Forecast next quarter 3 different ways.
Your forecast says one thing. The sales pipeline says another. Last year’s pattern suggests a third. Ask AI to calculate each view properly, then find out what you would have to believe for each one to be right.
Try it now
Copy the prompt into your AI tool of choice, then answer its questions as they come.
Paste it into any AI chat to start. For most people, this is all you need.
Everything on this page in one paste: the prompt plus the full plan, every step and what to check. Use it for longer moves, to plan ahead, to share with a colleague, or if your AI can't open links.
How it goes
What the AI will do
- Finds your company’s forecast methods and underlying data, then researches useful alternatives.
- Calculates next quarter three different ways, checks past performance where the records allow, and investigates why the numbers disagree.
What to have ready
- The business or metric you care about, if it is not already clear. Existing finance files or exports help when the AI cannot reach the records.
Where you come in
- Confirm what the forecast should cover and the decision it will inform.
- Fill consequential gaps the AI cannot resolve, then review the estimates and their assumptions.
What you'll get
Three comparable forecasts, an explanation of their differences, and inspectable calculations you can update—with the limits of the evidence made clear.
You're done when
You have three calculated next-quarter forecasts using distinct defensible methods, their key assumptions and disagreements, available historical checks, and the source-linked calculations. Any agreed narrower result is identified explicitly; a data request or three uncalculated outlines is not the completed forecast.
These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result. ### Start with the activity and the user's context Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear. Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them. ### Check the capabilities needed for this work Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit. When useful access is missing: - Explain what you cannot reach and how that material would help. - Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot. - If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response. - Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available. ### Ask focused questions and respect the answers Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful. Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements. ### Keep evidence and independent contributions distinct Distinguish evidence, inference, and missing information. Check your work against the source material. Use independent agents when your system actually supports delegation and independent investigation or review would improve the work: - Give each agent a bounded brief and the context appropriate to its role. - Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations. - Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately. ### Produce, check, and revise the result Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them. Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result. Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized. ### Find the question and the company’s forecasting habits Help me forecast next quarter three genuinely different ways, then explain what their disagreement tells us. Start by inspecting the information and tools you can actually access. Search relevant finance folders, actuals, previous forecast workbooks, budget packs, board materials, operating plans, CRM pipeline snapshots, and planning emails, chats and meeting transcripts. Follow project names, business units, forecast owners and file links to the underlying records. Look for forecast-versus-actual reviews, model formulas, planning calendars and explanations of why people adjusted a number. Distinguish the method actually used from a named software product or an aspirational planning policy. Use these records to identify the likely metric, fiscal-quarter dates, organizational boundary, currency, accounting basis, available history and decision the forecast serves. Find the people who consume it and their documented concerns: a cash forecast for treasury and a bookings forecast for a sales leader answer different questions. Cite those concerns rather than inventing stakeholder preferences. Distinguish targets, budgets, management commitments and forecasts of what is likely to happen. Present a short account of what you found and ask one focused question at a time about consequential unresolved choices. Confirm the target and boundary before estimating. Search first; do not ask me to assemble documents you can retrieve. If access fails, investigate the available connection or export route, then request the precise missing item and explain what it enables—for example, dated monthly actuals and frozen prior-quarter pipeline snapshots. Accept permitted CSVs, spreadsheets or pasted tables. If a read-only connection would unlock the work, offer to draft the specific IT request. An access request alone is not three forecasts. ### Prepare one evidence base and choose three methods Build a dated input table with sources, units, reporting periods, missing values, revisions and the cutoff for information known today. Reconcile totals against their source. Identify seasonality, one-off events, acquisitions, changed definitions and historical breaks; retain raw values and document every adjustment. Never silently mix bookings with recognized revenue or monthly with cumulative figures. Size the work to a meaningful business unit or metric before expanding a large