portfolio selection for Corvane Robotics, a half-solved staging example, an unsolved exercise for Tallow Creek Medical > Center, a self-test and a separate answer key. All companies and numbers are fictional. Source basis: March (1991), > Cooper (1990), Cooper, Edgett and Kleinschmidt (1999), McGrath and MacMillan (1995), Christensen and Bower (1996), > Christensen (1997), O'Reilly and Tushman (2004), Nagji and Tuff (2012) and Cohen and Levinthal (1990), written up in > original words and examples. Replace every [bracketed] value and every row marked *Example*. ## Document control | Field | Value | |---|---| | Course or project | [Technology strategy elective, session 6 / annual R&D review / venture board] | | Learner or team | [Your name or learning team] | | Organisation analysed | [Corvane Robotics (fictional) / your organisation] | | Budget and period | [USD 11.0m development budget, FY2027] | | Currency and unit | [USD m, present values at the hurdle rate] | | Estimates provided by | [Project leads, finance, market research] | | Reviewed by | [Instructor, portfolio committee or mentor] | | Date | YYYY-MM-DD | | Version | [0.1] | ## Contents - [When to use it](#when-to-use-it) - [The idea in plain language](#the-idea-in-plain-language) - [Formulas and notation](#formulas-and-notation) - [Assumptions and limits](#assumptions-and-limits) - [How managers use it in practice](#how-managers-use-it-in-practice) - [Worked example A (fully solved): Corvane Robotics development portfolio](#worked-example-a-fully-solved-corvane-robotics-development-portfolio) - [Worked example B (half solved): staging the outdoor yard robot](#worked-example-b-half-solved-staging-the-outdoor-yard-robot) - [Exercise C (unsolved): Tallow Creek Medical Center innovation budget](#exercise-c-unsolved-tallow-creek-medical-center-innovation-budget) - [Connections to other cards](#connections-to-other-cards) - [Common mistakes and misconceptions](#common-mistakes-and-misconceptions) - [Self-test questions](#self-test-questions) - [My notes and next steps](#my-notes-and-next-steps) ## When to use it **One-sentence rule:** use this card when an organisation must choose which of many uncertain innovation projects to fund, how much to put into each kind of ambition, and when to stop projects, with a fixed budget and limited people. | Use it when | Use something else when | |---|---| | An annual R&D, product or venture budget must be allocated across projects | You value one project in depth (`mba/mba-npv-irr-card`, `finance/investment-appraisal`) | | Leadership asks whether the firm invests enough in future businesses | The question is the overall growth target and gap (`mba/mba-growth-strategy-card`) | | Projects must pass gates and some must be killed | You need a detailed decision tree with many branches (`mba/mba-decision-trees-card`) | | A firm faces a new technology that could disrupt its core | You are prioritising features inside one product (`product/prioritization`) | | You must explain why a high-risk project deserves funding despite a low success probability | You need a launch plan for a chosen project (`marketing/go-to-market-plan`) | ## The idea in plain language **Exploration and exploitation.** March (1991) described the basic tension of organisational learning: exploitation (refining what the firm already knows: efficiency, execution, incremental improvement) gives returns that are positive, near and fairly certain, while exploration (search, experimentation, new technologies) gives returns that are uncertain, distant and often negative. Because exploitation pays off sooner and more reliably, organisations tend to drive out exploration over time, which makes them better at the present and worse prepared for change. An innovation portfolio is the management system that protects a deliberate amount of exploration. **Why incumbents underinvest in the wrong things.