Defining Cut, Keep, Add, and Test Rules

Assortment optimization becomes real when the organization moves from diagnosis to decisions. A SKU scorecard can identify strengths and weaknesses, but value is created only when leaders decide which products to protect, remove, introduce, expand, localize, substitute, or test. This is also where many assortment programs become difficult. Every SKU has an owner, a supplier, a history, and often a reason someone believes it should stay. Without explicit decision rules, the process can become political, inconsistent, and slow.

This chapter explains how to define practical cut, keep, add, and test rules. The aim is not to create rigid formulas that ignore merchant judgment. The aim is to create a disciplined operating language so that teams make similar decisions for similar situations, understand why exceptions exist, and can explain the expected customer and financial impact of each action. Good rules protect what matters, remove what does not, reveal what is missing, and create controlled experiments where uncertainty is too high for immediate rollout.

5.1 Keep Rules: Protecting Strategic, Profitable, and Customer-Critical SKUs

Keep rules define which SKUs must remain in the assortment and why. Many assortment projects focus heavily on what to cut, but a clear keep logic is the best defense against accidental damage. If the organization cannot explain why a SKU is protected, it will either preserve too much out of caution or cut too much in pursuit of productivity.

Protect strategic SKUs: Some products deserve protection because they support the retailer’s broader proposition. These may include private-label flagships, opening price point items, premium trade-up products, exclusive items, local hero products, or products that define category authority. A strategic SKU may not always rank at the top on sales or margin, but it must have a clear role. The role should be documented, not assumed.

Protect profitable SKUs: Profitability should be measured in margin dollars, margin rate, inventory productivity, markdown exposure, and supplier funding. A SKU that creates strong penny profit and turns efficiently is usually a keeper unless it conflicts with category strategy or creates excessive complexity. Retailers should not overvalue gross margin percentage alone. A product with a high rate but low demand may contribute less than a lower-rate item that builds baskets and moves steadily.

Protect customer-critical SKUs: Some items matter because customers expect them, rely on them, or use them as a reason to choose the retailer. These include trip drivers, loyalty items, need-state essentials, important sizes, trusted brands, and items with low substitutability. Removing these SKUs can create more damage than the scorecard suggests because the lost sale may include the entire basket or future visits.

Protect architecture SKUs: Good assortments need structure. A retailer may need a good-better-best ladder, clear price tiers, key brands, private-label alternatives, required sizes, major flavors, core colors, essential specifications, or solution-completing accessories. Some SKUs should stay because they make the category easier to shop and give the customer confidence that the retailer understands the mission.

Keep rules should be explicit enough to challenge. “Customer favorite” is not sufficient unless supported by repeat rate, loyalty penetration, ratings, customer feedback, or a clear category role. “Vendor priority” is not sufficient unless it translates into funding, innovation, supply reliability, exclusivity, or strategic leverage.

  • Economic keep: retain SKUs with strong sales, margin dollars, turns, and growth within the comparable group.
  • Customer keep: retain SKUs with loyalty, repeat purchase, trip driver value, low substitutability, or high satisfaction.
  • Architecture keep: retain SKUs required to complete price, brand, size, use-case, or good-better-best coverage.
  • Strategic keep: retain SKUs that support private label, exclusivity, localization, category authority, or supplier strategy.

The final keep list should not be a list of untouchables forever. It is a set of SKUs protected for this planning cycle because they currently earn their role. Even protected SKUs should have review dates and performance expectations.

5.2 Cut Rules: Identifying Weak, Redundant, Low-Rotation, and Operationally Costly SKUs

Cut rules identify SKUs that should leave the assortment, move to a narrower distribution, become online-only, be replaced by a stronger substitute, or be exited after inventory sell-through. The best cut rules do not simply target the bottom of a sales ranking. They identify products whose contribution is too small relative to the demand, profit, space, capital, and operational attention they consume.

Cut weak performers: Weak SKUs show low sales, low unit velocity, weak margin dollars, poor growth, low repeat purchase, or low customer engagement within a comparable group. A low-volume premium item should not be judged against a mass item serving a different mission. However, if a SKU underperforms within its true peer set and has no strategic role, it should be challenged.

Cut redundant SKUs: Redundancy is one of the largest sources of hidden assortment waste. A redundant SKU serves nearly the same customer need as another item but adds little incremental demand. Redundancy often appears in excessive flavors, scents, colors, pack counts, minor size differences, near-identical brands, overlapping technical specifications, or multiple items at the same price tier. The question is not whether the SKU sells. It is whether the assortment would lose meaningful demand if the SKU were removed.

Cut low-rotation SKUs: Low turns can indicate poor productivity, excessive inventory, weak demand, poor pack configuration, or overdistribution. A low-rotation SKU is especially problematic when it also has high inventory value, large minimum order quantities, markdown risk, or limited customer importance. The action may be full removal, narrower distribution, lower inventory depth, smaller pack ordering, or transition to online-only.

