Count the things you sell. Then count the things your team has to produce, price, quote, onboard, or deliver differently. If the second number is noticeably bigger than the first, you already have your answer about when to simplify a product or service portfolio. The honest answer is usually: before the next launch, not after the next bad quarter.
That gap between the two counts is the whole subject. Customers experience the first number. Your cost structure experiences the second. A business can look focused on its website and be quietly chaotic behind it: nine packages on the pricing page, but twenty-three ways a job actually gets delivered once you include legacy clients, inherited contracts, one-off discounts, and the custom build someone promised in 2023.
Variety is what customers pay for. Variance is what you pay for.
This distinction matters more than any list of symptoms, so it is worth sitting with.
Variety is a difference a customer can see, understand, and choose between. A larger size. A faster delivery window. A tier with a dedicated account manager. Customers will pay for those, and removing them removes revenue.
Variance is a difference in how the work gets done that the customer never asked for and often cannot detect. Two packages that look identical to a buyer but run through different approval steps. Three products built from four almost-identical components because nobody standardized. A service tier that exists because an enterprise prospect wanted one clause changed, and the clause became permanent.
McKinsey makes this point about consumer products: simplification is not only about culling SKUs. It can mean standardizing how things are designed and made while preserving the distinctions customers actually value and will pay for. That reframing applies just as cleanly to a consultancy, a managed service provider, or a subscription app. You are not trying to offer less. You are trying to stop paying for differences nobody buys.
So here is the first practical test. Do the two-count audit. Count the offers as a customer sees them. Then count the delivery paths, pricing exceptions, contract variants, and configurations as your operations team experiences them. The ratio tells you whether you have a portfolio problem or an architecture problem, and they do not get fixed the same way.
The signals that it is time
Nobody simplifies on a calendar. They simplify when something starts hurting. The trick is recognizing the pain early, because the usual triggers are lagging indicators.
Your operations start failing in patterned ways
Complexity shows up in quality metrics before it shows up in the P&L. A manufacturing case study published in a 2026 SEC exhibit, covering Middleby’s Pitco business, attributed 82% of first-pass-yield misses to complexity. The response was not simply discontinuing slow sellers. It included limiting configurators to set package options, and the programme reported roughly $9 million in inventory reduction after nine months.
Read that as a general warning. If your errors cluster around specific variants, if rework spikes on custom configurations, if onboarding runs long for one tier in particular, that is complexity presenting as an operations problem. A service business sees the same thing in different clothing: scope disputes, missed handoffs, proposals that take four days because somebody has to reconstruct what was promised to a similar client last year.
Buyers ask you which one they need
If prospects regularly ask a salesperson or a chat agent to explain the difference between two of your options, the options are not differentiated. They are confusing. Stripe’s guidance on self-serve pricing is blunt about this: more than three or four plan tiers can create decision paralysis. Its own Checkout pricing table caps out at four products with up to three prices each, which is a quiet signal about where sensible design stops.
There is a related piece of evidence worth sitting with, because it complicates the simple “less choice is better” line. Stripe reports that showing only one payment method that is not geographically relevant can cut conversion by up to 15%, while dynamically surfacing at least one additional relevant method beyond cards is associated with an average 12% increase in revenue and a 7.4% increase in conversion. Fewer options is not the goal. Fewer irrelevant options is.
That is the real rule. Relevance beats abundance, and relevance beats minimalism too.
Revenue grows and margin does not follow
This is the pattern that fools owners for years. You add an offer, revenue rises, and the cost of carrying it disappears into shared overhead where no line item will ever reveal it. Setup time, support tickets, training, inventory, extra pages on the website, an extra path through the billing system. None of it gets charged to the offer that caused it.
If your revenue line and your margin line have been diverging for a few quarters and nothing obvious explains it, assume complexity is absorbing the difference until proven otherwise.
Your lineup was inherited rather than chosen
Ask where each offer came from. Acquisitions, legacy plans you stopped selling but never closed, custom one-off deals that became standard, and internal requests from a team that wanted something to sell this quarter. Research published in Industrial Management & Data Systems in 2022 makes the point that product deletion decisions are neglected in both research and practice, and that deletion reasons span strategic, operational, and customer-related factors. In plain terms: companies build elaborate launch processes and almost never build a retirement process.
