But where did the 5% come from?
If it reflects relevant historical performance, it may be a reasonable forecasting input. If it represents an improvement the team hopes to achieve, it is a target. If it comes from a benchmark or management judgment, it is an assumption.
The arithmetic treats all three identically. Your business should not.
A credible funnel forecast requires traceability: every material input should have a clear definition, source, rationale, and level of confidence.
Without that discipline, a planning model can turn ambition into an apparently validated revenue projection.
A KPI Is a Measure, Not Evidence of Future Performance
A key performance indicator identifies what the business intends to monitor. It does not establish what the next result will be.
“Lead-to-customer conversion” is a KPI.
“Achieve 5% lead-to-customer conversion next quarter” is a target.
“Of 5,000 leads in the completed cohort, 200 became customers” is an observation.
“We expect the next comparable cohort to convert at 4%” is a forecasting assumption informed by that observation.
These statements play different roles:
| Category | What it describes | Example |
|---|---|---|
| KPI | The metric being monitored | Lead-to-customer conversion |
| Target | The desired result | Reach 5% conversion |
| Historical observation | A measured outcome in a defined context | A completed cohort converted at 4% |
| Planning assumption | An input applied to a future situation | The next cohort will convert at 4% |
| Scenario output | The mathematical consequence of selected inputs | At 5% conversion, the plan produces 500 customers |
| Forecast | The outcome currently expected, informed by evidence and judgment | The current outlook is 400 customers, subject to stated risks |
Sales forecasting generally combines historical performance, current pipeline information, and market conditions to estimate results for a specified period. It is not simply the act of entering a goal into a model.
The distinction does not reduce accountability. It makes accountability more useful.
How a Target Becomes a Misleading Forecast
Consider an illustrative six-month acquisition plan.
| Input | Evidence-led reference case | Target scenario |
|---|---|---|
| Leads | 10,000 | 10,000 |
| Lead-to-customer conversion | 4% | 5% |
| Customers | 400 | 500 |
| Revenue per customer within the period | $2,500 | $2,500 |
| Modeled revenue | $1,000,000 | $1,250,000 |
The target scenario produces $250,000 more revenue.
But that additional revenue comes entirely from an assumed conversion improvement. The model has not demonstrated that the improvement will occur.
If leadership funds hiring or other commitments against the $1.25 million output, it is making a decision that depends on an unvalidated condition.
That may be an acceptable risk. It should be an explicit one.
A clear presentation would say:
“The current reference case produces $1 million. Achieving the 5% conversion target would increase the modeled result to $1.25 million. The improvement remains subject to validation.”
It should not say:
“The forecast is $1.25 million because our conversion KPI is 5%.”
Historical Data Is Evidence—Within Its Context
Historical performance provides an important starting point. It is not a promise that the future will repeat.
Before carrying an observed rate into a forecast, ask whether the comparison remains relevant.
Are you measuring the same population?
A conversion rate for demo requests should not automatically be applied to webinar attendees.
Likewise, performance among existing customers may not represent prospects unfamiliar with the brand.
Has the cohort had enough time to convert?
Customers closed this month may have originated from leads generated several months ago.
Dividing this month’s customers by this month’s leads can combine different populations and distort the rate.
Follow defined lead cohorts through an appropriate observation window. Distinguish conversion achieved before the planning deadline from eventual conversion.
Has the operating environment changed?
A historical result may become less transferable when the business changes:
- Pricing.
- Customer segment.
- Geographic market.
- Acquisition channel mix.
- Qualification criteria.
- Sales capacity.
- Offer or product.
- Purchase timing.
A precise observation can still be a weak basis for a different future situation.
The confidence judgment belongs to the transfer from evidence to assumption—not merely to the accuracy of the historical report.
Build an Assumptions Register Beside the Forecast
A spreadsheet shows how inputs connect. An assumptions register explains why those inputs deserve consideration.
Strategyzer recommends making hypotheses explicit and prioritizing them according to their importance and supporting evidence. High-importance assumptions with limited evidence deserve early testing.
For a funnel model, the register should include:
- The assumption.
- The planning value.
- Supporting evidence.
- Limits to that evidence.
- Confidence in applying it.
- Business impact if it is wrong.
- An owner.
- A validation or review action.
Example: A Funnel Assumptions Register
The following register is illustrative. Its confidence labels describe evidence quality and applicability, not statistical probabilities.
| Assumption | Planning value | Supporting evidence | Main limitation | Confidence | Owner and next action |
|---|---|---|---|---|---|
| Existing acquisition activity can generate 8,000 leads over six months | 8,000 leads | Recent delivery from comparable programs | Seasonal variation and audience availability | Medium | Demand generation: review channel-level capacity monthly |
| A new channel can supply another 2,000 leads | 2,000 leads | Vendor estimates and a small pilot | Limited scale and customer-qua lity evidence | Low | Channel lead: run a bounded expansion test |
| Comparable inbound leads will convert at the historical rate | 4% | Mature cohorts under similar qualification rules | Future channel mix may differ | Medium | Revenue operations: monitor conversion by source and cohort |
| The proposed sales-process change will raise conversion | 5% target | A documented process issue and a proposed intervention | No validated uplift yet | Low | Sales lead: test before incorporating uplift into the reference forecast |
| New customers will generate $2,500 within the period | $2,500 per customer | Recent cohort revenue | Late acquisitions have less time to purchase | Medium | Finance: build acquisition-mont h revenue profiles |
| Contracted execution services will cost the agreed amount | Contracte d amount | Signed scope and payment schedule | Extra work and usage charges may sit outside scope | High within scope | Marketing operations: reconcile changes before approval |
The table makes an important distinction visible: confidence can be high in a contracted cost and low in the revenue improvement that cost is intended to produce.
