Book a call

Expand the Offer or Improve the Core: Which Move Should Come Next?

Expansion depends on the strength and capacity of the core. Preserve all independent study scopes, the approximate failure estimate and exact simplification counts. Sixteen tapered blades surround a stable dark green core on fresh green; one yellow blade extends the boundary while white modules retain the whole.

Improve the core when a measurable constraint blocks repeatable value; test an adjacent offer when the core is strong, the need is independently observed, distinctive capabilities transfer, and the operating system can absorb the added complexity. Summary

Two proposals arrive at the same growth meeting.

The first says the current offer needs work. Delivery is inconsistent. Customers reach value too slowly. Support carries exceptions that the product was supposed to remove. The proposal asks for six months to repair the core.

The second says the market is ready for more. Existing customers keep mentioning a nearby problem. A new tier, service, product, or channel could open another revenue line. The proposal asks for six months to build the adjacency.

Both sound responsible. Both can cite customers. Both can produce a handsome forecast. Only one can receive the team's undivided attention.

The choice is often framed as caution versus ambition. That framing is emotionally satisfying and strategically useless. Improving the core is not automatically cautious. Expanding the offer is not automatically ambitious. They are different learning investments with different failure modes.

The useful question is not, "Do we want growth?" It is: Which unresolved uncertainty can reverse the decision, and what is the smallest credible way to resolve it?

The two moves learn on different clocks

In 1991, James March gave organizations a durable vocabulary for this tension. Exploitation refines what is already known. Exploration searches beyond it. Exploitation favors efficiency, implementation, and selection. Exploration favors variation, experimentation, and discovery.

Core improvement usually leans toward exploitation. It asks whether a known promise can become faster, more reliable, easier to sell, cheaper to deliver, or more consistent. Offer expansion usually leans toward exploration. It asks whether a different need, buyer, channel, or configuration can produce value.

The returns arrive on different clocks. A core constraint can sometimes be measured and removed quickly. An adjacency may require time to reveal demand, repeat purchase, sales friction, service burden, and second-order effects. Exploration can protect the future while depressing the present. Exploitation can strengthen the present while making the organization overconfident in what it already knows.

That is why a universal resource split fails. A meta-analysis of 143 studies and 257 independent samples found performance value in exploration and exploitation, with important contextual differences. Another meta-analysis focused on small and medium firms covered 34 studies and 5,488 observations. Its central warning is especially useful for a constrained business: trying to do everything can consume the coordination capacity needed to do either mode well. A broader organizational ambidexterity meta-analysis likewise shows that the relationship depends on how ambidexterity is defined and where it is practiced.

Evidence scope

The trade-off is extensively studied, but not reducible to one rule

Three substantial research programs explain different parts of the decision without producing a universal expansion threshold.
Reading note

Shi et al., 2020; Wenke et al., 2021; Peng et al., 2023. The units and research questions differ and are intentionally not pooled.

The research does not remove judgment. It improves the judgment. Ask which learning mode the company needs now, which one it can execute faithfully, and how long it can wait for evidence.

A weak core corrupts the expansion test

Suppose the current offer has a retention problem. Customers buy, struggle to reach the promised result, and leave. A new offer could still attract attention. It might even create short-term revenue. But it enters the same operating system that failed to make the first promise repeatable.

The expansion test is now contaminated. If the new offer underperforms, the team cannot tell whether the need was weak or the delivery system was weak. If it grows, the business may celebrate while service cost, churn, and exceptions compound behind the revenue line.

This is the strongest case for improving the core first: a binding constraint makes every adjacent result harder to interpret.

A binding constraint is not a vague feeling that the product could be better. It is a measurable point where the current promise stops becoming repeatable value. It could be activation, reliability, fulfillment time, gross margin, sales conversion, retention, quality, support load, or implementation capacity. The exact measure depends on the offer.

Core improvement becomes strategic when it names four things:

  • the outcome the core is supposed to produce;
  • the constraint that prevents repeatability;
  • the intervention expected to remove that constraint;
  • the evidence that would show the constraint has moved.

Without those four elements, "improve the core" can become an elegant shelter for endless polish. Teams redesign screens, rewrite messages, clean a backlog, and call the activity focus. The work may be useful. It is not yet a growth argument.