** Christensen and Bower (1996) and Christensen (1997) showed how well-run firms can fail when a disruptive technology appears: their resource-allocation processes favour projects that serve current, profitable customers, so technologies that start in small or low-end markets are starved of funds until it is too late. The lesson for portfolio design is that good project-by-project decisions, made with the core business's criteria, can add up to a bad portfolio. Separate criteria, budgets or units are often needed for projects that serve future customers. **Ambition: core, adjacent, transformational.** Nagji and Tuff (2012) proposed an innovation ambition matrix that classifies projects by how far they move from current products and current customers: **core** (optimising existing products for existing customers), **adjacent** (expanding from existing business into new to the company business) and **transformational** (new offers for new markets or needs). They reported that firms allocating about 70% of innovation investment to core, 20% to adjacent and 10% to transformational initiatives tended to outperform in their data, while stressing that the right mix differs by industry, competitive position and stage of development. This is the same idea as the three horizons (`mba/mba-growth-strategy-card`), applied to the innovation budget. **Valuing risky projects: expected commercial value.** Early-stage projects have two kinds of risk: technical (will it work?) and commercial (will it sell?). Cooper, Edgett and Kleinschmidt (1999) surveyed portfolio management practices and reported that financial methods were the most widely used, but that businesses relying mainly on them did not achieve the best portfolio results, while strategy-based approaches were associated with better ones. A common financial measure from their work is **expected commercial value (ECV)**: the present value of the launched product, times the probability of commercial success, minus the commercialisation cost, all times the probability of technical success, minus the development cost. Dividing ECV by the scarce resource (development money or engineer-years) gives a **productivity index** used to rank projects. **Selecting a portfolio is a constrained choice.** Ranking by ECV alone favours large projects; ranking by productivity index favours small efficient ones but can leave budget unused. The correct selection maximises total ECV subject to the budget, which is a 0-1 (knapsack) problem that Excel Solver handles with binary variables. Managers then add strategic constraints: minimum spend on core projects, a ceiling or floor on transformational spend, balance across markets. **Stages and gates.** Cooper (1990) described the stage-gate process: work is split into stages, each followed by a gate where a cross-functional team decides go, kill, hold or recycle against stated criteria. Staging creates value because later, larger spending happens only if earlier stages succeed: an option to abandon at each gate (`mba/mba-options-real-options-card`). McGrath and MacMillan (1995) added **discovery-driven planning** for highly uncertain ventures: start from the profit the venture must deliver, derive what has to be true, list the assumptions, and fund milestones that test the riskiest assumptions first. **Organising for both.** O'Reilly and Tushman (2004) described the **ambidextrous organisation**: exploratory units are separated from the core business in processes, structure and culture, but integrated at the senior team, which shares resources and protects the new unit. Cohen and Levinthal (1990) added a reason to keep some exploration in house even when buying technology is possible: a firm's **absorptive capacity**, its ability to recognise and use outside knowledge, depends on its own prior related knowledge and R&D. ## Formulas and notation **Expected commercial value of a project:** $$ \text{ECV}_i = \left( PV_i \times P_{cs,i} - C_i \right) \times P_{ts,i} - D_i $$ **Productivity index and overall probability of success:** $$ PI_i = \frac{\text{ECV}_i}{D_i}, \qquad