A portfolio that has only ever grown is not a strategy. It is sediment.
The rules change underneath your plan structure
This is the trigger almost no article on portfolio simplification mentions, and it now matters a great deal for anything recurring. In October 2024 the FTC announced revisions to its Negative Option Rule, covering automatic renewals, free trials, and other negative option programmes, including business-to-business transactions. Some provisions were set to take effect within 60 days of Federal Register publication and most within 180 days. The rule requires disclosing material terms before sign-up and providing a cancellation method as easy as the sign-up method.
Plan sprawl used to be a conversion and UX problem. For subscription businesses it is now also a compliance surface. Every tier, trial variant, add-on, and grandfathered price is another set of material terms to disclose accurately and another path that has to be cancellable without friction. Eleven plans means eleven chances to get the disclosure wrong. Four means four.
Why cutting the slow sellers first is usually the wrong move
The standard advice is to rank everything by revenue, chop the bottom, and congratulate yourself. I think that is the least useful version of this work, and the evidence leans the same way.
Bain’s work on focused product portfolios includes a supermarket case in which total SKUs fell by 40% while overhead dropped, inventory days fell by 60%, and revenue rose 25%. It also reports businesses that cut SKUs by 50% or more with no damage to profitability or competitive position. Those are striking numbers, and they are tempting to read as “cutting works.” But the same body of work warns against simply whacking the tail, and stresses assessing what a given item is worth to your target customers rather than to a standalone P&L. Streamlining on its own does not improve performance unless the underlying cost structure and operating model change with it.
Cutting slow sellers removes revenue immediately and removes cost slowly, because the cost lives in shared capacity you still pay for. Standardizing the back end removes cost permanently and may remove no revenue at all.
So treat the low-seller list as a diagnostic, not a hit list. The interesting question is not “what sells least?” It is “what forces us to work differently for the least return?” Those are frequently different items. A low-volume product that runs on the same rails as everything else is cheap to keep. A mid-volume product that requires its own process, its own supplier, its own onboarding script, and its own support training may be the most expensive thing you own.
There is a related trap worth naming. One genuinely low-profit offer can earn its place by anchoring a strategic account, satisfying a channel requirement, or giving a buyer a reason to take you seriously. That is a legitimate reason to keep something, but only if you can say which account, which channel, and what happens if it disappears. “It’s strategic” without a name attached is usually just attachment.
What to measure before you cut anything
Run the numbers in this order, because each one changes how you read the next.
- Contribution margin after delivery cost, with real hours or real changeover time loaded in, not an allocation percentage.
- Operational burden per unit sold. Support tickets, rework rate, onboarding hours, exceptions raised, average quote turnaround.
- Overlap. Which offers compete for the same buyer with the same job to be done? Overlapping offers do not add demand; they split it and complicate the sale.
- Shared infrastructure. Does this offer reuse existing delivery rails, or does it have its own? This is the single most underrated number in the whole exercise.
- Customer concentration. Who buys it, how much of their spend is it, and would losing it put other revenue at risk?
- Exception count. How many customers are on a non-standard price, term, or scope for this offer? A high exception rate means the standard version is not solving the problem.
Then sort into four buckets rather than two: keep as is, keep but standardize the back end, merge into another offer, retire. The middle two buckets are where the quiet money usually sits, and they are the buckets a simple cut-list approach never generates. If you are unsure which constraint the portfolio is actually straining, it is worth separately identifying the real constraint limiting growth before deciding what to remove.
SKU rationalization, menu simplification, plan simplification: the same logic, different surfaces
These terms get used as if they were separate disciplines. They are the same decision applied to different parts of the business. SKU rationalization works on physical items and inventory. Menu simplification works on items, modifiers, and preparation steps. Plan simplification works on tiers, add-ons, trials, and billing logic. Portfolio strategy sits above all three and asks which categories you should be in at all.
The reason to keep them straight is sequencing. You can simplify plans in a sprint. You cannot simplify a manufacturing portfolio in a sprint, because component standardization has lead times measured in quarters.
