How to Stop Changing Business Strategy: A 90-Day Plan You Can Actually Follow

The strategy you are about to replace may be failing. It may also be six weeks into a test you have never actually run as planned.

This happens easily in an expert-led business. A few videos underperform, leads go quiet, or a new idea sounds more exciting. You add a channel, rewrite the offer, change the audience, and call the movement progress. The original strategy never produces enough clean evidence to judge because too many parts changed at once.

A 90-day business plan can create a stable test. Set an outcome range, work backwards to the inputs you control, build the supporting path once, name the behaviours likely to derail you, and review execution on a fixed schedule. The plan gives your work enough time to teach you something before you replace it.

Constant strategy changes destroy the evidence

When I recorded the source video for this article, I was 45 days into my first 90-day sprint. The early views and enquiries were lower than I wanted. My usual response would have been to interfere with everything. This time, the plan kept the inputs stable long enough for me to see early signs of movement.

That is a personal observation from halfway through one sprint, not proof that the process caused a commercial result. Its value is narrower and more useful: the process stopped a familiar reaction from ending the test before I had enough evidence.

Changing several things at once creates an attribution problem. If the results improve, you cannot tell which change helped. If they get worse, you cannot tell what broke. Each reset feels active while making the next decision less informed.

1. Set an outcome range for the next 90 days

Write the date 90 days from now. Then define three versions of the outcome:

1. Base: the minimum result that would make the sprint worthwhile.

2. Target: the result you are planning towards.

3. Stretch: a strong result that remains plausible without distorting the work.

The range gives the plan direction without turning one number into a daily verdict on your ability. You can include the measures relevant to the business, such as subscribers, enquiries, buyers, revenue, or completed offers, but each number needs a clear definition.

Ninety days is a useful decision container. Some strategies need a shorter or longer test because sales cycles, cash constraints, seasonality, and risk differ. Choose the horizon deliberately and record why it is long enough to create evidence.

2. Turn outcomes into controllable inputs

Outcomes tell you where you want to go. Inputs tell you what to do today.

Work backwards from each outcome and list the recurring actions you believe can influence it. In the source sprint, those inputs included videos, short-form content, text posts, emails, and offers. Completing them did not guarantee an outcome. It created a consistent body of work that could be reviewed.

Use an input table with five columns:

1. Outcome served: the result this work is meant to influence.

2. Input: the action you will complete.

3. Cadence: the daily or weekly commitment.

4. Completion rule: the observable condition that counts as done.

5. Capacity check: the time, energy, or resource the commitment requires.

A completion rule prevents vague scoring. ‘Work on YouTube’ cannot be audited. ‘Publish the planned long-form video’ can. The capacity check catches plans that look disciplined on paper but cannot survive an ordinary week.

Measure the inputs daily. Inspect the outcomes at the review point. That separation reduces the temptation to treat every quiet day as a strategic emergency.

3. Plan the ordinary day

A plan can point towards an attractive result while requiring a working life you dislike. Describe the average day that would produce the inputs instead.

Write down when the work begins, what gets made first, when selling or follow-up happens, how you respond to a disappointing result, and when work stops. The source process uses this picture to remove repeated morning decisions. The work is already visible, so attention can go into completing and improving it.

This is also a realism check. If the ordinary day needs more hours, energy, or attention than you can repeatedly supply, reduce the commitments before the sprint begins. A smaller plan that gets executed creates more evidence than an impressive plan that exists only in a document.

4. Separate recurring inputs from build-once work

Activity needs somewhere useful to lead. Content without an offer path can attract attention without creating a business result. Outreach without a clear next step can create conversations that go nowhere.

Create a separate build-once list for the assets that support the recurring work. Depending on the strategy, this might include an offer page, landing page, lead magnet, email path, webinar, booking route, or measurement setup. Build and test these dependencies early, then keep them out of the daily input count.

This distinction prevents a common planning mistake. Publishing more is not the same as building the path that lets an interested person understand what to do next.

5. Write the sabotage list before you need it

Most plans name the work and ignore the person who has to do it. The sabotage list is a short pre-mortem: the predictable behaviours and situations that could pull you away from the planned inputs.

My list included hearing about an exciting new strategy and wanting to add it immediately. It also included celebrating early progress in a way that damaged the following day's work. Your patterns may differ, but the useful structure stays the same:

1. Derailer: the behaviour, event, or temptation.

2. Early sign: what you will notice before it takes over.

3. Default response: the action you will take when the sign appears.

4. Recovery rule: how you return after a missed input without rewriting the whole plan.

Naming a derailer does not remove it. It gives you a prepared response at the moment when improvisation is least reliable.

6. Review execution without reinventing the strategy

Keep the strategy stable between scheduled reviews. Use the weekly review to improve how you execute it.

Record the planned inputs, completed inputs, observed signals, friction, and one improvement for the next week. A useful review distinguishes four different diagnoses:

1. Execution problem: the planned work was not completed.

2. Quality problem: the work was completed, but a specific part can be improved.

3. Infrastructure problem: the recurring activity has a broken or missing next step.

4. Strategy problem: the core assumption is contradicted after the agreed test has actually been run.

The first three diagnoses permit changes inside the strategy. The fourth may justify replacing it. This keeps learning active without turning every review into a new plan.

A strategy can change before day 90 when a material fact invalidates it, the downside threatens the business, or the test cannot be run safely or honestly. The point is to define those conditions in advance. Persistence should never become an excuse to ignore decisive evidence.

The one-page 90-day anti-pivot worksheet

Copy the following structure into a document or notebook. Keep it visible for the full sprint.

A. Decision

End date; strategy in one sentence; reason this test deserves 90 days; base outcome; target outcome; stretch outcome.

B. Input scoreboard

For every recurring input, record the outcome served, input and cadence, completion rule, capacity required, and where completion will be recorded.

C. Build-once dependencies

For every supporting asset, record the asset, the recurring input it supports, owner, completion date, and the test that proves it works.

D. Sabotage responses

For every likely derailer, record the derailer, early sign, default response, and recovery rule.

E. Review boundaries

Record the weekly review time, evidence reviewed, execution changes allowed, strategy changes forbidden between reviews, evidence that would justify replacing the strategy, and conditions that require an immediate stop.

F. Weekly decision log

Finish each review with one recorded decision: continue unchanged; adjust one part of execution and keep the strategy; repair a build-once dependency and keep the strategy; or replace the strategy because the pre-agreed evidence threshold has been met.

The worksheet will not make a strategy effective. It will show whether you executed the test, what changed, and why the next decision is justified.

Let one plan teach you something

A useful 90-day plan reduces unnecessary decisions. It turns a desired result into recurring work, protects that work from predictable derailers, and creates a review record that separates discomfort from evidence.

The aim is a better next decision. Give one strategy a fair test, improve its execution as you learn, and change direction when the recorded evidence earns that change.