Insights / Company Building · · Updated · 14 min read

Continue, stop or scale: a 30/60/90-day decision framework

Use 30, 60 and 90-day checkpoints to review activation, retention, costs and readiness. A practical framework for deciding when to continue, stop or scale.

Ninety days is a review horizon, not an automatic deadline to scale. A useful 30/60/90-day plan tests whether people reach value, whether the product remains useful over its natural usage cycle, and whether a controlled increase in demand can be served responsibly. If the evidence is incomplete at day 90, name the uncertainty and decide whether a bounded further test is worthwhile.

Without a clear decision process, new companies tend to drift. Nobody wants to be the person who says "this isn't working", so the company keeps running on hope. Months pass. Attention and money that could have gone to something stronger are spent on something that stopped improving long ago.

This note proposes a decision framework for a digital company or feature. It is educational guidance, not a report that every Oryvelon company follows these checkpoints or has achieved particular results. The options are continue, stop or scale; the timing and evidence needed depend on the product, its customers and the decision at stake.

Write the targets before launch

The most important rule is about timing: KPIs are written before launch, not after.

After launch, it is very easy to look at whatever happened and find a way to call it a success. "We didn't hit our sign-up target, but engagement is great." "Revenue is low, but we learned a lot." Sometimes those statements are true. Often they are a way of avoiding a decision.

Write assumptions and decision criteria before gathering the evidence. Strategyzer’s Test Card separates the hypothesis, test, measurement and success threshold. This can help make a review auditable. If a target changes, retain the earlier definition and explain why the new one is appropriate; do not rewrite the past result.

Choose KPIs that measure behaviour

Useful checkpoint KPIs connect to a customer action and the business question being tested. Treat page views, feature delivery and interview comments as context; none alone establishes that people obtained value. The table suggests a sequence, not universal success thresholds.

Stage Weak KPI Strong KPI
30 days Pages published, visitors Share of sign-ups who complete a first meaningful action
60 days Features shipped Share of active users still active after several weeks
90 days Total revenue in isolation Paying customers, retention and gross margin per customer

Choose the action for the actual product. For example, a hypothetical store tool might measure a completed reporting task, while a hypothetical marketplace might measure requests that receive a useful response. These are possible definitions, not claims about features or tracking deployed by Oryvelon companies. Specify the eligible population, numerator, denominator and time window.

Keep the decision metrics focused. Record a few primary measures and separate quality or safety guardrails. A long list makes it easy to highlight one favourable number while overlooking a material problem.

The three checkpoints

Day 30: is anything working?

Use the first review to check delivery and first value. Thirty days may or may not provide enough observations for a particular business; inspect what has actually happened:

  • Are the right people arriving?
  • Do they understand what the product does?
  • Are they completing the first meaningful action?
  • Is anything technically broken or unexpectedly expensive?

Choose the response to the evidence. If a broken step prevents intended users from reaching value, a bounded fix-and-retest cycle may be useful. If the central premise is disproved or the product cannot be operated responsibly, do not wait for a later calendar checkpoint to pause or stop.

Day 60: are people coming back?

At the next review, examine repeat value where repeat use is expected:

  • Do activated users return?
  • Are early paying customers staying?
  • Is the cost of serving each customer in line with expectations?
  • Is at least one acquisition channel showing sustainable economics?

Match the observation window to the product. A daily workflow and a seasonal purchase should not share an arbitrary return target. Compare cohorts with equal time to return; recent sign-ups have not had the same opportunity as older ones. Infrequent use can be appropriate, so define the expected behaviour before interpreting an absence.

Day 90: is this a business?

At the day-90 review, ask whether the available evidence supports the next commitment:

  • Is revenue growing from paying customers who stay?
  • Does each customer generate more than they cost to acquire and serve, or is there a credible path to that?
  • Is the team learning faster than the problems are growing?
  • Compared with other opportunities in the group, does this deserve more, the same or less attention?

