How to Preserve Units, Currencies, and Date Ranges in an AI Brief

Summary

• Units, currencies, and date ranges get distorted when an AI brief leaves room for interpretation (symbols, locale, time zone, inclusive/exclusive ranges).
• Fix ambiguity by declaring a single “measurement and time contract” at the top of every brief: units, currency, FX rule, date format, time zone, and range semantics.
• Use a small, reusable data block (copy/paste) that forces the model to keep the same unit system and date boundaries throughout.
• Add targeted checks: unit consistency, currency symbol vs ISO code, FX date, rounding, and whether date ranges are inclusive.
• Save your contract + checks as a reusable prompt so every new project starts with the same guardrails.

When an AI output “looks right” but quietly changes units (miles vs km), currencies (USD vs CAD, symbol confusion), or date ranges (inclusive vs exclusive, time zone drift), the problem is rarely the model’s math. It’s the brief: it didn’t lock down the measurement and time assumptions.

This guide shows the specific ambiguity patterns that cause unit/currency/date-range drift, then gives you a reusable “contract” format, a before-and-after example, and a set of targeted checks you can run on every response.

Why AI briefs lose units, currencies, and date ranges

AI systems fill in missing assumptions. If your brief doesn’t explicitly define the assumptions, the model may choose defaults that differ from yours. The most common ambiguity triggers are:

• Unit system not specified: “distance,” “weight,” “temperature,” “volume,” “revenue per user” without stating metric/imperial or the exact unit (km vs miles; kg vs lb; C vs F).
• Currency symbol ambiguity: “$” could be USD, CAD, AUD, SGD, etc. “£” could be GBP or other pound-denominated contexts. Even “kr” varies.
• FX conversion rule missing: If you ask for conversions but don’t specify the exchange rate source/date, the model may invent a rate or assume a generic one.
• Date format ambiguity: 03/04/2026 can be March 4 or April 3 depending on locale.
• Date range semantics missing: “From June 1 to June 30” can be interpreted as inclusive, exclusive, end-of-day, or “through June 30 00:00.”
• Time zone not specified: “last week,” “this month,” or “Q2” can shift when the model assumes a different time zone or uses UTC boundaries.
• Rounding and precision not specified: “Round to 2 decimals” vs “nearest whole unit,” and whether to round intermediate steps.

Before-and-after: the same request, with and without a “measurement and time contract”

Before (ambiguous brief)

Prompt: “Summarize our ad performance from 03/04/2026 to 04/03/2026. Include spend, CPA, and distance-based delivery radius. Convert spend to dollars and show totals.”

What can go wrong:

• The model may interpret 03/04/2026 as March 4 (US) or 3 April (international).
• “Dollars” may be assumed as USD even if your business reports in CAD or AUD.
• “Distance-based radius” may be output in miles when you operate in kilometers (or vice versa).
• Totals may be computed across the wrong date window (swapped start/end) or with unclear inclusivity.

After (unambiguous brief with a contract block)

Prompt: “Summarize ad performance for the date range below. Follow the Measurement & Time Contract exactly and do not change units/currency/date interpretation.”

Result: You’ve removed the model’s need to guess. Even if the model still makes a mistake, you now have explicit criteria to catch it quickly.

A reusable format you can drop into any AI brief

The goal is to make the “contract” small enough to reuse, but strict enough to prevent drift. Use this template at the top of your brief (or right before the task instructions):

Targeted checks to prevent silent drift (copy/paste checklist)

After the AI responds, run these checks. You can also instruct the AI to self-check against them and report any conflicts.

1) Unit checks

• Unit presence: Every numeric quantity that depends on a unit should show the unit (e.g., “12 km,” not “12”).
• Unit consistency: The same concept uses one unit throughout (no mixing km and miles in the same table unless requested).
• Temperature and thresholds: If you have thresholds (e.g., “above 30”), confirm whether that is 30 C or 30 F.
• Derived metrics: Confirm the unit of rates (e.g., “GBP/day,” “kg per shipment,” “km per delivery”).

2) Currency checks

• ISO code vs symbol: Require “USD 1,200.00” rather than “$1,200” when ambiguity matters.
• Conversion rule: If conversion appears, confirm the exchange rate and the rate date were provided (or the model asked for them).
• Rounding: Confirm money rounding is consistent (2 decimals is common, but your finance rules may differ).
• Totals: Ensure totals are in the same currency as line items (no mixed-currency summation).