corpus. Search reputable forecasting research and original technical documentation for methods suited to this metric, horizon and available data, including productive alternatives to company practice. Explain the practical difference between your three choices and why each is feasible. Ordinarily include the company’s established approach as one of the three when you can reproduce it; distinguish an approximation when you cannot. The other methods should supply different information or reasoning, not simply optimistic, central and pessimistic assumptions in the same model. Useful starting points, to adapt after inspecting the data: - A historical-series approach estimates future values from observed level, trend and seasonality. A same-season-last-year or latest-value forecast supplies a transparent benchmark; use a more elaborate fitted model only when the history supports it. [Simple forecasting methods](https://otexts.com/fpp3/simple-methods.html). - A business-driver approach builds the result from operational quantities and relationships: retained customers × revenue per customer, plus new business; or scheduled work × realizable price, constrained by delivery capacity. Establish the actual company formula, timing and source of each driver. If estimating relationships statistically, use predictors available at forecast time or explicitly forecast their future values. [Forecasting with regression](https://otexts.com/fpp3/forecasting-regression.html). - Another distinct approach might aggregate validated deal-level or business-unit forecasts, model cohorts, or use a researched outside demand indicator. Pick the approach that provides a defensible third view in this business. Avoid dressing the same spreadsheet in three names. If the data cannot support three defensible estimates, explain exactly why and ask for the missing material or agreement to a narrower comparison. Do not fabricate historical observations or parameter estimates to satisfy the title. ### Produce the forecasts independently When subagents are available, assign one analyst per method. Give each the confirmed target, common data snapshot, source references, method, required outputs and explicit cutoff. Do not show them one another’s forecast before they finish. Divide very large inputs by relevant business unit or period, track coverage and reconcile their outputs before aggregation. Run successive waves if capacity is limited. Without subagents, use separate bounded passes or fresh sessions with the same handoff; disclose shared context rather than claiming independence. Each analyst must implement its calculation using available code or spreadsheet tools, return next-quarter totals and useful within-quarter detail, preserve formulas or executable code, identify the assumptions that move its result most, and explain its uncertainty. Use sourced estimates for conversion, churn, price, lead time or capacity; mark judgmental replacements explicitly. Check that components reconcile and that constraints are respected. A prose description of a model is not a calculated forecast. Where statistical prediction intervals are available, explain their meaning and assumptions. Where only sensitivity cases are defensible, label them as scenarios. In particular, a regression interval conditional on assumed future drivers does not account for uncertainty in those drivers; vary them separately and show their contribution to uncertainty. Do not invent probabilities from the agreement of three agents. ### Test past performance and explain divergence Where dated history permits, replay each method at prior forecasting cutoffs, using only information available then. Evaluate the same next-quarter horizon and target for all three, ideally across several historical quarters. Do not use later CRM outcomes, revised budgets or future driver values to improve a simulated past forecast. [Time series cross-validation](https://otexts.com/fpp3/tscv.html) describes this rolling-cutoff comparison. Report the number of usable test periods, errors in understandable business units, systematic over- or underprediction and performance against the simple benchmark. With too little history, say that accuracy has not been established. Have a separate reviewer inspect the calculations, source table and results without the analysts’ defense of them. Ask it to find leakage, duplicated components, incompatible definitions and unjustified certainty. Repair material findings. If independent review is unavailable, make a separate check and disclose the limitation. Compare the estimates side by side. Trace important differences to concrete assumptions or information: seasonality, pipeline slippage, renewal timing, pricing, delivery limits. Recalculate the largest differences under common assumptions to show what explains the gap. Do not average away disagreement automatically. Explain which forecast is most useful for the confirmed decision, what the evidence supports and what would change your judgment. ### Deliver a forecast the reader can inspect Lead with the three numbers or ranges, their methods and the main reason they differ. Provide the comparison, historical checks, material assumptions, source links, remaining gaps and the underlying runnable calculation or inspectable formula table. Explain how to refresh the inputs next quarter. Stop with this analysis; do not replace the official forecast or distribute it to stakeholders. These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.
View full AI instructions
Forecast next quarter 3 different ways.
What the AI will do
- Finds your company’s forecast methods and underlying data, then researches useful alternatives.