A defensible scale proposal should state the repeatable source of demand, the cost of serving it, delivery capacity and the conditions for pausing. If a renewal cycle has not elapsed, record retention as unobserved rather than successful. If the sample is small, say which limited decision it can support. A further test needs a specific question, resource limit and end date; another 90 days by default is not a plan.

What each decision means

The three decisions are deliberately simple. Each has a clear meaning.

Continue means funding a defined next step within an agreed limit. Record what remains uncertain, what will be tested and what result would change the decision. Continuing is not evidence of success and should not hide an indefinite commitment.

Stop means the company or initiative ends in its current form. That might mean shutting it down, pausing it until conditions change, folding its useful parts into another company, or reshaping it into something different that starts the protocol again. Stopping is described more below, because it is the hardest decision to make well.

Scale means the evidence is strong enough to put more resources behind it: more budget for acquisition, more engineering time, more of the founders' attention, perhaps outside partners. A scale decision should also specify what must stay true — margins, retention, quality — as the company grows.

Evidence to weigh at a checkpoint

For a recurring product, the following evidence can inform a checkpoint. Choose its weight according to the hypothesis and record the limits of each observation.

Heavily weighted: - activation and retention curves; - paying customers and whether they renew or repurchase; - cost to serve each customer, including AI costs; - qualitative feedback from customers who use the product regularly; - whether a sustainable acquisition channel is emerging.

Lightly weighted: - total visitors; - social media attention; - feedback from people who do not use the product; - the team's excitement — important for morale, weak as evidence.

Context, not proof: competitors’ funding and newly discovered metrics may prompt a useful question, but they do not validate the original hypothesis. Record genuinely new evidence separately rather than discarding it or retrospectively claiming it met a target.

Why stopping is a skill

Stopping or narrowing an initiative can be a deliberate operating decision. A company builder can assess the resources a further test requires alongside the evidence it is likely to produce.

Attention is finite. Every company in the group competes for the founders' time, engineering capacity and budget. A company that should have stopped takes those resources from one that should have scaled.

Separate past spending from the next commitment. Ask what can be learned or delivered with the next unit of time and money. A pre-agreed review helps keep the current decision distinct from a desire to justify earlier work.

Preserve what can responsibly be reused. Record findings and review ownership, licences, contracts and data restrictions before moving assets. A shared group relationship does not automatically permit transferring customer information or third-party materials.

Plan the effect on customers. If a live product must close, identify service commitments, communication needs and any required export, refund or deletion process before choosing dates. A decision document should identify the responsible person and unresolved obligations, not promise options that have not been checked.

Planning a responsible stop

For a proposed closure of a live product, work through the applicable commitments. This is an operating checklist, not a statement that Oryvelon has closed a company or a substitute for reviewing the actual contracts and obligations:

  1. Tell customers early and clearly, with a reasonable timeline.
  2. Offer data export so customers leave with what belongs to them. See Retention, export and deletion.
  3. Refund fairly where customers have paid for service they will not receive.
  4. Delete what should be deleted according to the product's retention policy.
  5. Keep the domain and, where appropriate, a page explaining what happened.
  6. Write down the lessons, including what the checkpoints showed and when.
  7. Reuse useful assets only after verifying ownership, permissions and any restrictions on their transfer.

Checkpoints for features, not only companies

The framework can also apply to a feature, pricing experiment or new market. Define the smallest decision worth making and the evidence it needs. A feature may be improved, narrowed, paused or retired without treating the whole company as the unit of analysis.

How checkpoints connect to capital allocation

A group can compare the next proposed commitment across initiatives: what remains uncertain, what the test costs and what would change the decision. See Capital allocation inside a company builder for related questions. A consistent format supports comparison; it does not make different products’ metrics interchangeable.

Before comparing economics, verify revenue, serving costs, shared-cost assumptions and the period covered. Mark estimates and unavailable costs explicitly. See Unit economics per product for related questions to include in the review.