P_{s,i} = P_{ts,i} \times P_{cs,i} $$ **Portfolio selection** (choose $x_i \in \{0, 1\}$): $$ \max_{x} \sum_i x_i\, \text{ECV}_i \quad \text{subject to} \quad \sum_i x_i\, D_i \le B, \qquad \text{share}_k = \frac{\sum_{i \in k} x_i D_i}{\sum_i x_i D_i} $$ **Staged project with two development stages and a launch decision** (stage $j$ costs $D_j$ and succeeds with probability $p_j$; launch only if worth it): $$ \text{EV}_{\text{staged}} = -D_1 + p_1 \left[ -D_2 + p_2 \max\left( PV \times P_{cs} - C,\ 0 \right) \right] $$ | Symbol | Meaning | Unit in the examples | |---|---|---| | $PV_i$ | Present value of the launched product's future cash flows (after launch costs other than $C_i$) | USD m | | $P_{cs,i}$ | Probability of commercial success, given technical success | 0 to 1 | | $C_i$ | Commercialisation (launch) cost, spent only if technically successful | USD m | | $P_{ts,i}$ | Probability of technical success | 0 to 1 | | $D_i$ | Remaining development cost (the budgeted, scarce resource) | USD m | | $\text{ECV}_i$ | Expected commercial value | USD m | | $PI_i$ | Productivity index: ECV per unit of development budget | Ratio | | $P_{s,i}$ | Overall probability of success | 0 to 1 | | $x_i$ | 1 if project $i$ is funded, 0 otherwise | Binary | | $B$ | Development budget | USD m | | $\text{share}_k$ | Share of funded spend in ambition class $k$ (core, adjacent, transformational) | % | | $D_1, D_2, p_1, p_2$ | Stage costs and stage success probabilities (Worked example B) | USD m, 0 to 1 | ## Assumptions and limits 1. **Inputs are guesses with a bias.** Project champions overstate PV and probabilities. Calibrate with the firm's own history (what share of projects at each gate actually launched?) and with independent estimates. 2. **Independence.** Summing ECVs assumes projects do not share risks, people or markets. Projects that rely on the same technology succeed or fail together; projects that compete for the same customers cannibalise each other. 3. **One budget constraint.** Real portfolios are limited by scarce people (software engineers, clinical staff) as much as by money. Add a resource constraint for each bottleneck. 4. **Present values hide timing.** ECV treats all values as present values at one hurdle rate. Transformational projects pay later and may deserve a different discount rate or a scenario view. 5. **ECV is a single-gate model.** It allows one stop (after technical failure). Projects with several gates are worth more than ECV suggests (Worked example B); use a decision tree for them. 6. **Strategic value is not in the formula.** Learning, absorptive capacity, platform positions and defensive value are real but hard to put in PV. Make them explicit with strategic buckets or constraints rather than inflating PV. 7. **Ambition ratios are descriptive, not rules.** The 70/20/10 split is an average reported in one study; a firm whose core is being disrupted may need far more in adjacent and transformational work. ## How managers use it in practice **Set strategic buckets first.** Leadership decides, before seeing individual projects, roughly how much of the budget goes to each ambition class (and sometimes to each market or platform). Projects then compete within their bucket. This prevents core projects, which always look safer, from crowding out exploration. **Use different criteria for different ambitions.** Core projects are judged on ECV, payback and fit with the roadmap. Adjacent projects add market evidence and capability fit. Transformational projects are judged on learning milestones: what assumption will this stage test, at what cost, and what would make us stop? **Run gates with kill discipline.** A gate that never kills is a formality. Write the go and kill criteria before the stage starts, and reallocate the money of killed projects quickly. Celebrate well-run kills: they release budget. **Review the portfolio, not only projects.