3) Date range checks

• Format: Dates are in ISO (YYYY-MM-DD) to avoid locale confusion.
• Boundary inclusion: Confirm whether the end date is included and what “included” means (end-of-day in the specified time zone vs midnight).
• Time zone: Confirm the time zone used for “day” boundaries and any timestamps.
• Range direction: Start date is before end date; no accidental swap.
• Relative dates: Avoid “last 30 days” unless you also specify the anchor date and time zone.

Practical examples you can reuse

Example A: Marketing performance brief (currency + date range)

Prompt:

“Create a performance summary and a table of daily spend and CPA. Use the contract below. If any input conflicts with the contract, ask one clarification question before proceeding.”

• DATE_FORMAT: YYYY-MM-DD
• TIME_ZONE: America/Los_Angeles
• DATE_RANGE: 2026-06-01 to 2026-06-30
• RANGE_SEMANTICS: Inclusive start and inclusive end (whole days)
• CURRENCY: USD (use “USD”)
• ROUNDING: Money 2 decimals; CPA 2 decimals

Example B: Logistics brief (units + derived metrics)

Prompt:

“Rewrite this ops update into a structured report. Keep all distances in km and weights in kg. Do not convert to miles/lb. If you see miles/lb in the input, keep the original value and add a note that it conflicts with the contract.”

• UNITS: Metric (km, kg)
• CONSISTENCY_RULE: No unit conversion unless explicitly requested

Example C: Finance brief (currency + FX rule)

Prompt:

“Summarize revenue by region. Present in EUR. If any source numbers are in other currencies, ask me for the FX rate and FX date; do not invent rates.”

• CURRENCY: EUR
• FX_RULE: Ask for FX rate + date; do not invent

How to instruct the AI to self-audit (without turning your brief into a novel)

You can add a short “output contract” section that forces the model to surface assumptions and conflicts:

• Assumptions line: “At the top, list the contract values you used (currency, units, time zone, date range).”
• Conflicts line: “If any input conflicts with the contract, list it under ‘Conflicts’ and do not silently convert.”
• Checks line: “Before finalizing, verify: ISO dates, currency ISO code present, and no unit mixing.”

Preserving these details across repeated work with CopyCharm

If you write briefs frequently, the hard part is not creating a good contract once - it is reusing the same contract reliably across new chats, new documents, and new stakeholders without retyping it or accidentally editing it.

A concrete workflow: save, find, reuse

• Save: Copy your “Measurement & Time Contract” block and your “Targeted Checks” checklist, then save them as Saved Prompts in CopyCharm (separately from favoriting ordinary copied clips).
• Find: When you start a new brief, search your past clips or open your saved prompt in CopyCharm to retrieve the exact contract text you used last time.
• Reuse: Paste the contract into your AI brief, then fill in only the variables (time zone, date range, currency). For Claude, Gemini, Cursor, email, documents, and other apps, the workflow is manual: retrieve in CopyCharm, then copy/paste into the destination.

When ChatGPT access matters: authenticated connector vs manual reuse

If you want ChatGPT to pull your saved contract text without you manually pasting it, CopyCharm offers an authenticated ChatGPT connector backed by optional AI Access sync. After you sign in with an eligible active CopyCharm purchase, authorize the CopyCharm Desktop connection, enable and complete AI Access sync, and authorize the ChatGPT connector, ChatGPT can search and retrieve supported Synced Data (such as Favorite Clips, Saved Prompts, and optional Other Clips within your selected time range). ChatGPT cannot access unsynced local CopyCharm data.

This can be useful when you want to say: “Find my ‘Measurement and Time Contract’ saved prompt and insert it,” then proceed with the rest of the brief.

Try it if you want a repeatable contract you can quickly reuse across projects: CopyCharm

Common pitfalls (and how to fix them fast)

Using symbols instead of ISO codes

Pitfall: “$” in a multi-country context.
Fix: Require ISO codes in outputs: “USD”, “CAD”, “AUD”.

Asking for “last month” without an anchor

Pitfall: “Summarize last month” when “last month” depends on today’s date and time zone.
Fix: Provide explicit ISO start/end dates and a time zone.

Mixing “whole days” with timestamps

Pitfall: “June 1 to June 30” plus a dataset with timestamps in UTC.
Fix: State whether you want whole days in a named time zone, or exact timestamps with offsets.

Letting the model convert units “helpfully”

Pitfall: “Distance is 10” and the model decides it is miles.
Fix: Add “Do not convert or infer units; if missing, ask.”

Frequently Asked Questions

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Back to FAQ Table of Contents

Related Guides

• What Is an AI Context Pack?
• How to Prepare Better Context for ChatGPT
• How to Organize Copied Text for AI Work
• Source-Labeled Context for AI: Why It Matters
• ChatGPT Context Pack for Consultants