TOWS matrix
Combine evidenced SWOT factors into SO, ST, WO, and WT action candidates, then state their assumptions, resource limits, and validation steps.
Use TOWS after a focused SWOT has enough evidence to support a discussion. Pair internal strengths or weaknesses with external opportunities or threats. A useful entry explains a possible action and how to test it, rather than simply joining two factor names.
A shared goal and identified SWOT factors, ideally with IDs and sources. Keep unverified demand, competitive conditions, and resource estimates visible.
Traceable action candidates with a reason to consider them, assumptions to check, and conditions for continuing or stopping.
How to use it
- State the goal and record internal S/W and external O/T factors with their evidence and IDs.
- Explore SO actions that use strengths to pursue opportunities and ST actions that use strengths to respond to threats.
- Explore WO actions that address weaknesses to pursue opportunities and WT actions that limit exposure where weaknesses meet threats.
- Attach factor IDs, the proposed mechanism, a test, and relevant constraints to each candidate.
- Choose a manageable next test; combine overlapping candidates and keep unsupported combinations blank.
Keep its limits in mind
The four combinations are not a ranking and do not require four simultaneous initiatives. Unsupported combinations can stay blank. TOWS does not calculate priorities or establish that an action will work.
Test a household replenishment service
Fictional retail example. A team has gathered initial SWOT evidence for household replenishment. Demand scale and the duration of a competing promotion remain unknown.
On a small screen, scroll across the diagram to read it.
Evidence trail: The SO candidate uses customer trust and reported interest. Neither input establishes that 30 invited households will place an order.
Resource check: Limited staffing affects both delivery experiments and exposure to low-price competition. Confirm capacity and the trial budget before committing.
Next step: Choose one affordable test and define the evidence needed to continue. Do not treat all four candidate actions as approved work.
Three reasons to postpone a TOWS session
First, postpone the matrix if the question has no boundary. “Grow the business” can produce dozens of attractive combinations without helping anyone decide. “Should this shop test a replenishment offer using its current staff?” names a decision, an organization and a constraint. Write that sentence above the matrix and remove factors that do not affect it.
Second, do not cross unsupported claims. A salesperson’s impression that customers want subscriptions is a research lead, not established demand. Keep the speaker, context and uncertainty. A blank opportunity cell is more useful than a plausible market claim invented to complete the pattern. If you cannot explain where a factor came from, return to observation before generating strategy.
Third, do not use TOWS to settle an urgent operational incident, prove a cause, or choose a winner by counting filled cells. An incident needs containment and investigation. A choice between feasible options needs explicit constraints and tradeoffs. TOWS helps generate candidates; it cannot supply evidence, authorization or delivery capacity.
Choose the question before choosing the diagram
These methods can support different parts of a discussion. They are not mandatory consecutive stages. Use the smallest method that answers the current question, and carry its unresolved assumptions into any later work.
| Current question | Method | Useful output and limit |
|---|---|---|
| What internal and external conditions matter? | SWOT analysis | An evidenced factor inventory; it does not choose a strategy. |
| What action could follow from these particular factors? | TOWS matrix | Traceable action candidates and tests; four combinations are not four priorities. |
| Which usage situations have we failed to imagine? | Scenario Graph | Explicitly imagined who/when/where/what stories; their frequency is unknown. |
| Which feasible candidate fits our constraints? | Decision matrix | A comparison with scoring reasons and sensitivity checks; scores are judgments. |
A worked continuation of the replenishment example
The following is an original fictional continuation of the example above. It adds teaching assumptions rather than new evidence about a real shop. Begin with four source notes: S1 is a follow-up note describing trust from regular customers; W1 is a roster showing limited delivery staff; O1 is interview interest in saving shopping time, with its prevalence unknown; T1 is a competing promotion page with no known end date. None of these notes supplies a trial budget or confirms willingness to pay.
Keep the raw note separate from its interpretation. “An interviewee asked about replenishment” can support testing an offer. It cannot support “most local households want a subscription.” Likewise, limited staffing makes fulfillment a constraint; it does not establish how many orders can be handled safely. Ask for the roster window, other duties and the time required by each order before promising capacity.
For the SO candidate, invite the existing example’s 30 households to consider pickup replenishment. Record invitations, responses, actual orders, completed collections and repeat orders separately. Thirty invitations are an exposure count, not the demand denominator for an unrelated neighborhood survey. A favorable reply is not an order, and a repeat rate needs both a defined eligible group and an adequate observation window.
For ST, propose checking whether selection and after-sales service matter to customers who actually choose the shop; do not assume these compensate for a price difference. For WO, measure the work required by a fixed delivery window before offering more slots. For WT, propose a service radius and order cap, with the actual limits still awaiting an owner’s agreement. The same staffing constraint affects several candidates; do not count its supposed benefit several times.
| Candidate | Evidence to collect | Decision still needed |
|---|---|---|
| SO: S1 × O1, pickup invitation | Invitations, orders, completed pickup and later repeat use | Offer, window, effort allowance and approval |
| ST: S1 × T1, service differentiation | Recorded customer reasons, without attributing every repeat to service | Whether the evidence warrants a separate test |
| WO: W1 × O1, fixed delivery window | Person-minutes, other duties and incomplete orders | Whether staffing can cover the proposed scope |
| WT: W1 × T1, limit exposure | Trial spending against an agreed cap | Cap, review time and who can pause the trial |
Write the conclusion and review before claiming success
A defensible conclusion is: “The factor pairs suggest a limited pickup test worth discussing. Demand, effort and budget remain unconfirmed. We propose one test after the owner agrees to its scope and stopping conditions.” It is not “TOWS shows that subscriptions will increase revenue.” Preserve the other candidates in the record so a later reviewer can see why they were deferred.
Before any real trial, agree on the offer, observation window, records, spending limit, responsibilities and what would cause a pause. Then distinguish execution from outcome at review: sending invitations means the invitation happened; it does not mean customers ordered or the trial was profitable. Record returns, incomplete orders and additional staff work. If conditions changed, explain the difference before comparing periods. A small self-selected group cannot establish neighborhood-wide demand or a causal improvement.
For a strategy meeting or a weekly update, keep the final request narrow: approval to investigate or run a bounded trial, rather than approval for all four cells. Record who actually agreed and when. If the meeting produces no authorization, the document remains a proposal even when the matrix is complete.
Check your work
- S/W describe internal conditions; O/T describe external conditions.
- Each candidate traces to the relevant factor pair.
- A possible effect is not written as an achieved result.
- The test fits the resource limits and has an explicit stopping condition.
Framework references
References explain the framework and its scope; they do not validate the fictional example's results.
Put the framework to work
Copy the outline and add your question, evidence, and open questions.
Read a practical guide →Practice with an example
Trace an action to its conditions, then retain assumptions, a test, effort limits, and stopping conditions in a small plan you can review.
After SWOT: turn an idea into a testable action →