Cut operationally costly SKUs: Some SKUs create disproportionate operational burden through special handling, fragile packaging, frequent returns, high damage rates, poor case-pack fit, content issues, supplier service problems, or high store labor needs. If customers value the SKU highly, the answer may be to fix execution. If customers do not value it, complexity is a strong reason to exit.

Cut legacy SKUs: Many assortments include products that remain because they have always been there. A legacy SKU may once have mattered but no longer fit current customer behavior, category strategy, brand positioning, or channel economics.

Cut decisions require demand-transfer thinking. Customers may transfer to another SKU, trade up, trade down, buy private label, buy online, delay the purchase, or leave the retailer. A cut decision is most attractive when the SKU has weak performance, high substitutability, low strategic value, and high complexity. It is risky when the SKU has low substitutability, strong loyalty, or category credibility value despite weak sales.

  • Performance: Is the SKU below threshold on sales, margin, turns, growth, or availability-adjusted velocity?
  • Differentiation: Does the SKU serve a distinct need, or is it duplicative within the decision tree?
  • Transfer: Is there a credible substitute that can capture most demand?
  • Complexity: Does the SKU create inventory, space, supplier, store, or digital burden?
  • Risk: Would removal harm loyalty, trip behavior, brand image, or category authority?

The strongest cut rules also include exit mechanics. A SKU recommended for removal still needs inventory disposition, markdown guidance, supplier communication, system updates, planogram changes, digital delisting, substitution mapping, and post-exit measurement.

5.3 Add Rules: Finding Gaps in Good-Better-Best, Price Points, Brands, Sizes, Flavors, and Use Cases

Assortment optimization is not only about reducing SKUs. In many categories, the best answer is to cut weak duplication and add where the range has genuine gaps. A retailer can be over-assorted overall and under-assorted in the places that matter most to customers. Add rules help the organization distinguish meaningful innovation from supplier-driven proliferation.

Add for price architecture: Many categories require a clear price ladder. Customers should understand the opening price point, mainstream value, better-quality option, and premium trade-up. If a category lacks a credible opening price point, the retailer may appear expensive. If it lacks a premium tier, it may miss margin and trade-up opportunity.

Add for good-better-best coverage: Good-better-best logic helps customers navigate quality, features, performance, and price. The “good” option should be credible, the “better” option should explain why it costs more, and the “best” option should offer meaningful benefits. Additions are justified when a tier is missing, weak, or not clearly represented.

Add for customer needs and use cases: Customer decision trees often reveal missing needs such as dietary requirements, sustainability claims, technical compatibility, local tastes, family size, single-serve convenience, professional-grade performance, gifting occasions, or project completion. The best ad decisions are grounded in observed demand, search behavior, lost sales, competitor assortment, store feedback, or loyalty insight.

Add for brand and private-label strategy: A category may need a national brand to establish credibility, a challenger brand to signal innovation, a local brand to build relevance, or a private-label item to improve margin and differentiation. However, brand additions should be controlled. Adding a new brand without clarifying which customer need it serves and which existing SKU it may replace is a common cause of assortment creep.

Add for size, flavor, color, or specification gaps: Variants create value when they correspond to real demand. A missing size that customers regularly request may justify an ad. A flavor that duplicates an existing need with little incremental demand may not. A technical specification required for compatibility may be essential even if volume is modest.

Add rules should include a one-in, one-out or one-in, one-review discipline where appropriate. This does not mean every addition must force an immediate removal, especially in growth categories. It does mean every addition should trigger a review of whether the new SKU deserves space, inventory, and management attention more than the weakest existing alternative.

  • Gap statement: the customer, price, brand, size, use-case, channel, or local gap addressed.
  • Evidence: the demand insight from search data, loyalty behavior, competitors, stores, lost sales, or suppliers.
  • Expected role: the intended impact on trips, baskets, margin, conversion, private label, or assortment architecture.
  • Trade-off: the existing SKU, space, module, or inventory allocation that may be reduced.
  • Success metric: the threshold for sales, margin, turns, repeat purchase, conversion, or customer response.

Add decisions should be held to a high standard because every addition creates future exit work. Supplier-funded launches are attractive at introduction, but they can leave long-term complexity after launch support ends.

5.4 Test-and-Learn Rules: Pilots, Control Groups, and Success Metrics

Not every assortment decision should be made with full confidence at the conference table. Some decisions require testing because customer response is uncertain, demand transfer is hard to estimate, supplier economics may change, or store execution may influence results. Test-and-learn rules give the organization a disciplined way to learn before scaling.

A test is most valuable when the decision is material and uncertainty is high. If the SKU is clearly redundant, weak, and low risk, a test may slow the organization unnecessarily. If the SKU is strategically important but economically uncertain, a test may prevent an expensive mistake. Tests are also useful for new brands, private-label substitutions, localized modules, premium tier expansion, online-only migration, assortment compression, and changes to pack architecture.

