How to Fill In a Business Model Canvas (the Right Order)

Print out a Business Model Canvas and hand it to a team, and almost everyone does the same thing: they start at the top-left box, Key Partnerships, and work across the page like reading a book. It feels natural. It is also the single most reliable way to produce a canvas that looks complete and means nothing.

This guide walks through the order professionals actually use, block by block, with one worked example running through all nine steps, the consistency checks reviewers run before calling a canvas done, and what to do with it afterward. If you need the basics of what each block means first, start with what the Business Model Canvas is and come back.

The mistake almost everyone makes

The canvas is laid out with infrastructure on the left and customers on the right, so a left-to-right fill starts with what you have: your team, your technology, your suppliers, your assets. The reasoning slides in unnoticed: “We have a factory, a skill, a codebase. What can we sell with it?”

That is designing the engine before deciding where to drive.

The canvas has a deliberate geometry. The right side (Customer Segments, Value Propositions, Channels, Customer Relationships, Revenue Streams) is the front stage: market-facing, about whether anyone wants this. The left side (Key Activities, Key Resources, Key Partnerships, Cost Structure) is the back stage: operational, about whether you can deliver it. Value Propositions sit in the center as the hinge between the two.

Start on the left and you optimize the back stage for a front stage you have not defined yet. Every downstream block inherits the error: you pick channels that suit your operations rather than your customers, price what you built rather than what they value, and end up with a plausible-looking page that customers never asked for.

There is one legitimate exception. Established companies exploring how to leverage an existing asset (a distribution network, a dataset, a licensed technology) sometimes start from a resource on purpose. Even then, experienced facilitators force the team over to the right side within minutes, because an asset without a customer is a cost.

The professional consensus is a different sequence entirely: right side first, then left side, then the bottom row. Value before infrastructure before finances.

The right order: 9 steps

To keep this concrete, every step follows the same fictional company: Ledgerly, a B2B SaaS that automates month-end close for small accounting firms. One business, nine blocks, in order.

Step 1: Customer Segments

Key question: for whom are we creating value, and who are our most important customers?

Name specific groups, not categories. “SMEs” and “businesses” are labels, not segments. A usable segment is specific enough that you could find a real person in it, call them tomorrow, and interview them. Separate groups whenever they have distinct needs, are reached through different channels, or pay for different things. Star the segment the model is optimized for, and note who the user is versus who pays, because in B2B they are often different people.

Ledgerly: not “accountants.” The primary segment is owner-led bookkeeping practices with 3 to 15 staff that still run month-end close in Excel. A second note marks the buying committee inside that segment: the firm owner pays, the senior bookkeeper uses the product daily. The owner-led practice gets the star.

There is a full breakdown of segment types and mistakes in the Customer Segments guide.

Step 2: Value Propositions

Key question: which of our customer’s problems are we solving, and why do they choose us over the alternative?

State value in the customer’s language, not yours. “AI-powered dashboard” is a feature. “Close the monthly books in 2 hours instead of 2 days” is value. Write one value proposition per segment, and run the completion test: “Our customers choose us because ______.” If a competitor could paste your sentence onto their canvas, it is not a value proposition.

Do not fill this block once and move on. Segments and value propositions are a pair. Facilitators bounce between Steps 1 and 2, tightening both until they click, because this pair is the heart of the whole model. Everything else on the canvas exists to deliver and monetize it.

Ledgerly: for the firm owner, “take on 30% more clients without hiring, because close time per client drops from 2 days to 2 hours.” For the senior bookkeeper, “stop rebuilding the same reconciliation spreadsheet every month.” Same product, two propositions, each pointed at a named segment. The current alternative is named too: Excel plus overtime, which costs nothing in software and a great deal in hours.

The Value Propositions guide covers Osterwalder’s full checklist of value types.

Step 3: Channels

Key question: how does each segment discover us, evaluate us, buy, receive the product, and get help?

Channels are the entire customer journey across five phases: awareness, evaluation, purchase, delivery, and after-sales. The classic error is listing only marketing (“Instagram, SEO, ads”) and forgetting how the product is delivered and supported. Distinguish channels that work today from channels you hope will work, and mark the hopes as assumptions.

