
Small Business Strategy: 12 Owner Questions
On this page
- Do I need a written business strategy?
- What is the difference between a strategy, a plan and goals?
- How far ahead should a small business plan when tools change every quarter?
- Should I compete on price or on something else?
- How do I decide what to stop doing?
- How do I make bets when I can't predict what happens?
- How can I use scenario planning without building a forecasting department?
- How often should I revisit the strategy?
- How do I involve a small team without turning it into a committee?
- How do I tell whether the strategy is working?
- Should AI be the strategy?
- What does a one-page strategy look like?
Business
A small business strategy is a short written answer to one question: what is the main obstacle in front of us, and how will we get past it? In Richard Rumelt's terms, it has three parts: a diagnosis, a guiding policy and a set of coherent actions. Our one-page version adds a list of things the policy rules out.
The answer can fit on one page. Goals, budgets and tools can support it, but they are not part of that three-part definition. Sources cited here were read on 30 October 2026.
Do I need a written business strategy?
Yes. Our recommendation is to write the strategy on one page so employees can use it to test decisions.
The SBA's business plan formats show that a useful written document need not be long. Traditional business plans can run dozens of pages and are commonly requested by lenders and investors. Lean plans can take as little as one hour to make and are typically one page.
Our distinction is that a business plan describes the company to a reader, such as a lender, while a strategy states the challenge, the policy for dealing with it and the actions that carry out that policy. You may need both documents. A lender's questions about financing do not disappear because your strategy fits on a page.
Writing the strategy also makes disagreement visible. If a proposed service conflicts with the guiding policy, the team can discuss the conflict before spending money on it.
Check: Give the page and a recent proposal to someone who did not write either. Ask them to identify the diagnosis and say whether the proposal follows the policy.
What is the difference between a strategy, a plan and goals?
A strategy explains how you will deal with a challenge; a plan schedules actions; a goal states a wanted result. Richard Rumelt's Good Strategy Bad Strategy describes the strategy's kernel as a diagnosis, a guiding policy and coherent actions.
The diagnosis defines or explains the challenge. The guiding policy sets an approach to that challenge and channels choices in some directions while ruling others out. Coherent actions carry the policy into practice and support one another.
“Grow revenue 20%” is a goal. A hypothetical kernel is: “Our service contracts are under-sold because quotes arrive slowly; focus on commercial maintenance in one metro; introduce contract tiers, reserve technician capacity and call customers before renewal.” A plan would assign dates and owners to those actions.
Rumelt's 2011 McKinsey Quarterly article, The perils of bad strategy, names four hallmarks of bad strategy: fluff, failure to face the challenge, mistaking goals for strategy and bad strategic objectives.

Notice the two paths from the guiding policy: it directs the actions and sends incompatible choices into the ruled-out bin.
A goal list, a budget or a tool rollout is not a strategy. Ask: what obstacle does this answer, and what does it rule out?
Check: Underline the challenge, circle the guiding policy and mark the actions on your page. If one is missing, fill that gap before adding another target.
How far ahead should a small business plan when tools change every quarter?
Our rule is to review the guiding policy yearly or when the diagnosis changes, plan coherent actions about 90 days ahead and treat each tool trial as reversible. This gives the company a stable direction while leaving room to change a tool that fails.
Pierre Wack observed in Scenarios: Uncharted Waters Ahead that forecasts had worked reasonably well in the relatively stable 1950s and 1960s. From the early 1970s, forecasting errors became more frequent and were sometimes dramatic and unprecedented in magnitude.
A 90-day action plan can specify the work, owner and observable result without pretending you know which software will be best several years from now. Our suggested tool trial has a defined end date, a person responsible for evaluating it and a written cost to undo the change. A trial stops being easy to reverse if it creates a long contract or moves essential records into a format you cannot recover.
Check: Look at the next quarter's actions. For each tool trial, find the end date and the written exit cost; add them before the trial starts if either is missing.
Should I compete on price or on something else?
Choose a position you can support economically; a low price alone does not establish a cost advantage. In Competitive Strategy, Michael Porter identifies overall cost leadership, differentiation and focus as generic strategies.
Cost leadership means holding the lowest-cost position in the industry, which is different from charging the lowest price. Differentiation means offering something customers value as distinct. Focus serves a narrow segment through cost or differentiation. Porter calls the position of a firm that fails to choose a viable approach “stuck in the middle.”