An adjacency must borrow more than a name

Now take the opposite case. The current offer retains well. Delivery is stable. The business understands its economics. Customers repeatedly expose a nearby unmet need. The team can explain why its capabilities make it unusually able to serve that need.

Here, refusing expansion can become its own risk. The organization may optimize a mature offer while the market changes around it.

But "nearby" needs a stricter definition than similar branding or a familiar customer. Chris Zook and James Allen's five-year study of growth outside the core reported that roughly 75 percent of adjacency expansions failed. Their practical argument is not that companies should remain small. It is that successful adjacency usually begins with a strong core and repeats a proven capability into a nearby customer, geography, channel, product, or need.

Adjacency warning

Most adjacent expansions in one five-year study failed

Proximity to the core does not make expansion safe when customer fit, transferable capability, and operating readiness remain unproven.
Reading note

Zook and Allen, 2003. The approximation is a preparation warning, not a forecast for a specific business.

The brand must have permission to travel as well. David Aaker and Kevin Keller showed that consumers evaluate extensions through associations with the parent brand and the perceived fit of the extension. A later meta-analysis of 2,134 effect sizes from research published between 1990 and 2020 gives a much larger view of extension success and its moderators.

Perceived fit matters, but it is only one layer. A credible adjacency passes three tests:

Need: The problem appears in customer behavior and evidence before the proposed solution enters the conversation.

Permission: Buyers can understand why this company belongs in the new decision.

Capability: The business can reuse a distinctive process, asset, relationship, technology, or body of knowledge.

Dynamic-capabilities theory makes the final test sharper. Advantage does not come from noticing an opportunity that everyone can see. It comes from sensing, seizing, integrating, and reconfiguring capabilities in a way competitors cannot easily reproduce.

An extension with customer permission but no operating capability is a good story with an expensive second act.

Variety has a revenue side and an operating side

Offer expansion changes more than the catalog. It changes demand planning, sales training, configuration, pricing, inventory, delivery, documentation, data, support, architecture, and the number of exceptions the organization must recognize.

The revenue arrives on one line. The complexity is distributed across the company.

In a field study using weekly data from 108 distribution centers over three years, product variety had an inverted-U relationship with total sales. Variety helped, then hurt. The source does not give every business a transferable optimum. It establishes a more important principle: the value of another option is not linear.

Variety threshold

More offers help until the operating system starts paying the bill

Variety can add demand and still reduce total performance after forecasting, inventory, and service complexity exceed the system's headroom.
Reading note

Wan, Evers, and Dresner, 2012. Shape reproduced conceptually; source coefficients are not reconstructed.

At first, variety can help customers find a better fit. It can serve more situations, reduce compromise, and open useful segments. After a point, the operating system begins to pay a tax. Forecasts become noisier. Inventory spreads. Fill rates weaken. Service teams learn more exceptions. Product architecture accumulates dependencies. Salespeople spend more time explaining the portfolio.

Research on product complexity treats this as a system, not a count. A systematic review examined 93 empirical articles. A design-for-variety study covered 702 manufacturers in 22 countries. Another study followed 283 distribution centers across 26 four-week periods to examine variety, forecast bias, and inventory. Work on product portfolio architectural complexity separates multiplicity, diversity, and interrelatedness.

Complexity evidence

Offer count is only the visible edge of portfolio complexity

Multiplicity, diversity, and interrelatedness determine whether a new offer is a small addition or a system-wide operating change.
Reading note

Marti et al., 2019; Eroglu et al., 2019; Wan and Sanders, 2017; Jacobs and Swink, 2011.

Those distinctions matter. Ten offers built from one modular platform can be less complex than three offers with different data models, fulfillment processes, compliance rules, and service teams. A new offer is not expensive because it is new. It is expensive when it multiplies difference across a tightly connected system.

New ventures face the same nonlinearity. Patel, Azadegan, and Ellram found an inverted-U relationship around product variety in new ventures, shaped by operational capability. Again, the useful conclusion is not "less variety." It is "variety must fit the system that carries it."

Improving the core can be the growth move

The most striking counterexample comes from heavy equipment, not a software dashboard.