- Calculates next quarter three different ways, checks past performance where the records allow, and investigates why the numbers disagree.
What you can provide
- The business or metric you care about, if it is not already clear. Existing finance files or exports help when the AI cannot reach the records.
Where you'll take part
- Confirm what the forecast should cover and the decision it will inform.
- Fill consequential gaps the AI cannot resolve, then review the estimates and their assumptions.
What you'll get
Three comparable forecasts, an explanation of their differences, and inspectable calculations you can update—with the limits of the evidence made clear.
Activity context
Your forecast says one thing. The sales pipeline says another. Last year’s pattern suggests a third. Ask AI to calculate each view properly, then find out what you would have to believe for each one to be right.
For the user
Start with a metric, business unit or forecast you already use; you do not need to prepare a data packet first. The AI will search what it can access, establish your company’s current method and ask for missing material. You will confirm the scope before it runs the comparison. If the evidence cannot support three forecasts, it will explain what is missing and agree a smaller exercise with you.
Guidance for the AI
These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result.
Start with the activity and the user's context
Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear.
Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them.
Check the capabilities needed for this work
Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit.
When useful access is missing:
- Explain what you cannot reach and how that material would help.
- Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot.
- If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response.
- Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available.
Ask focused questions and respect the answers
Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful.
Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements.
Keep evidence and independent contributions distinct
Distinguish evidence, inference, and missing information. Check your work against the source material.
Use independent agents when your system actually supports delegation and independent investigation or review would improve the work:
- Give each agent a bounded brief and the context appropriate to its role.
- Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations.
- Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately.
Produce, check, and revise the result
Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them.
Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result.
Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized.
Your AI prompt
Find the question and the company’s forecasting habits
Help me forecast next quarter three genuinely different ways, then explain what their disagreement tells us. Start by inspecting the information and tools you can actually access. Search relevant finance folders, actuals, previous forecast workbooks, budget packs, board materials, operating plans, CRM pipeline snapshots, and planning emails, chats and meeting transcripts. Follow project names, business units, forecast owners and file links to the underlying records. Look for forecast-versus-actual reviews, model formulas, planning calendars and explanations of why people adjusted a number. Distinguish the method actually used from a named software product or an aspirational planning policy.
Use these records to identify the likely metric, fiscal-quarter dates, organizational boundary, currency, accounting basis, available history and decision the forecast serves. Find the people who consume it and their documented concerns: a cash forecast for treasury and a bookings forecast for a sales leader answer different questions. Cite those concerns rather than inventing stakeholder preferences. Distinguish targets, budgets, management commitments and forecasts of what is likely to happen.
Present a short account of what you found and ask one focused question at a time about consequential unresolved choices. Confirm the target and boundary before estimating. Search first; do not ask me to assemble documents you can retrieve. If access fails, investigate the available connection or export route, then request the precise missing item and explain what it enables—for example, dated monthly actuals and frozen prior-quarter pipeline snapshots. Accept permitted CSVs, spreadsheets or pasted tables. If a read-only connection would unlock the work, offer to draft the specific IT request. An access request alone is not three forecasts.
Prepare one evidence base and choose three methods
Build a dated input table with sources, units, reporting periods, missing values, revisions and the cutoff for information known today. Reconcile totals against their source. Identify seasonality, one-off events, acquisitions, changed definitions and historical breaks; retain raw values and document every adjustment. Never silently mix bookings with recognized revenue or monthly with cumulative figures. Size the work to a meaningful business unit or metric before expanding a large corpus.
Search reputable forecasting research and original technical documentation for methods suited to this metric, horizon and available data, including productive alternatives to company practice. Explain the practical difference between your three choices and why each is feasible. Ordinarily include the company’s established approach as one of the three when you can reproduce it; distinguish an approximation when you cannot. The other methods should supply different information or reasoning, not simply optimistic, central and pessimistic assumptions in the same model.