** Once or twice a year, plot all projects on a bubble chart (probability of success against value, bubble size for spend, colour for ambition), check the mix against the buckets, and look for gaps (no projects in a strategically vital area) and clusters (many projects betting on one assumption). **Organise the exploratory work.** Decide which projects stay in business units and which go to a separate exploratory unit with its own processes and a senior sponsor (the ambidextrous model). Plan how successful explorations will be handed back to the core. ```mermaid flowchart LR I["Idea pool"] --> G1{"Gate 1: strategic fit"} G1 -->|"Go"| S1["Stage 1: scoping and riskiest assumption test"] G1 -->|"Kill"| K["Release budget to the portfolio"] S1 --> G2{"Gate 2: business case, ECV, bucket"} G2 -->|"Go"| S2["Stage 2: development and prototype"] G2 -->|"Kill"| K S2 --> G3{"Gate 3: launch decision, PV x Pcs > C?"} G3 -->|"Go"| L["Launch and post-launch review"] G3 -->|"Kill"| K K --> R["Annual portfolio review: mix, gaps, clusters"] L --> R R --> I ``` ## Worked example A (fully solved): Corvane Robotics development portfolio **Situation (fictional).** Corvane Robotics, the warehouse-robotics company (`mba/mba-growth-strategy-card`, Worked example A), has a development budget of USD 11.0m for FY2027 projects that are past scoping. Six projects compete for it. Values are present values at Corvane's 12% hurdle rate; the estimates were reviewed by finance and engineering (*Example* values, USD m). **Step 1: ECV, productivity index and overall probability of success.** | ID | Project | Ambition | $PV$ | $P_{cs}$ | $C$ | $P_{ts}$ | $D$ | ECV | PI | $P_s$ | |---|---|---|---|---|---|---|---|---|---|---| | IP-01 *(Example)* | Fleet analytics module v2 | Core | 18.0 | 0.80 | 2.0 | 0.90 | 3.0 | 8.16 | 2.72 | 0.720 | | IP-02 *(Example)* | Battery hot-swap kit | Core | 10.0 | 0.70 | 1.5 | 0.85 | 2.0 | 2.68 | 1.34 | 0.595 | | IP-03 *(Example)* | Third-party robot connector | Adjacent | 30.0 | 0.50 | 3.0 | 0.75 | 4.0 | 5.00 | 1.25 | 0.375 | | IP-04 *(Example)* | Piece-picking arm | Adjacent | 45.0 | 0.50 | 6.0 | 0.50 | 6.0 | 2.25 | 0.38 | 0.250 | | IP-05 *(Example)* | Hospital logistics robot | Adjacent | 40.0 | 0.40 | 5.0 | 0.60 | 4.5 | 2.10 | 0.47 | 0.240 | | IP-06 *(Example)* | Outdoor yard robot | Transformational | 120.0 | 0.30 | 10.0 | 0.25 | 3.5 | 3.00 | 0.86 | 0.075 | | | **Total** | | | | | | **23.0** | **23.19** | | | Excel, with PV in D2, $P_{cs}$ in E2, $C$ in F2, $P_{ts}$ in G2 and $D$ in H2: ECV `=(D2*E2-F2)*G2-H2`, PI `=I2/H2`, $P_s$ `=G2*E2`. For IP-01: $(18.0 \times 0.80 - 2.0) \times 0.90 - 3.0 = 12.4 \times 0.90 - 3.0 = 8.16$. All six projects have positive ECV, but together they need USD 23.0m, more than twice the budget. **Step 2: rank by productivity index.** Taking projects in PI order while they fit funds IP-01, IP-02 and IP-03 (USD 9.0m, total ECV USD 15.84m). The next project in PI order, IP-06 (USD 3.5m), does not fit in the remaining USD 2.0m, and neither does any other; USD 2.0m stays unspent. **Step 3: optimise the selection.** Solver set-up: binary decision cells K2:K7; objective `=SUMPRODUCT(I2:I7,K2:K7)` to maximise; constraint `=SUMPRODUCT(H2:H7,K2:K7)<=11`; method Simplex LP with K2:K7 binary. Checking all 64 combinations gives the same answer: | Selection | Projects | Spend (USD m) | Total ECV (USD m) | Core | Adjacent | Transformational | |---|---|---|---|---|---|---| | Rank by PI *(Example)* | IP-01, IP-02, IP-03 | 9.0 | 15.84 | 55.6% | 44.4% | 0.0% | | ECV-optimal *(Example)* | IP-01, IP-03, IP-06 | 10.5 | 16.16 | 28.6% | 38.1% | 33.3% | Shares are of funded spend, for example core in the optimal set `=3.0/10.5` = 28.6%. The optimal portfolio swaps the battery kit (core, ECV 2.68) for the outdoor yard robot (transformational, ECV 3.00), uses more of the budget and adds about USD 0.33m of ECV (16.160 less 15.835 before rounding; IP-02's ECV is exactly 2.675). ```vega-lite { "$schema": "https://vega.github.io/schema/vega-lite/v5.json", "description": "Corvane Robotics (fictional) innovation portfolio: probability of success against ECV, bubble size = development cost", "width": 360, "height": 240, "data": {"values": [ {"id": "IP-01", "ambition": "Core", "p_success": 0.720, "ecv": 8.16, "dev_cost": 3.0}, {"id": "IP-02", "ambition": "Core", "p_success": 0.595, "ecv": 2.68, "dev_cost": 2.0}, {"id": "IP-03", "ambition": "Adjacent", "p_success": 0.375, "ecv": 5.00, "dev_cost": 4.0}, {"id": "IP-04", "ambition": "Adjacent", "p_success": 0.250, "ecv": 2.25, "dev_cost": 6.0}, {"id": "IP-05", "ambition": "Adjacent", "p_success": 0.240, "ecv": 2.10, "dev_cost": 4.5}, {"id": "IP-06", "ambition": "Transformational", "p_success": 0.075, "ecv": 3.00, "dev_cost": 3.5} ]}, "layer": [ { "mark": {"type": "circle", "opacity": 0.6}, "encoding": { "x": {"field": "p_success", "type": "quantitative", "title": "Overall probability of success", "scale": {"domain": [0, 0.8]}}, "y": {"field": "ecv", "type": "quantitative", "title": "ECV (USD m)"}, "size": {"field": "dev_cost", "type": "quantitative", "title": "Development cost (USD m)", "scale": {"range": [100, 1200]}}, "color": {"field": "ambition", "type": "nominal", "title": "Ambition"} } }, { "mark": {"type": "text", "dy": -16}, "encoding": { "x": {"field": "p_success", "type": "quantitative"}, "y": {"field": "ecv", "type": "quantitative"}, "text": {"field": "id", "type": "nominal"} } } ] } ``` **Step 4: add a strategic guardrail.** Suppose the board insists on the 70/20/10 logic in a softened form: core at least 40% and transformational at most 20% of funded spend. The best feasible portfolio is then IP-01, IP-02 and IP-03 again (ECV USD 15.84m, USD 9.0m spent): the guardrail costs about USD 0.33m of ECV and removes the only transformational project. **Interpretation and recommendation.** The numbers do not settle the question; they show its price. Funding the yard robot raises expected value slightly and buys the only option on a new market, at a 7.5% chance of success. Excluding it saves the board from a project it may not understand but leaves Corvane, whose core is crowded (see `mba/mba-five-forces-card`), with no exploration at all. A good recommendation: fund IP-01, IP-03 and IP-06, but fund IP-06 in stages with a kill gate after stage 1 (Worked example B shows staging is worth more than ECV assumes), and fund the battery kit next year from the money released if IP-06 is killed. IP-04 and IP-05 have the lowest PI: return them for a cheaper first stage or a partner (borrow) route, as in `mba/mba-growth-strategy-card`. ## Worked example B (half solved): staging the outdoor yard robot **Situation (fictional).** IP-06's development budget of USD 3.5m can be split into two stages: stage 1, a feasibility study of navigation in rain and snow (USD 1.5m, success probability 0.5), and stage 2, a field prototype (USD 2.0m, success probability 0.5). The overall technical success probability is $0.5 \times 0.5 = 0.25$, as in Worked example A. At the launch gate, Corvane pays USD 10.0m to commercialise, and the launched product is worth USD 120.0m with probability 0.30. **Step 1: value at the launch gate.** $120.0 \times 0.30 - 10.0 = 26.00$, positive, so Corvane would launch. **Step 2: staged expected value.** $$ \text{EV}_{\text{staged}} = -1.5 + 0.5 \times \left[ -2.0 + 0.5 \times 26.00 \right] = -1.5 + 0.5 \times 11.00 = 4.00 $$ Excel: `=-1.5+0.5*(-2+0.5*MAX(120*0.3-10,0))`. **Step 3: compare the three ways of funding it** (complete the table). | Funding approach | Expected value (USD m) | Excel | |---|---|---| | Commit everything up front (no gates: spend 3.5 and 10.0 whatever happens) | [ ] | [ ] | | One gate after all development (the ECV model of Worked example A) | 3.00 | `=(120*0.3-10)*0.25-3.5` | | Two development stages plus the launch gate | 4.00 | see Step 2 | **Step 4: discovery-driven questions.