Operating Model Design: A Practical Guide for SMBs

Operating model design is the work of deciding how your business actually delivers what it sells: which processes run the same way everywhere, which are left to local judgment, who decides what, and which tools support it. A good design turns strategy into daily routines. A poor one leaves a smart strategy stuck behind handoffs, duplicate tools and unclear ownership. This guide shows what an operating model contains, the four basic choices, and a step-by-step way to design the target operating model for a company with 10 to 500 people, not 50,000.

RaremindMake Your Operating Model Work BetterRaremind maps how your business generates revenue to identify process inefficiencies and opportunities to improve operations without pushing unnecessary technology.Explore Raremind consulting

TL;DR

• An operating model describes how people, processes, technology and structure create and deliver the value your business model promises.

• You design one by choosing how much your units should be integrated and how much their processes should be standardized, then building around that choice.

• Start from how the business makes money, not from software. Technology comes last and only where it removes a real gap.

• Write down decision rights, review rhythm and a few outcome metrics, or the design stays a diagram.

• Aim for a short, high-level model you can read in one sitting, then refine it as the business changes.

Table of Contents

• What an operating model is, and what it is not

• The parts of an operating model

• Four operating models and how to choose

• How to design your target operating model

• Which framework should you use?

• An example: moving decisions to the work

• Decision rights: the part most designs skip

• Measuring whether the design works

• Our perspective

• Getting help with the design

• FAQ

• Sources

What an operating model is, and what it is not

A business model says what value you bring to market and to whom. An operating model says how you create and deliver it. Penn State’s Propel resource frames the difference this way: the business model is the value brought to market, while the operating model is how people, processes, technology and structure make that value real.

Andrew Campbell of the Operational Excellence Society describes an operating model as a visualization that explains how an organization operates to deliver value to customers or beneficiaries. The word that matters is visualization. If nobody can draw it on one page, nobody is running the company from it.

An operating model is also not an org chart, a process manual or a software architecture. Each of those is one part of it. A company can have a tidy org chart and still lose money because pricing approvals take a week, invoices go out late, or three tools hold three versions of the same customer record.

The parts of an operating model

Different authors cut the model into different pieces. Three useful cuts, from the simplest to the most detailed:

The Operating Model Canvas elements are often remembered as POLISM. TechTarget notes that the elements should be linked to strategy, success metrics and enabling technology, which is the step many teams skip.

For a smaller business, translate these into six plain questions:

1. Work: what are the five to eight processes that create revenue and deliver the product or service?

2. People: who owns each one, and how are teams organized around them?

3. Information: what data does each step need, and where does it live?

4. Partners: which suppliers, contractors or platforms does the work depend on?

5. Places: where is the work done, in which offices, regions or remote setups?

6. Management: how are decisions made, how often do leaders review results, and what numbers do they look at?

The parts only work as a system. A new sales process with no owner fails. A new owner with no data fails. A new dashboard with no review meeting fails.

Four operating models and how to choose

The most useful single idea in operating model design comes from MIT’s Center for Information Systems Research. It says an operating model is the level of business process integration and standardization the company needs to deliver its goods and services. Pick low or high on each axis and you get four models.

The same source describes Replication as units performing tasks the same way on the same systems, which produces efficiency and brand consistency while local managers keep the customer relationship.

Some quick, rough translations for smaller companies:

• A franchise-style set of service locations leans toward Replication.

• A holding company with unrelated brands leans toward Diversification.

• A firm where sales, delivery and finance all touch the same customers and must share one record leans toward Coordination or Unification.

Two cautions from the same research matter for any owner. Unification takes substantial time, money and management attention, so it is not a free upgrade. And different levels of an organization may need different models: your finance back office may be standardized hard while your client-facing teams keep room to adapt.

How to design your target operating model

A target operating model is simply the operating model you want to be running in 12 to 24 months. Here is a sequence that works without a large consulting team.

Step 1: Start with how you make money

Before drawing anything, trace the money. Where does revenue come in, what does it cost to earn it, and where does it leak? Typical leaks in small and mid-sized firms are quotes that never get followed up, unbilled work, discounts nobody approved, renewals that lapse quietly, and manual rework between systems. This is the order we use with clients, and it keeps the design anchored in profit rather than in an org diagram.

Step 2: Declare what makes you distinct

MIT CISR’s advice is to declare which processes distinguish the company, then choose a viable model and build reusable capabilities around it. If your edge is fast custom quoting, that process gets your best people and tools. Everything else can be made boring and standard.

Step 3: Map the current state

Draw the real workflow, not the one in the handbook. For each of your five to eight core processes, note the owner, the steps, the tools, the handoffs and the wait times. Ask the people doing the work. Mark every spot where someone retypes data, waits for approval, or keeps a private spreadsheet.

Step 4: Choose your integration and standardization level

Use the grid above. Decide, process by process, whether it should be the same everywhere (standardize), shared across teams (integrate), both, or neither. Write one sentence per process explaining why. If you cannot explain it, you have not chosen it.

Step 5: Set design principles

Translate strategy into a handful of simple rules that can settle arguments later, for example: “one customer record”, “approvals sit with the person closest to the customer up to a set limit”, “we buy a tool only after the process works on paper”. Principles keep the design coherent when you are not in the room.

Step 6: Design the target state

For each core process, draw the future version on one page: owner, steps, data, tools, decision points, metrics. Keep the whole model short. The Operating Model Canvas is built for high-level models of one to ten pages, not hundred-page documentation, and that is the right scale for a small business.

Step 7: Plan the transition

List the gaps between current and target, rank them by money recovered or saved against effort, and sequence them. Start with changes that need no new software: clear an owner, remove an approval, merge two reports. Add automation only where a manual step remains after the process has been simplified.