Our reading is that a company with 10 to 50 people rarely has the scale for industry-wide cost leadership, so focus with differentiation is usually the more realistic choice. A rival can match a price cut, and if your costs are higher, the cut leaves you with less room to serve the customer. The guide to competitive moats for small businesses examines how a chosen position can be defended.
Check: Name your target segment and the specific reason it would choose you. Then compare the margin on that offer with the margin after a price cut you are considering.
How do I decide what to stop doing?
Put activities that repeatedly fail your policy and economic tests on a stop list with an end date. Porter argues in What Is Strategy? that strategy requires trade-offs, and distinguishes operational effectiveness from strategy.
The essence of strategy is choosing what not to do.
Getting better at an activity can improve operations without deciding which activities belong in your chosen position. A faster quoting process, for example, does not by itself answer which customers you intend to serve.
Our stop-list test applies to each product, customer segment or recurring activity. Record a pass or fail for each question; two failures put the item on the stop list with an end date.
| Test | Observable check |
|---|---|
| Does it serve the guiding policy? | Point to the policy it advances and explain the connection. |
| What margin remains after the owner's time? | Subtract the cost of the owner's time from the margin and decide whether the remainder meets your threshold. |
| Would we start it today? | Write down whether you would choose it now, knowing its actual demands and results. |
A stop date can include time to honor existing commitments. It should still name the last date for accepting new work in that category.
Check: Run the three tests on one offer at your next review. If it fails two, write its end date and the person responsible for the transition.
How do I make bets when I can't predict what happens?
Sort decisions by how consequential and reversible they are, then match the decision process to the cost of getting back out. Jeffrey P. Bezos's 2015 Letter to Shareholders describes consequential, irreversible or nearly irreversible choices as Type 1, or one-way doors, and changeable, reversible choices as Type 2, or two-way doors.
The letter says most decisions are Type 2 and warns that larger organizations can drift into applying the heavy Type 1 process to them. Its instruction is: “Type 2 decisions can and should be made quickly by high judgment individuals or small groups.”
Our classifications below are starting points. The actual exit cost, contract terms and customer commitments can change the classification.
| Decision | Our classification | Cost to undo | Who decides | How fast |
|---|---|---|---|---|
| Trial an AI scheduling tool for 30 days | Two-way | Trial cost and time to move notes back | Operations owner | Promptly, after an exit check |
| Change prices on new quotes | Two-way | Possible lost quotes and time to revise pricing | Pricing owner | Promptly, after a margin check |
| Sign a three-year software contract | Closer to one-way | Contractual exit cost and migration work | Owner | Carefully, after reviewing terms |
| Drop a customer segment | Closer to one-way | Lost relationships and possible difficulty returning | Owner | Carefully, after reviewing commitments |
| Open a second location | One-way | Lease, setup and closure costs | Owner | Carefully, after reviewing obligations |

The doors separate choices that need careful deliberation from choices a small group can make and reverse quickly.
Check: Pick a pending decision and write down the actual cost to undo it. If the cost is unknown, find it before choosing the decision process.
How can I use scenario planning without building a forecasting department?
Our approach is to choose the uncertainty that matters most, describe two or three plausible futures and test whether the guiding policy survives each. The exercise prepares decisions for different conditions without requiring you to declare one forecast correct.
Wack's account of forecasting errors is useful here: a forecast can be reasonable and still fail when conditions change. For a small firm, the value of a scenario is the question it forces. What would you do if the assumed source of demand weakened, and what evidence would make you act?
For example, suppose your main uncertainty is how many inquiries arrive through AI answers. Write one future in which that channel accounts for little of your inbound demand and another in which it accounts for a measurable share. For each, ask whether your policy still directs resources toward customers you can serve profitably, and whether an action needs to change.
Check: Put the two futures beside the written policy. Record one observable signal for each future and one action you would reconsider if that signal appeared.
How often should I revisit the strategy?
Our cadence is a 60-minute review every quarter, a yearly rewrite and an early review when a written trigger occurs. The quarterly meeting checks whether the existing diagnosis and actions still hold; the annual rewrite gives you a chance to reconsider the policy itself.
Our four quarterly questions are:
- Is the diagnosis still true?
- What did we refuse this quarter?
- Which actions shipped?
- Which leading numbers moved?
Bring evidence for each answer. A refusal might be a declined proposal or a segment you chose not to pursue. An action has shipped when the agreed work is available or operating; a meeting about it does not count.
Write early-review triggers before you need them. Examples include a top three customer leaving, a rival cutting the price of your core offer or a new channel taking a measurable share of inquiries. Set the inquiry threshold in your own records so the trigger can be observed.