Pilot design: A pilot should define the assortment change, stores or channels involved, duration, operational steps, and expected impact. The pilot group should be large enough to produce reliable signals and representative enough to support rollout decisions. Pilots should avoid periods distorted by promotions, holidays, weather, supply disruptions, or competitor actions unless those events are part of the intended learning.

Control groups: A control group separates the effect of the assortment action from normal business movement. Control stores or customer groups should be similar to test stores in size, sales, customer mix, category role, seasonality, competition, and historical performance.

Success metrics: Each test should have clear success thresholds before launch. Metrics may include retained sales, sales transfer, gross margin dollars, margin rate, inventory reduction, turns, out-of-stocks, digital conversion, search conversion, attachment, complaints, returns, and store execution quality. For SKU cuts, the key question is how much demand is retained. For adds, the key question is whether the new SKU creates incremental value or merely cannibalizes existing products.

Decision timing: Tests must run long enough to observe behavior but not so long that the organization avoids decisions. The right duration depends on purchase frequency, seasonality, replenishment cycle, and sample size. The decision date should be agreed before launch.

Scale rules: Before a pilot begins, leaders should define what happens if the test wins, fails, or produces mixed results. A winning test may roll out nationally, to selected modules, or to additional clusters. A failed test may be reversed. A mixed test may lead to localization, revised pricing, packaging changes, supplier renegotiation, or another test.

  • Hypothesis: the specific belief being tested and why it matters.
  • Action: the SKUs added, removed, substituted, localized, expanded, or moved online-only.
  • Test design: the test group, control group, duration, timing, and operational requirements.
  • Success thresholds: the financial, customer, inventory, and execution metrics required for rollout.
  • Decision rule: the action to take after success, failure, or inconclusive results.

Test-and-learn discipline prevents two bad habits: rolling out uncertain changes too broadly and using “we need more data” as an excuse to avoid decisions. The purpose of testing is not endless analysis. It is faster, safer learning.

5.5 Decision Tree for SKU Actions

A SKU action decision tree converts the rules into a repeatable sequence. It helps teams move from facts to decisions without re-litigating every principle from the beginning. The decision tree should be simple enough for category teams to use, but robust enough to handle strategic exceptions.

The first question is whether the SKU fits the category role and customer decision tree. If it does not fit, it should be challenged even if it has some sales. Products that do not support the strategy can distract customers, consume space, and complicate execution.

The second question is whether performance is strong, average, or weak after normalizing for distribution, availability, seasonality, and lifecycle stage. Strong SKUs are candidates to keep or expand. Weak SKUs become cut or substitute candidates unless they serve a documented customer or strategic need.

The third question is whether complexity is justified. A SKU with moderate performance but high operational burden needs action: supplier improvement, pack change, reduced distribution, online-only migration, or removal. The final question is whether evidence is strong enough to act now or whether the decision should be tested.

  • Step 1: Does the SKU fit the category role and customer decision tree? If no, challenge for cut, substitution, or relocation. If yes, continue.
  • Step 2: Does the SKU perform strongly within its comparable group? If yes, keep or expand. If not, continue.
  • Step 3: Does the SKU serve a distinct customer, price, brand, size, use-case, local, or strategic need? If not, cut or substitute. If yes, continue.
  • Step 4: Is the SKU’s complexity justified by its value? If not, simplify, localize, move online-only, renegotiate, or cut. If yes, continue.
  • Step 5: Is the decision low risk and evidence-backed? If yes, implement. If not, run a test with clear controls and thresholds.

Each final action should have an owner, timing, and measurement plan. Keep: confirm the SKU’s role and performance expectations. Cut: define the exit path and demand-transfer assumptions. Add: define the gap, launch support, and success thresholds. Expand: increase distribution, space, visibility, or inventory based on proven demand. Localize: assign the SKU to specific clusters or channels with clear eligibility rules. Substitute: map the customer to a stronger alternative. Test: define the hypothesis, control group, duration, and decision rule.

Good rules create speed and consistency, but they do not eliminate accountability. Merchants still need judgment. Finance still needs to challenge economics. Stores and supply chains still need to validate execution. Digital teams still need to manage content, search, and conversion. The value of rules is that they let teams focus on the few decisions where trade-offs are real.

When cut, keep, add, and test rules are clear, assortment optimization becomes a repeatable management process rather than a one-time cleanup. The retailer protects what matters, removes what does not, fills gaps intentionally, and learns before scaling uncertain moves. That is the practical discipline behind a smarter, more productive assortment.

How to get started

1

arrow-down-blue

Tell us about your project

2

arrow-down-blue

Interview candidates

(We’ll provide bios within 48 hours on average)

3

Select your consultant and start work

Find a Consultant

or email us at: [email protected]