Step 8: Install the rhythm

Set a weekly operating review and a monthly results review with a fixed agenda and the same few numbers. This is the management-system part of the model, and it is what keeps the design alive after the workshop energy fades.

Which framework should you use?

You do not need to pick one religion. Each lens answers a different question:

• Operating Model Canvas (POLISM): what are the building blocks and how do they fit on a page?

• MIT CISR four models: how integrated and standardized should we be?

• Business model canvas and value-chain analysis: where does value come from, and which internal activities create it?

• Componentized design: how do we split into accountable teams without losing alignment? MIT CISR says this works by balancing the autonomy of empowered teams with alignment through strategic missions, clear accountability and shared objectives.

For most SMBs, a sensible combination is the four-model grid to choose a direction, the canvas to document it, and a short list of metrics to run it.

An example: moving decisions to the work

Large-company examples are useful mainly for the choices they show. MIT CISR’s case on Allstate reports that its Claims organization moved operational decision rights to durable cross-functional teams focused on strategic objectives and continuous solving of customer problems. The lesson scales down. A 30-person firm can do the same thing in miniature: replace a chain of handoffs with one small team that owns a customer outcome from request to cash, with authority to decide within agreed limits.

Decision rights: the part most designs skip

Most operating model problems are decision problems. Who can approve a discount? Who decides to take on a job, and who can say no? When those answers are vague, work waits, or gets done inconsistently. Even a simple written rule helps. A bid/no-bid scorecard for small businesses is a good illustration: a few weighted criteria let a team decide quickly whether an opportunity deserves effort, without escalating every case.

For each recurring decision, record four things: who decides, who must be consulted, what limit applies, and how fast the answer is due.

Measuring whether the design works

Track outcomes, not activity. A useful starter set for a small business:

• Time from customer request to cash collected

• Share of revenue that leaks (unbilled, discounted without approval, lapsed renewals)

• Cost to serve per customer or per job

• Rework rate between steps

• Number of tools per core process

The Harvard Data Science Review’s 2026 playbook on redesigning work with AI recommends auditing workflows, gauging them against outcomes, engineering redesigned flows and tracking value through outcome metrics. Its perspective is that redesign should remove unnecessary steps rather than simply automate the existing ones. That matches what we see: automating a bad process gives you a faster bad process.

Common pitfalls

• Designing around the org chart. Start with the flow of money and work, then fit the structure.

• Buying tools first. Software picked before the process is settled usually locks in the old mess.

• Skipping decision rights. The diagram looks finished, but nothing moves faster.

• Over-documenting. A 60-page model that nobody reads is worse than a 4-page one that people use.

• Treating it as one-off. Revisit the model when you add a product line, a location or a major customer segment.

Our perspective

We come to operating model design from the revenue side. Most owners who call us are not short of software; they have too much of it, bought one problem at a time, and no clear picture of where money is lost. So we begin by mapping how the business earns, which is what our Money Map does, and only then ask whether a tool, an automation or just a changed habit closes the gap.

We do not sell software and we take no vendor commissions, so the answer is sometimes “you already own what you need” or “remove this tool”. Our method is standard, but the recommendations are not: a ten-person services firm and a fintech scale-up need different integration and standardization choices, and a template pretending otherwise does harm.

If you remember one thing: design the operating model so that your business becomes more profitable without turning into a technology company. Technology is a supplier to the model, not its author.

— Mirko

Getting help with the design

If you would rather not carry this alone, our Your Digital & AI Office plan gives you ongoing independent advice for €995 per month: we map how your business makes money, find the leaks, and guide the changes. Bigger, hands-on initiatives are scoped separately as Larger Projects. For background on choosing an adviser, see our piece on strategic advisory versus traditional consulting, and on where cost savings usually hide, our cost reduction strategies for small businesses. If AI is part of your plan, read when to use AI in business strategy first.

FAQ

What are the four operating models?

MIT CISR names them Diversification, Coordination, Replication and Unification. They come from combining low or high process integration with low or high standardization. Which one fits depends on whether your units share customers and data, and whether they should run the same way.

What is the difference between a target operating model and an operating model?

An operating model is how the business runs. A target operating model is the version you intend to run after a planned change, usually 12 to 24 months out. Designing it means mapping the current state, choosing a direction and listing the gaps to close.

How long should an operating model document be?

Short. The Operating Model Canvas is meant for high-level models of one to ten pages. For an SMB, a few pages that show core processes, owners, decision rights and metrics are enough, as long as people use them in weekly management.

How often should we redesign our operating model?

Revisit it whenever strategy changes materially: a new product line, a new market, an acquisition, or sustained margin pressure. Otherwise, a yearly review of the metrics and decision rights keeps the design current without constant restructuring.

Do we need new technology to change our operating model?

Usually not at first. Many gains come from clarifying ownership, removing approvals and merging duplicate reports. Add automation or AI after the process works, and only where a manual step still costs you time or money.

Sources

• Propel Business, Operating Models (Penn State)

• MIT CISR, Operating Models (2005)

• Operational Excellence Society, What is an operating model?

• Operating Model Canvas, chapter 1

• Operating Model Canvas

• MIT CISR, The Digital Operating Model: Building a Componentized Organization

• TechTarget, What is an operating model?

• MIT CISR, Allstate’s Digital Operating Model

• Harvard Data Science Review, The Agent-Centric Enterprise

• Bid/No-Bid Criteria: A Scorecard for Small Business

Recommended

• 17 Cost Reduction Strategies for Small Businesses: Operations First

• AI for Business Strategy: When to Use It (and When You Shouldn’t)

• AI Readiness Assessment: Map Revenue Workflows Before Buying Tools