** 1. What PV of the launched product makes the staged project break even? [ ] 2. If stage 1 succeeds, what is the project worth at gate 2? [ ] 3. If new evidence lowers the commercial success probability to 0.20, what is the staged expected value? [ ] 4. [Name the assumption stage 1 should test first, and the result that would kill the project.] ## Exercise C (unsolved): Tallow Creek Medical Center innovation budget **Situation (fictional).** Tallow Creek Medical Center, a regional hospital, has a USD 5.0m innovation budget for the coming year. Five proposals have passed scoping (*Example* values, USD m, present values at the hospital's hurdle rate; for a non-profit hospital, PV includes cost savings and new service margin). | ID | Project | Ambition | $PV$ | $P_{cs}$ | $C$ | $P_{ts}$ | $D$ | |---|---|---|---|---|---|---|---| | TC-01 *(Example)* | Digital patient check-in | Core | 6.0 | 0.85 | 0.8 | 0.90 | 1.2 | | TC-02 *(Example)* | Bed-management analytics | Core | 9.0 | 0.70 | 1.0 | 0.80 | 1.5 | | TC-03 *(Example)* | Remote monitoring after discharge | Adjacent | 14.0 | 0.50 | 2.0 | 0.70 | 2.0 | | TC-04 *(Example)* | Home infusion service | Adjacent | 12.0 | 0.55 | 2.5 | 0.75 | 1.8 | | TC-05 *(Example)* | AI triage assistant | Transformational | 30.0 | 0.30 | 3.0 | 0.40 | 2.5 | 1. Compute ECV, PI and overall probability of success for each project. Show the Excel formulas. 2. Select the portfolio that maximises total ECV within USD 5.0m, with Solver. Does ranking by PI give the same answer? 3. Compute the ambition mix of your selection and compare it with the 70/20/10 benchmark. 4. TC-05 has a negative ECV. Argue for and against funding a small first stage anyway, and say which assumption that stage should test. Consider patient safety and the governance questions in `mba/mba-ai-risk-governance-card`. 5. Which non-financial criteria should a hospital add to the selection, and how would you build them into the model (constraint, bucket or score)? ## Connections to other cards - `mba/mba-growth-strategy-card`: the growth gap, the three horizons and build, borrow or buy; this card allocates the innovation budget inside that plan. - `mba/mba-options-real-options-card` and `mba/mba-decision-trees-card`: staging, abandonment and the value of waiting, in full. - `mba/mba-npv-irr-card`: the PV inputs for each project. - `mba/mba-technology-strategy-guide`: the elective this card belongs to; disruption, technology cycles and appropriability. - `mba/mba-platform-strategy-card`: a platform move is usually an adjacent or transformational bet (IP-03 above). - `mba/mba-resource-based-view-card`: dynamic capabilities and why some projects build capabilities the firm needs. - `mba/mba-change-management-card`: moving people and processes when an exploratory project scales. - `mba/mba-ai-strategy-card`: scoring and valuing AI use cases with the same discipline. ## Common mistakes and misconceptions 1. **Ranking by NPV alone.** Large projects crowd out several smaller ones with more value per dollar; check PI and solve the constrained selection. 2. **Ranking by PI and stopping.** A greedy PI ranking can leave budget unused and miss a better combination, as in Worked example A. 3. **One probability for everything.** Merging technical and commercial risk hides which gate should test what. 4. **Treating 70/20/10 as a law.** It is an average from one study; the right mix depends on the threat to the core. 5. **No kill criteria.** Projects that are never stopped absorb the budget of better ones; write kill criteria before each stage. 6. **Judging exploration by core metrics.** Transformational projects fail on short-term ROI by design; judge them on learning milestones. 7. **Optimism bias in inputs.** Champions overstate PV and probabilities; calibrate with historical gate pass rates. 8. **Ignoring shared resources.** Two projects that need the same five engineers cannot both run on schedule. 9. **Counting correlated bets as diversification.** Five projects that all depend on one sensor technology are one bet. 10. **Valuing staged projects as if committed.** Up-front commitment can make a good staged project look negative (Worked example B). 11. **Separating exploration without a bridge.