Money spent is not evidence of effectiveness.
Define Confidence Levels Consistently
Avoid labels such as “high confidence” without an agreed meaning.
A practical qualitative scale is:
| Confidence | Working definition | Planning treatment |
|---|---|---|
| High | Relevant, repeatable evidence or firm commitments, with limited material change | Use in the reference case while monitoring |
| Medium | Useful evidence exists, but transferability, variability, or timing remains material | Use with sensitivity analysis and a defined review |
| Low | Evidence is limited, indirect, small-scale, or mainly judgment-based | Treat as a hypothesis; limit dependent commitments |
| Unknown | No adequate evidence supports the value | Do not present as established; design discovery or testing |
These categories are not equivalent to “90%, 60%, and 30% likely.”
Do not convert qualitative confidence labels into probability weights unless you have a defensible method. Similarly, a statistical confidence interval around an observed conversion rate does not automatically account for changes in market, channel, or execution.
Prioritize Uncertainty by Its Business Consequence
Not every weak assumption deserves immediate attention.
An uncertain design-production estimate may have little effect on the decision. An uncertain conversion rate may determine whether the program is viable.
Use a simple prioritization matrix:
| Business impact if wrong | Stronger evidence | Weaker evidence |
|---|---|---|
| High | Monitor carefully and check whether conditions change | Prioritize validation before major commitments |
| Low | Maintain proportionate monitoring | Use a low-cost check or accept the risk explicitly |
This follows the importance-versus-evidence logic of assumptions mapping: focus testing on claims the business depends on but cannot yet adequately support.
For each priority assumption, ask:
What observation would change our spending, staffing, or revenue decision?
If the team cannot answer, the assumption may not be defined precisely enough.
Use Sensitivity Analysis to Show What Is at Stake
A sensitivity table reveals how much the forecast depends on an input.
For 10,000 leads and $2,500 in period revenue per customer:
| Lead-to-customer conversion | Customers | Modeled revenue |
|---|---|---|
| 3% | 300 | $750,000 |
| 4% | 400 | $1,000,000 |
| 5% | 500 | $1,250,000 |
| 6% | 600 | $1,500,000 |
A one-percentage-point change in conversion moves modeled revenue by $250,000.
That is useful for deciding whether a conversion assumption needs stronger evidence before approval.
The table is not a probability distribution. It does not imply that each outcome is equally likely, or that the highest row represents a credible upside case.
Sensitivity answers, “What happens if this input changes?” Evidence is needed to assess which values are plausible.
Turn Low-Confidence Assumptions Into Defined Tests
“Improve conversion” is an objective, not a testable explanation.
A more useful hypothesis might be:
“For the specified inbound segment, the revised qualification and follow-up process will increase lead-to-customer conversion enough to justify its additional operating cost.”
Define the test before launch:
| Test element | What to specify |
|---|---|
| Population | Which leads are eligible |
| Intervention | What changes and what remains constant |
| Comparison | How the result will be evaluated against an appropriate reference |
| Primary outcome | The relevant commercial result |
| Economic threshold | The minimum improvement needed to justify the cost |
| Guardrails | Margin, customer quality, sales workload, and retention |
| Observation window | Time needed for outcomes to mature |
| Evidence requirement | Sample and analytical requirements appropriate to the decision |
| Decision rule | Expand, revise, stop, or continue gathering evidence |
A positive result should update the specific assumption tested. It should not automatically validate every segment, channel, or spending level.
An inconclusive result should remain inconclusive.
Keep Targets, Forecasts, and Actuals Visible
A useful dashboard should show separate columns for:
- Original target.
- Current forecast.
- Actual performance to date.
- Remaining requirement.
- Evidence status.
For example, retain the 5% conversion target even if the current evidence-led forecast remains at 4%. The difference is a management issue to address—not an inconsistency to hide.
Do not overwrite the original plan whenever expectations change. Preserve versions so the team can understand:
- Which assumptions changed.
- What evidence prompted the revision.
- How the revision affected revenue and cost.
- Which decisions followed.
This enables honest reforecasting without silently moving the goalposts.
Update Confidence as Well as the Number
When new data arrives, review both the input value and the evidence supporting it.
A pilot might produce 5% conversion but still leave confidence low because the cohort is small or unusually selected. A larger, comparable cohort may support a 4.3% estimate with greater confidence.
The higher number is not necessarily the better forecasting input.
Use a consistent review sequence:
- Verify definitions and data quality.
- Check cohort maturity.
- Assess comparability with the planned activity.
- Update the assumption where justified.
- Recalculate dependent outcomes.
- Revise the confidence label and limitations.
- Record the resulting decision.
The register should evolve alongside the financial model.
A Forecast Should Explain Its Own Limits
A transparent funnel forecast should allow a reader to identify:
- What management wants to achieve.
- What has actually been observed.
- What is assumed about the future.
- Which outputs depend on unvalidated improvements.
- What evidence would justify a revision.
- Which commitments are exposed if an assumption fails.
In a scenario tool such as PlotGTM, selecting a KPI value can help explore a possible outcome. It should not make that outcome appear proven.
The most useful forecast is not the one that most closely matches the target. It is the one that makes the distance between ambition and evidence clear enough for the business to act.