Useful starting points, to adapt after inspecting the data:
- A historical-series approach estimates future values from observed level, trend and seasonality. A same-season-last-year or latest-value forecast supplies a transparent benchmark; use a more elaborate fitted model only when the history supports it. Simple forecasting methods.
- A business-driver approach builds the result from operational quantities and relationships: retained customers × revenue per customer, plus new business; or scheduled work × realizable price, constrained by delivery capacity. Establish the actual company formula, timing and source of each driver. If estimating relationships statistically, use predictors available at forecast time or explicitly forecast their future values. Forecasting with regression.
- Another distinct approach might aggregate validated deal-level or business-unit forecasts, model cohorts, or use a researched outside demand indicator. Pick the approach that provides a defensible third view in this business. Avoid dressing the same spreadsheet in three names.
If the data cannot support three defensible estimates, explain exactly why and ask for the missing material or agreement to a narrower comparison. Do not fabricate historical observations or parameter estimates to satisfy the title.
Produce the forecasts independently
When subagents are available, assign one analyst per method. Give each the confirmed target, common data snapshot, source references, method, required outputs and explicit cutoff. Do not show them one another’s forecast before they finish. Divide very large inputs by relevant business unit or period, track coverage and reconcile their outputs before aggregation. Run successive waves if capacity is limited. Without subagents, use separate bounded passes or fresh sessions with the same handoff; disclose shared context rather than claiming independence.
Each analyst must implement its calculation using available code or spreadsheet tools, return next-quarter totals and useful within-quarter detail, preserve formulas or executable code, identify the assumptions that move its result most, and explain its uncertainty. Use sourced estimates for conversion, churn, price, lead time or capacity; mark judgmental replacements explicitly. Check that components reconcile and that constraints are respected. A prose description of a model is not a calculated forecast.
Where statistical prediction intervals are available, explain their meaning and assumptions. Where only sensitivity cases are defensible, label them as scenarios. In particular, a regression interval conditional on assumed future drivers does not account for uncertainty in those drivers; vary them separately and show their contribution to uncertainty. Do not invent probabilities from the agreement of three agents.
Test past performance and explain divergence
Where dated history permits, replay each method at prior forecasting cutoffs, using only information available then. Evaluate the same next-quarter horizon and target for all three, ideally across several historical quarters. Do not use later CRM outcomes, revised budgets or future driver values to improve a simulated past forecast. Time series cross-validation describes this rolling-cutoff comparison. Report the number of usable test periods, errors in understandable business units, systematic over- or underprediction and performance against the simple benchmark. With too little history, say that accuracy has not been established.
Have a separate reviewer inspect the calculations, source table and results without the analysts’ defense of them. Ask it to find leakage, duplicated components, incompatible definitions and unjustified certainty. Repair material findings. If independent review is unavailable, make a separate check and disclose the limitation.
Compare the estimates side by side. Trace important differences to concrete assumptions or information: seasonality, pipeline slippage, renewal timing, pricing, delivery limits. Recalculate the largest differences under common assumptions to show what explains the gap. Do not average away disagreement automatically. Explain which forecast is most useful for the confirmed decision, what the evidence supports and what would change your judgment.
Deliver a forecast the reader can inspect
Lead with the three numbers or ranges, their methods and the main reason they differ. Provide the comparison, historical checks, material assumptions, source links, remaining gaps and the underlying runnable calculation or inspectable formula table. Explain how to refresh the inputs next quarter. Stop with this analysis; do not replace the official forecast or distribute it to stakeholders.
Done when
You have three calculated next-quarter forecasts using distinct defensible methods, their key assumptions and disagreements, available historical checks, and the source-linked calculations. Any agreed narrower result is identified explicitly; a data request or three uncalculated outlines is not the completed forecast.
These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.
These instructions were written in September 2026 using the frontier models available at the time. You may now have better tools or models, or restrictions that make some tasks harder. Your AI should adapt to the tools and access you actually have and find the best available way to help. Everything here is a starting suggestion for you and your AI, not a proven best practice for every setup. Feel free to rewrite any of it to better meet your needs.