** An exploratory unit with no link to the core cannot hand back its successes; ambidexterity needs integration at the top. ## Self-test questions 1. State March's (1991) distinction between exploration and exploitation and why organisations drift towards exploitation. 2. Why, according to Christensen and Bower (1996), do well-managed incumbents underinvest in disruptive technologies? 3. Name the three ambition classes of Nagji and Tuff (2012) and give a Corvane example of each. 4. Write the ECV formula and explain why $C$ is multiplied by $P_{ts}$ but $D$ is not. 5. A project has PV 50, $P_{cs}$ 0.6, $C$ 8, $P_{ts}$ 0.7 and $D$ 5. Compute ECV and PI with Excel formulas. 6. Why can ranking by productivity index fail to give the best portfolio? Use Worked example A. 7. What does a gate decide, and why does staging add value? Use the numbers of Worked example B. 8. What is discovery-driven planning (McGrath and MacMillan 1995)? 9. What is an ambidextrous organisation (O'Reilly and Tushman 2004)? 10. What is absorptive capacity (Cohen and Levinthal 1990), and what does it imply for outsourcing all R&D? 11. Give two reasons why summing project ECVs can overstate a portfolio's value. ## My notes and next steps | Item | My answer | |---|---| | Portfolio I am analysing and its budget | [Portfolio], [budget] | | Buckets I would set before seeing projects | [Core %], [adjacent %], [transformational %] | | Input I trust least and how I will test it | [Input], [test] | | Project that needs staging and its first kill criterion | [Project], [criterion] | | Organisational home for exploratory work | [Unit and sponsor] | | Review dates | [Two days], [one week] | ## Sources and further reading **Primary works.** - March, J. G. (1991). Exploration and exploitation in organizational learning. *Organization Science*, 2(1), 71-87. - Cohen, W. M., and Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and innovation. *Administrative Science Quarterly*, 35(1), 128-152. - Cooper, R. G. (1990). Stage-gate systems: A new tool for managing new products. *Business Horizons*, 33(3), 44-54. - Cooper, R. G., Edgett, S. J., and Kleinschmidt, E. J. (1999). New product portfolio management: Practices and performance. *Journal of Product Innovation Management*, 16(4), 333-351. - McGrath, R. G., and MacMillan, I. C. (1995). Discovery-driven planning. *Harvard Business Review*, 73(4). - Christensen, C. M., and Bower, J. L. (1996). Customer power, strategic investment, and the failure of leading firms. *Strategic Management Journal*, 17(3), 197-218. - Christensen, C. M. (1997). *The Innovator's Dilemma*. Boston: Harvard Business School Press. - O'Reilly, C. A., and Tushman, M. L. (2004). The ambidextrous organization. *Harvard Business Review*, 82(4). - Benner, M. J., and Tushman, M. L. (2003). Exploitation, exploration, and process management: The productivity dilemma revisited. *Academy of Management Review*, 28(2), 238-256. - Nagji, B., and Tuff, G. (2012). Managing your innovation portfolio. *Harvard Business Review*, 90(5). **Books and textbooks (further reading only).** Cooper, Edgett and Kleinschmidt, *Portfolio Management for New Products*; Schilling, *Strategic Management of Technological Innovation* (McGraw-Hill); O'Reilly and Tushman, *Lead and Disrupt* (Stanford University Press); Christensen and Raynor, *The Innovator's Solution* (Harvard Business School Press); Dixit and Pindyck, *Investment under Uncertainty* (Princeton University Press). **Open resource.** OpenStax, *Entrepreneurship* (CC BY-NC-SA 4.0), for background on opportunity recognition and testing assumptions. ## Answer key **Worked example B.