Ledgerly: awareness through bookkeeping association newsletters and a podcast the owners actually listen to. Evaluation through a 14-day trial on the firm’s own client data. Purchase through self-serve checkout. Delivery is the cloud app plus a guided data migration. After-sales is in-app chat with a 4-hour response promise. “Partnership with a major accounting platform’s app marketplace” goes on the canvas as a pink hypothesis note, because no conversation has happened yet.

Step 4: Customer Relationships

Key question: what type of relationship does each segment expect, and what does it cost us to provide?

Pick from a real taxonomy: personal assistance, dedicated account management, self-service, automated services, communities, co-creation. “Good customer service” means nothing. The test is fit: high-touch relationships for high-value accounts, self-serve for the long tail. Dedicated account managers for $10-per-month customers is as broken as pure self-service for enterprise buyers who expect a human.

Ledgerly: automated onboarding with email sequences for the bookkeeper user, plus one 45-minute setup call per firm for the owner, because migration is the moment firms churn. No dedicated account managers below 10 seats. The model is optimized for retention over acquisition, since the firm’s client data compounds inside the product.

Step 5: Revenue Streams

Key question: for what value is each segment actually willing to pay, and how?

Choose a mechanism, not just a number: asset sale, usage fee, subscription, licensing, brokerage, advertising. Then a pricing approach: fixed list price, volume-dependent, negotiated. Note whether each stream is one-time or recurring, and roughly what share of total revenue it contributes. The single riskiest untested assumption on most canvases lives here: “customers will pay $X per month” written without evidence.

Ledgerly: subscription at $40 per seat per month, plus a one-time $500 migration fee that offsets onboarding cost. Estimated split: 85% recurring subscription, 15% migration fees. The willingness-to-pay number is flagged pink until real firms have pre-ordered.

Notice what just happened: five blocks in, and nothing yet about servers, staff, or suppliers. That is the point. You now know what the infrastructure must deliver. See how real companies structure this in the 10 worked canvas examples.

Step 6: Key Resources

Key question: what assets does delivering all of the above actually require?

Now, and only now, the left side. List 3 to 6 non-negotiable assets across four categories: physical, intellectual (brand, data, proprietary knowledge, licenses), human, and financial. Apply the kill test to every candidate: what happens if we lose this tomorrow? If the answer is “we’re fine,” it is not key. Watch for intangibles; data, community, and specific people are frequently the real moat and frequently omitted.

Ledgerly: the reconciliation engine and its codebase, the growing dataset of categorized transactions that improves accuracy for every firm, the two engineers who built the bank-feed integrations, and 12 months of runway. Laptops and the office do not make the list.

Step 7: Key Activities

Key question: what must we do exceptionally well that competitors cannot easily match?

Three to five activities, typed as production, problem solving, or platform management. Routine operations (accounting, HR, email) are necessary but not key. Keep activities distinct from resources: “our engineers” is a resource; “maintaining 99.9% accurate bank-feed integrations” is an activity. And check the mirror: if your value proposition promises 2-hour closes, the activity that makes closes fast had better appear here.

Ledgerly: continuous development of the reconciliation engine, maintaining bank-feed integrations, and running the guided migrations that Step 4 promised. Three items, each traceable to something on the right side.

Step 8: Key Partnerships

Key question: what should we not build or own ourselves?

Only partners whose loss would materially damage the model. The electricity provider and the accountant are vendors, not key partners. Classify the motivation: cost optimization, risk reduction, or access to resources and capabilities you will not build in-house. Then name the dependency risk out loud: what happens if this partner triples prices or cuts you off?

Ledgerly: the bank-data aggregator that supplies transaction feeds (access to a capability Ledgerly should not build), and cloud hosting (cost optimization). The aggregator is also flagged as the model’s biggest dependency risk: if it changes terms, the core value proposition wobbles. The hoped-for accounting-platform marketplace deal stays a hypothesis, not a partnership.

Step 9: Cost Structure

Key question: what do the resources, activities, and partnerships we just listed actually cost?

This block is deliberately last, because it is calculable only after Steps 6 to 8 exist. List the 3 to 5 dominant cost drivers, not 30 line items of bookkeeping. Classify the model as cost-driven or value-driven and check coherence: a premium white-glove value proposition paired with a cheapest-possible cost obsession is an incoherent model. Then run the back-of-envelope viability check: at what point does revenue exceed cost?