Simplifying without losing the customers you want
The fear here is rational. Cutting badly creates churn, and churn in your best accounts costs far more than the complexity you removed.
Three principles make the difference.
Every retirement needs a named successor. Not “we are discontinuing this,” but “you are moving to this, it covers the same need, and here is what changes.” A migration is a redirect, not a deletion. If you cannot name the successor, you are not simplifying, you are abandoning a segment, which is a different decision and should be made consciously.
Protect price for a defined period, and say when it ends. Open-ended grandfathering is how plan sprawl is reborn. A date-bounded transition removes the offer from your operating reality rather than hiding it in a legacy column forever.
Consider removing the exception instead of the offer. Starbucks did a version of this. From 7 November 2024, company-operated stores in the US and Canada stopped charging extra for non-dairy milk, which the company said was the second most requested customization after an extra espresso shot. Starbucks noted that almost half of US customers who paid to modify a beverage at those stores would see a price reduction of more than 10%. The drinks did not go away. The pricing exception did, and with it a layer of decision friction at the counter and a layer of complexity at the register.
That is often the highest-return move available in a service business: keep the thing customers want, delete the surcharge, the carve-out, or the conditional rule that made it awkward to buy and awkward to deliver.
There is also an offensive version of this. In July 2024 McDonald’s framed its US focus around value, core menu, and digital execution, including a $5 Meal Deal. Under pressure, a large chain chose to push a clear hero offer rather than widen the assortment. That instinct is worth copying. When demand softens, the reflex to launch something new is usually the wrong one, and learning to say no is what protects the offers already working.
The one-page test
Before committing to cuts, write the business on one page. The US Small Business Administration notes that a lean startup plan is typically a single page and can take as little as an hour, and that the format is useful for visualizing tradeoffs and the fundamental facts about a company. That is exactly what you need here, and almost nobody uses it this way.
Force the whole portfolio onto that page alongside your target market, value proposition, key activities, and cost structure. The constraint does the work. When everything has to fit, you quickly see which offers serve the same customer with the same promise, which ones require activities that appear nowhere else in the business, and which ones you kept because removing them felt like a loss rather than because they earned space.
Pair it with a short market analysis in the SBA’s sense: industry outlook, target market, competitor research, and the trends shaping both. Simplification decisions made without that context tend to cut whatever is currently inconvenient rather than whatever is strategically finished. For a fuller version of this exercise, it fits neatly into a simple annual planning process.
Build the retirement process you never built
The structural fix is the one almost no business makes, and it is the reason this problem recurs every three years.
You have a launch process. Someone proposes an offer, it gets reviewed, resourced, priced, and shipped. You almost certainly do not have the mirror image: a standing process for retiring offers, with an owner, a cadence, and criteria.
Two changes fix it.
First, attach a sunset condition to every new offer at launch. Not a date necessarily, but a threshold: if it has not reached a defined revenue or adoption level by a defined review point, it gets merged or retired by default. Default matters. Without it, inertia always wins, because killing something requires a decision while keeping it requires nothing. This belongs in the same gate you use when deciding whether a new opportunity fits the business.
Second, review the portfolio on a fixed rhythm rather than in a crisis. Quarterly is enough for plans, packages, and services. Annually is usually enough for anything with long manufacturing lead times, with a quarterly check on exceptions and custom deals, which is where sprawl enters fastest. It sits comfortably alongside the other strategic questions worth reviewing each quarter.
Bain’s research suggests low-complexity companies have grown substantially faster than high-complexity ones, though that brief is now some years old and is better read as a direction than a benchmark. The mechanism is what persuades me rather than the multiple. Simpler businesses launch faster, train faster, quote faster, explain themselves more clearly, and spend less management attention maintaining things nobody is excited about. Those advantages compound, and they are almost impossible to buy back once a portfolio has been allowed to sprawl for a decade.
The question to leave with is not “what should we cut?” It is “what would we build if we were starting today with the customers we actually want?” Then compare that to what you have. The difference is your work list, and it will rarely be just the slow sellers.