A checkpoint template

A proposed checkpoint record can use the following template. Assign someone to assemble it and someone with authority to make the decision:

  1. KPIs as written before launch, with the actual numbers next to them.
  2. What surprised us, positively and negatively.
  3. What customers told us, quoting real words where possible.
  4. Cost to serve, broken down.
  5. The proposed decision — continue, stop or scale — and the reasoning.
  6. What we expect at the next checkpoint, if the decision is continue or scale.

Keep the main decision concise and link to the underlying reports. Include the reporting dates, cohort definitions, missing data, competing explanations and proposed resource limit. A short summary should make evidence accessible, not omit uncertainties to meet a page limit.

Common failure modes

Check for patterns that can undermine a review:

  • Moving the goalposts — changing KPIs after seeing the results.
  • Averaging away the signal — reporting totals that hide weak retention.
  • Permanent "continue" — continuing without stating why or what should change.
  • Scaling too early — putting money into acquisition before retention is proven.
  • Stopping too quietly — letting a product decline without a clear decision.

For each issue, record a concrete correction and who will verify it. Naming a risk does not establish that a particular team has avoided it.

Examples of checkpoint reasoning

The scenarios below are hypothetical decision exercises. They describe possible responses, not Oryvelon results or a rule that one metric determines the outcome.

A strong start that fades. A hypothetical consumer product attracts sign-ups, but few eligible users return within the expected cycle. Before increasing acquisition, investigate whether users reached value and whether tracking or cohort definitions explain the drop. A bounded product test may be reasonable; expansion without resolving the gap would leave the core question unanswered.

A slow start with loyal users. A hypothetical business tool has a small group of repeat users. That supports investigating their needs and testing whether similar customers can be reached. It does not establish market size or repeatable acquisition. Keep the sample size and observation window visible before proposing expansion.

Healthy usage, unhealthy economics. A hypothetical product costs more to serve than it earns. A limited pricing or delivery-cost experiment may be worth testing if there is a plausible path to improvement; otherwise continuing may only extend the loss. See Subscription plus usage for pricing considerations, not a guaranteed fix.

A marketplace that fills one side only. In a hypothetical marketplace, supplier registrations grow while useful buyer requests remain scarce. Inspect matching, response and fulfilment before buying more supply. Define a test of the missing side and its stopping condition. See Two-sided marketplaces and the cold-start problem.

Nothing moves. If valid tests repeatedly fail to support the core premise, a stop or material change may be appropriate. First distinguish a failed hypothesis from an unobserved outcome caused by broken tracking, an unavailable product or too little exposure. Do not call absent evidence either success or failure by default.

Who makes the decision

Name a decision owner with the appropriate authority and invite challenge to the evidence. Record disagreements and the chosen action. Where uncertainty remains, compare the cost and reversibility of a further test with the consequences of stopping or expanding.

The record matters. Six months later, when someone asks why a company was scaled or stopped, the answer should be a short document with the numbers and the reasoning, not a memory.

What scaling actually involves

Treat scaling as a set of explicit commitments rather than a single switch. A proposed scale decision should cover:

  • Acquisition budget: the limited increase, intended channel, measurement window and acceptable cost per acquired customer.
  • Product capacity: which improvements the extra engineering time will go to, prioritised by what retained customers value most.
  • Operational readiness: support, onboarding and documentation that can handle more customers without quality dropping.
  • Guardrails: the retention, margin and quality levels that must hold as volume grows. If they slip, scaling pauses.
  • Next review: a date and owner for checking whether the added resources produced the expected result and whether the next increase is justified.

Start with an increase that can be evaluated and reversed where practical. Define what would pause it: for example, a deterioration in a chosen quality measure, an unaffordable serving cost or an unresolved delivery constraint. The threshold must fit the actual product and be supported by its evidence; there is no universal scale-ready percentage in this framework.

Setting targets when you have no history

The first objection to writing KPIs before launch is obvious: how can you know what "good" looks like for a product nobody has used yet? You can't, precisely. But you can do much better than guessing.