Caterpillar faced a product portfolio with 37,920 configurations. A simple reduction could have destroyed demand by removing combinations customers valued. The company instead modeled customer substitution and complexity cost. The resulting method reduced the portfolio to 135 configurations. Sales increased almost 7 percent.

Core simplification case

Caterpillar removed almost every configuration and sold more

Core improvement can create growth when the business models what customers will substitute and which complexity no longer earns its cost.
Reading note

Shunko et al., 2018. The method modeled customer substitution and complexity cost; it was not indiscriminate assortment cutting.

The case, documented by Shunko and colleagues, is not permission to delete 99 percent of every catalog. It shows why core improvement can produce growth when it removes complexity without removing the attributes customers value.

That is a deeper form of focus than doing fewer things. It changes the architecture of choice and delivery.

This matters because expansion proposals often compare their visible upside with the core's current performance. That is the wrong counterfactual. The proper comparison is the adjacency's expected result versus the result of removing the core's most consequential constraint.

The core option deserves a forecast too.

Compare tests before you compare programs

Executives often ask for two business cases: one for core improvement and one for expansion. Each team then spends weeks making its preferred future look inevitable.

A better comparison begins one level smaller. Design one test for each uncertainty.

For the core, the test might remove one activation barrier, standardize one delivery exception, simplify one configuration family, repair one reliability constraint, or change one part of the service model. The test needs a measurable baseline and a result that can mature soon enough to influence the decision.

For the adjacency, the test might be a paid pilot, manual service, prototype with a real commitment, channel experiment, limited geography, constrained tier, or pre-sale with explicit delivery conditions. The test must expose willingness to act, not merely willingness to compliment the idea.

Then compare the tests on four dimensions:

  1. Decision value: Which result can genuinely reverse the choice?
  2. Reversibility: Which test limits irreversible cost and customer harm?
  3. Time to evidence: When will the signal be mature enough to trust?
  4. Displacement: What work, attention, or operating capacity does the test remove from something else?

Portfolio practice supports this wider view. Cooper, Edgett, and Kleinschmidt describe leading portfolio management as pursuing value, balance, and strategic alignment. A single financial estimate cannot carry all three goals. Neither can a single enthusiasm score.

Next-move decision

Resolve the earliest uncertainty that can reverse the choice

Core repeatability, observed adjacent need, transferable capability, and complexity headroom create a sequenced test rather than a generic growth score.
Reading note

Editorial synthesis of the paper's organizational-learning, adjacency, portfolio, complexity, and dynamic-capabilities evidence.

The sequence is intentionally asymmetric. If the core is not repeatable, diagnose the constraint before treating expansion results as clean evidence. If the core is strong, the adjacent need is observed, capabilities transfer, and complexity has headroom, test the adjacency before another round of internal optimization makes the company better at yesterday.

The next move is the uncertainty worth buying down

There is no permanent identity called a core company or an expansion company. Healthy organizations move between exploitation and exploration as their evidence changes.

The discipline is to avoid financing two vague programs with the same optimistic language.

Improve the core when the current promise cannot produce repeatable value, when its constraint also threatens an adjacency, or when simplification can release meaningful capacity. Test expansion when the core is stable, the adjacent need is independently visible, customers grant permission, distinctive capabilities transfer, and the operating system can carry the additional variety.

When those conditions remain uncertain, do not debate harder. Run the two smallest tests that make the uncertainty comparable.

Growth is not the act of adding more. It is the act of increasing the organization's capacity to create repeatable value. Sometimes that requires a new offer. Sometimes it requires making the existing one finally work as promised.

References

Summary

Treat core improvement and offer expansion as competing learning investments: identify the uncertainty that can reverse the decision, then fund the smallest reversible test that resolves it without hiding operating cost.

  1. Name the measurable constraint that prevents the current offer from producing repeatable value.
  2. Verify the adjacent customer need independently from enthusiasm for the proposed solution.
  3. List the distinctive capabilities, assets, and relationships the new offer can genuinely reuse.
  4. Estimate the added forecasting, architecture, service, inventory, support, and coordination load.
  5. Define one test for the core constraint and one test for the adjacency's demand and fit.
  6. Compare the two tests by decision value, reversibility, time to evidence, and work displaced.

What Are the Formulas for Virality in Social Media? (Part 1)

Is SEO Actually the Right Growth Channel for Your Business?