Check: Put the next quarterly review and annual rewrite on the calendar. Assign someone to watch each early-review trigger.
How do I involve a small team without turning it into a committee?
Our approach is to involve customer-facing employees in the diagnosis, leave the guiding policy decision with the owner and give every coherent action one named owner. People who hear why customers decline or leave can supply evidence that an owner may not see directly.
Ask those employees for specific lost deals, recurring objections and changes in customer requests. Record the examples alongside the proposed diagnosis. Their contribution is evidence for the choice, while the owner remains accountable for choosing a policy that makes trade-offs.
Once the policy is decided, an action owner needs authority to carry out the assigned work and a way to report obstacles. One named owner does not mean one person does every task. It means the team knows who is responsible for moving that action forward.
Check: Ask three staff members separately to state the guiding policy in one sentence and name one thing it rules out. If their answers conflict, revise the wording or explain the decision before assigning more work.
How do I tell whether the strategy is working?
Give each coherent action one leading measure and one lagging measure, then review both against the diagnosis. A leading measure tracks activity or an earlier signal; a lagging measure records a result that follows. Treat the leading measure as a guess to check: compare it with the lagging result over past quarters.
For a hypothetical renewal action, the share of expiring contracts contacted before renewal could be the leading measure. The share of expiring contracts renewed could be the lagging measure. If calls are happening but renewals remain weak, the team has evidence to inspect the offer or its diagnosis of the problem; the two numbers do not identify the cause on their own.
Use the guide to KPIs and targets that drive action to define each measure. The guide to decisions, not dashboards explains why more data does not settle a strategic question: someone still has to choose what to do with the evidence.
Check: Pick one action and locate its leading and lagging measures in the next review record. If either is missing, define it before judging the action's result.
Should AI be the strategy?
No. Our position is that AI is a means inside coherent actions: “use AI” names a tool without answering the obstacle or specifying what the company will rule out.
The 2026 Report on Employer Firms reports the 2025 Small Business Credit Survey, fielded from 3 September to 14 November 2025. Its 6,525 responses came from a convenience sample of employer firms with 1 to 499 employees, so it was not a random sample.
Of responding firms, 46% said the business or employees currently used AI. Among AI users, about half were experimenting, 44% had partially integrated it and 7% had fully integrated it. The reported uses included writing or marketing at 83%, individual productivity at 61%, and planning or analysis at 51%; accuracy was a challenge for 46% and adapting tools to business needs for 43%.
Reaching customers and growing sales remained the most commonly reported operational challenge. The guide to what running a business means in the AI era examines our view that AI re-prices tasks while customers, cash, people and decisions remain the owner's job. Once the policy identifies repeat work worth automating, Iron Goo's operations service can audit, build and operate chatbots for repeat inbound questions and an agentic operations layer for internal ops work.
Check: For every proposed AI action, name the obstacle it addresses and the person who will check its output. Remove “use AI” from the strategy page if it stands alone as the policy.
What does a one-page strategy look like?
A one-page strategy states the diagnosis, guiding policy and coherent actions, then records what the policy rules out and when to review it. The example is hypothetical: the company, its conditions and its choices are invented to show the format.
The diagnosis makes an assertion the owner can test. The policy selects a customer segment and geography, while the actions support that choice. The AI task sits inside an action because scheduling notes and first-draft quotes are means of carrying out the policy.
COMPANY
A fictional 25-person commercial HVAC service company
DIAGNOSIS
New installation work goes to low bidders found through search
and AI answers. Service contracts renew at high rates, but we
do not offer them consistently to suitable building owners.
GUIDING POLICY
Become the maintenance contract firm for buildings over
50,000 square feet in one metro area.
COHERENT ACTIONS
1. Create contract pricing tiers for those buildings.
2. Reserve technician capacity for contract maintenance;
trial AI for scheduling notes and first-draft quotes,
with a staff review before either is used.
3. Call contract customers before renewal to discuss service
needs and renewal terms.
RULES OUT
- Residential one-off installs.
- Statewide expansion.
REVIEW TRIGGERS
- A top three customer leaves.
- A rival cuts prices on the core maintenance offer.
- A new channel reaches the inquiry share set in our records.
Check: Replace the fictional details with your own evidence. Ask whether every action supports the policy and whether the ruled-out choices are specific enough for someone to decline a request.
Put your next steps in order
Build on what you have just read with a free ebook on semantic SEO and AI visibility. Get worked examples, decision rules, checklists and a do-this-week action in every chapter, plus a six-week plan for your site.