For a product without history, separate the evidence you have from assumptions still to test:

  • Validation evidence. Record what people actually did in the relevant test. An email signup or stated willingness to pay is not paid retention and does not establish a minimum activation rate. Link the target to the observation and keep that limitation visible.
  • The cost floor. List known fixed and variable costs and state which are estimates. Use plausible demand scenarios to identify what must be tested, rather than inventing a day-90 revenue target from a desired outcome.
  • Patterns from similar products. External examples can suggest questions, but their audiences, definitions and usage cycles may differ. Treat them as context unless the comparison is demonstrably relevant.

Write the target with its reasoning and uncertainty. A range can describe an assumption, but it is not measured evidence. When results differ from the expectation, record which assumption changed and what the next test will resolve.

Worked example: a hypothetical store alert

The following is a hypothetical store-reporting feature, not an Oryvelon company deployment or measured customer result.

Imagine a store tool that flags unusual refund activity and links an alert to the relevant underlying records. Before release, test whether the signal is accurate, understandable and actionable. See Shopify operations signals for related design questions.

The KPIs written before release might be:

Checkpoint Question KPI
Day 30 Do merchants notice and understand the alert? Share of alerts opened; share followed by a suggested action
Day 60 Is it useful enough to keep? Share of stores that keep the alert on; share of alerts merchants mark as "not useful"
Day 90 Does it earn its place? Retention of stores that use it versus those that do not; cost per alert against plan price

At the first checkpoint, hypothetical users open alerts but rarely act. Investigate several possible explanations: the task may already be complete, the signal may be wrong, or the next step may be unclear. At the second, users flag alerts as noisy during a sale; test whether seasonality explains the pattern before changing the rule. At day 90, compare usefulness, false alerts and serving costs against the criteria set for this experiment.

Higher retention among stores that chose to use the alert would be an association, not proof that the alert caused the difference. Microsoft Research explains how properly designed randomized experiments can assess causal effects. Where an appropriate experiment is impractical or the sample is insufficient, state that limitation and keep the decision proportionate to the evidence.

Before the checkpoint: making sure the evidence is real

Before reviewing the result, check the evidence itself. Use the following as a proposed verification checklist, not a claim about checks already performed:

  • [ ] Tracking for every KPI has been tested end to end, not assumed.
  • [ ] Test accounts, staff activity and internal orders are excluded from the numbers.
  • [ ] Cohorts are defined by sign-up date, so new arrivals do not flatter retention.
  • [ ] Costs for the period are attributed to the company, including AI usage.
  • [ ] At least a handful of real customer quotes have been collected from people who use the product.
  • [ ] Any change to a KPI since launch is listed with the date and reason.

Collect only the information needed for the decision and verify what the measurement setup actually records. A checkpoint does not require tracking people across unrelated websites. See Analytics without surveillance for related review questions.

A broken event or an unexcluded test account can distort a result. Check the underlying definitions and inputs before changing the product because of a dashboard number. A missing row is unavailable evidence until its meaning is established; it is not automatically zero.

Summary

Use 30, 60 and 90-day reviews to make explicit, evidence-based commitments. Continue with a bounded test, stop or change a disproved approach, or scale with recorded limits and a further review. The calendar alone proves none of these choices. Oryvelon’s company directory explains the portfolio relationship; this guide supplies a decision method rather than company performance claims.

Questions and answers

What are 30/60/90-day KPIs?

Measures chosen for reviews around 30, 60 and 90 days, with defined populations, time windows and decision criteria. These are proposed checkpoints, not universal success thresholds.

Does a 90-day plan mean the company should scale on day 90?

No. Review demand, repeat value, costs and delivery capacity. If important evidence is missing, record the uncertainty and decide whether a bounded further test is worthwhile; the date alone does not justify scaling.

What if the sample is too small at the checkpoint?

Report the sample and what remains unobserved. Verify tracking and exposure, then define a limited next test if it is justified. Do not label missing evidence a success or failure by default.

NextDesigning a recurring revenue loop from day one →