** | Funding approach | Expected value (USD m) | Excel | |---|---|---| | Commit everything up front | -4.50 | `=-(1.5+2)-10+0.5*0.5*0.3*120` | | One gate after all development (ECV) | 3.00 | `=(120*0.3-10)*0.25-3.5` | | Two stages plus launch gate | 4.00 | `=-1.5+0.5*(-2+0.5*MAX(120*0.3-10,0))` | The second development gate adds USD 1.00m over ECV (stage 2's USD 2.0m is spent only if stage 1 succeeds), and all gates together add USD 8.50m over up-front commitment (the USD 10.0m launch cost is spent only after technical success). Discovery-driven answers: (1) break-even PV solves $-1.5 + 0.5[-2.0 + 0.5(0.30\,PV - 10)] = 0$, so $PV = ((1.5/0.5 + 2.0)/0.5 + 10)/0.30 = 66.67$ (Excel: `=((1.5/0.5+2)/0.5+10)/0.3`); (2) at gate 2 the project is worth $-2.0 + 0.5 \times 26.00 = 11.00$; (3) with $P_{cs} = 0.20$ the launch value is $120 \times 0.20 - 10 = 14.00$ and the staged value $-1.5 + 0.5 \times (-2.0 + 0.5 \times 14.00) = 1.00$, still positive. (4) Stage 1 should test the riskiest technical assumption first (reliable navigation in rain and snow at a set accuracy); failure to reach it kills the project before the larger stage 2 spend. **Self-test.** 1. Exploitation refines existing knowledge (efficiency, execution); exploration searches for new knowledge (experimentation, variation). Exploitation's returns are nearer and more certain, so learning processes reinforce it and crowd out exploration. 2. Their resource-allocation processes respond to current, profitable customers and to projects with large, near and measurable returns; disruptive technologies first serve small or low-end markets that these processes reject. 3. Core: fleet analytics module v2 (IP-01). Adjacent: third-party robot connector (IP-03) or hospital logistics robot (IP-05). Transformational: outdoor yard robot (IP-06). 4. $\text{ECV} = (PV \times P_{cs} - C) \times P_{ts} - D$. Commercialisation is paid only if the product works (probability $P_{ts}$); development must be paid to find out. 5. ECV $= (50 \times 0.6 - 8) \times 0.7 - 5 = 22 \times 0.7 - 5 = 10.40$ (`=(50*0.6-8)*0.7-5`); PI $= 10.40/5 = 2.08$ (`=10.4/5`). 6. Ranking by PI fills the budget greedily and can leave money unused when the next project does not fit. In Worked example A, PI ranking funds IP-01, IP-02 and IP-03 (ECV 15.84, USD 2.0m unused); the optimal set IP-01, IP-03 and IP-06 has ECV 16.16. 7. Go, kill, hold or recycle against stated criteria. Staging adds value because later spending happens only after earlier success: the yard robot is worth 4.00 staged, 3.00 with one gate and -4.50 committed up front. 8. Planning for highly uncertain ventures that starts from the required profit, derives what must be true, lists assumptions and funds milestones that test the riskiest assumptions first. 9. A firm that runs exploratory units separately from the core in processes, structure and culture, while integrating them through the senior team, which shares resources and protects them. 10. The ability to recognise, assimilate and apply outside knowledge, which depends on prior related knowledge. A firm that outsources all R&D can lose the ability to judge and use external technology. 11. Projects may share risks (correlated success), compete for the same customers (cannibalisation) or need the same scarce people; inputs also tend to be optimistic. ## Revision history | Version | Date | Author | Change summary | |---|---|---|---| | 0.1 | YYYY-MM-DD | [Your name] | First copy for [course, case or portfolio review] | | 0.2 | YYYY-MM-DD | [Your name] | *Example:* Replaced Corvane projects with our FY pipeline, added an engineer-hours constraint | ## Standard and licence note Structure aligned with the innovation management literature named above (March, Cooper, Cooper, Edgett and Kleinschmidt, McGrath and MacMillan, Christensen and Bower, O'Reilly and Tushman, Nagji and Tuff, Cohen and Levinthal). The explanations, notation, examples, exercises and answers are original; no textbook, article or published case text is copied or paraphrased, and Corvane Robotics, Tallow Creek Medical Center and every project named are fictional. This template is independent and not affiliated with or endorsed by any school, author or publisher. Check your own course materials and your organisation's portfolio rules before relying on any figure. Template licence: MIT.