Ledgerly: engineering salaries (largest, fixed), the bank-data aggregator’s per-connection fees (variable, scales with customers), and migration labor. Value-driven, which matches a proposition built on accuracy and time saved rather than on being the cheapest tool. Roughly 120 firms at current pricing covers the fixed base. That is a napkin number, not a financial model; the canvas and the spreadsheet are complements, not substitutes.

The 6 consistency checks before you call it done

A canvas is a system, not nine independent lists. The most cited whole-canvas failure is filling blocks in isolation, which produces a page that reads well and collapses under the first sharp question. Before you call the draft done, run these six checks. If you cannot draw the arrow, delete the note.

1. Every value proposition points at a named customer segment. Color-coding sticky notes per segment makes orphans visible instantly.
2. Every channel and every relationship serves a specific segment. A channel nobody in your segments uses is decoration.
3. Every revenue stream comes from a specific segment paying for a specific value proposition. “Revenue: subscriptions” attributed to no one is a red flag.
4. Every key resource, activity, and partner supports something on the right side. Orphan entries on the left signal padding, or worse, waste you are already paying for.
5. Every major cost traces to a key resource, activity, or partnership. Free-floating costs mean the bottom row was filled independently of everything above it.
6. Change propagation holds. Pick one block and imagine changing it. Swap Ledgerly’s segment from small firms to enterprise accounting departments, and the channels, relationships, pricing mechanism, and support model all have to shift. If changing one block would not ripple through the others on your canvas, the blocks were never actually connected.

How long should this take?

Shorter than you think. A solo first draft takes about 30 minutes. A good team first pass is a 2 to 3 hour workshop: under 90 minutes is rushed, over 4 hours is overthinking. Spend 10 to 15 minutes per block with a visible timer, then move on; you can always come back, and the Segments-Value Propositions pair usually earns a second loop.

Mechanics that consistently help: 4 to 7 cross-functional people, laptops banned, the canvas printed poster-size on a wall, and one idea per sticky note written in thick marker (if it does not fit on the note, the thought is not sharp enough). Sticky notes signal that everything on the page is a hypothesis that can be moved or thrown away, where typed documents get treated as decisions. Many teams color-code: yellow for validated facts, pink for assumptions, green for new ideas to test. For an existing business, map the current state first and put the future state on a separate canvas.

Any free Business Model Canvas template works for the draft; remote teams run the identical protocol on a shared digital canvas.

After the canvas: rank your riskiest assumptions

Do not end the session with a “finished” canvas. End it with ranked risks. Every note on the page is an assumption, and fill order matters less than validation order: as Ash Maurya puts it, draft fast, then stop debating the boxes and start testing the claims that would kill the model if wrong.

In practice: plot your assumptions on an importance-versus-evidence grid and attack the important-but-unevidenced quadrant first (Strategyzer calls this Assumptions Mapping). The killers usually hide in Customer Segments, Value Propositions, and Revenue Streams, which is why desirability gets tested before feasibility. Then design the cheapest experiment that could prove each one wrong: customer interviews, a landing page test, pre-orders or letters of intent, a concierge MVP, a pricing test.

Bring the evidence back and update the page: pink assumption notes become yellow fact notes, or the affected blocks pivot. Revisit at minimum quarterly, or after every experiment sprint. For Ledgerly, the first test is obvious: will 10 firm owners pre-pay the $500 migration fee before the migration tooling exists? That answer is worth more than any amount of canvas polishing.

If you are pre-product with zero customer evidence, consider whether the Lean Canvas fits better for this stage; and when the draft holds up, the Complete Toolkit adds the revenue and break-even models the canvas deliberately leaves out.

Common questions

Can I fill it in alone? Yes, and a 30-minute solo draft is a fine starting point. But a solo canvas encodes one person’s blind spots, so treat it as a draft to challenge in a team session, not a conclusion.

What if I have multiple customer segments? Color-code notes per segment on one canvas, and once segments diverge seriously (different channels, pricing, or economics), give each its own canvas plus an optional consolidated view. Averaging five segments into one canvas serves none of them.

Do I need a business plan too? Usually not at this stage. The canvas replaces the plan for modeling and testing; a full plan matters mainly when a bank or institutional investor demands one. The canvas versus business plan comparison covers when each earns its keep.

What does the canvas not cover? Competition, the external environment, and mission or vision. It is also the wrong tool for deep value-proposition work (zoom into a dedicated worksheet for that) and for financial planning. Layer those separately rather than cramming them into the nine boxes.