# Get Better Results with Mo Mo is your AI planning assistant inside Moselle — built to help you make faster, smarter inventory and demand planning decisions. This guide (10-minute read) will show you how to get the most out of every conversation. {% hint style="info" %} **Quick Answer:** Write specific prompts that name the SKUs, channel, and time window you care about. Use Mo's prompt template: \[What you want] + \[SKUs/channels/dates] + \[Extra context] + \[Output format]. Save your best prompts as Favourites for one-click reuse. {% endhint %} ## The Golden Rule **Specific prompts get specific answers. Vague prompts get generic ones.** Mo is not a search engine — it's a reasoning agent. Give it a clear job to do, and it will do it well. | Instead of… | Try… | | -------------------------------- | --------------------------------------------------------------------------------------------------------------- | | *"How's our inventory looking?"* | *"How many weeks of cover do we have on our top 10 SKUs across Shopify and Amazon, based on current forecast?"* | | *"Which SKUs are at risk?"* | *"Which SKUs are at risk of stocking out before the end of Q2, across our Shopify and Amazon channels?"* | | *"What should I order?"* | *"What should I order for our wholesale channel, assuming a 10-week lead time from our main supplier?"* | *** ## What Mo Can (and Can't) Do **Mo can:** * Analyze your sales, inventory, forecast, BOM, and supplier data * Draft purchase orders and replenishment plans * Model "what if" scenarios * Generate charts, tables, and Excel exports * Walk you through its reasoning step by step **Mo can't:** * Remember previous conversations — every session starts fresh * See data that hasn't been entered into Moselle (e.g. an unannounced lead time change) * Access the internet or live market data {% hint style="info" %} **Keep your Moselle data clean and up to date.** It's the single biggest factor in the quality of Mo's answers. {% endhint %} ### How Mo works through a question When you send Mo a message, it moves through a few stages before responding. **1. It identifies what you're asking**\ Mo reads your question and determines the business concept — stockout risk, sell-through rate, reorder quantity, demand trend — along with the scope: which SKUs, which channel, which time period. If something is ambiguous, Mo will make a reasonable assumption and tell you what it assumed, or ask a clarifying question if it can't resolve it on its own. **2. It pulls from your connected data**\ Mo works from the data Moselle has access to: your sales history, current inventory levels, SKU catalog, supplier lead times, and channel breakdown. It draws on whichever of these is relevant to your question and cross-references across them to build its answer. **3. It checks whether the answer makes sense**\ Before responding, Mo evaluates whether the result is internally consistent — if a number looks anomalous, it will flag it rather than presenting it as fact. Mo will tell you when it's uncertain and explain why. **4. It responds and shows its reasoning**\ Mo formats its answer based on what you asked for — a recommendation, a number, an analysis, a ranked list — and explains the logic behind it so you can decide whether to act on it or adjust it. ### What Mo can see Mo's answers are only as good as the data behind them. It can work with: * **Sales history** — order data synced from your connected channels (Shopify, Amazon, Walmart) * **Inventory levels** — current stock positions across locations, synced from your fulfillment integrations * **SKU catalog** — your products, variants, bundles, and bill of materials * **Supplier data** — lead times and supplier details you've added to Moselle * **Forecasts** — the 12-month projections Moselle has generated for your SKUs * **Current conversation** — everything you've said in this session, including context you've added along the way * **Files you upload** — CSV, XLSX, and PDF files shared directly in the chat What Mo can't see: * **Information not in Moselle** — if a supplier lead time hasn't been entered, or a channel isn't connected, Mo won't know. It may make a reasonable assumption, but it will say so. * **Previous conversations** — Mo starts fresh each session. Context you established in a past conversation doesn't carry over. If there's something Mo should always know — a planning assumption, a key definition — the best place for it is in your Moselle setup, not a previous chat. * **External signals** — Mo doesn't browse the web or pull in market data. Its view is your business, not the broader market. *** ## Part 1 — How to Write a Good Prompt ### The Prompt Template When in doubt, use this structure: > **\[What you want Mo to do] + \[SKUs / channels / dates] + \[Extra context] + \[Output format]** **Example:** > *"Show me sell-through for our core skincare collection across Shopify and Amazon for the last 90 days, as a table ranked by sell-through rate."* ### The Five Fundamentals **1. Be specific**\ Name the SKUs, channel, and time window. The more precise you are, the more useful the answer. **2. Add context Mo can't see**\ Mention anything outside of Moselle — an upcoming promotion, a factory closure, a new retail partner, a lead time change. **3. Ask for the format you want**\ "As a table," "as a bar chart," "as an Excel file," "in three bullet points." If you don't specify, Mo will choose for you. **4. Ask one thing at a time**\ Combining five questions in one prompt gets shallow answers to all five. Use follow-up prompts instead. **5. Refine — don't restart**\ If the first answer isn't quite right, correct it in the same conversation. Starting over loses all the context Mo has built up. ### Use the Right Language Mo responds best when you speak in planning terms: SKU, channel, sell-through, weeks of supply (WOS), forward weeks of supply (FWOS), reorder point (ROP), safety stock, lead time, MOQ, coverage period. If a term has a specific meaning at your company, define it upfront: > *"For this conversation, define 'slow mover' as any SKU with WOS above 26."* *** ## Part 2 — Tips for Better Results ### Re-state context every session Mo does not remember previous conversations. At the start of each session, briefly re-share the context that matters — which channel you're focused on, any active promotions, your standard lead times. ### Save your best prompts Hover over any Mo response and click ☆ to save it as a Favourite. Reuse it in one click next time. This is the fastest way to get consistent results across sessions. {% content-ref url="/pages/DAsnRIRTbbjM6bgR0RLM" %} [Save Your Favourite Prompts](/mo/tips/save-favourite-prompts.md) {% endcontent-ref %} ### Ask Mo to show its work For any complex answer, ask Mo to explain its reasoning. This helps you catch errors and builds confidence in the output. > *"Walk me through your reasoning step by step before giving me the final answer."* Mo's **Explain This** button on any white forecast cell does this automatically. ### Upload files for context Mo can't see Mo accepts CSV, XLSX, and PDF files via the 📎 icon or drag-and-drop. Use this to share promotional calendars, supplier lead time updates, retail set dates, and manual targets. {% hint style="warning" %} Don't upload files containing customer PII or confidential internal data. Use mock or anonymised data where possible. {% endhint %} {% content-ref url="/pages/bhnbOdyzHJU16Tc1oHKc" %} [Upload Your Files to Mo](/mo/tips/upload-your-files-to-mo.md) {% endcontent-ref %} ### Set a role for better outputs Telling Mo who it's acting as at the start of a conversation consistently improves results: > *"Act as a senior demand planner reviewing my Q4 forecast for risk."*\ > \&#xNAN;*"You're an inventory analyst preparing a summary for our leadership team."* ### Run scenarios in threes For any planning decision, ask Mo to model three versions — conservative, moderate, and aggressive. This forces a range of outcomes and surfaces risk on both ends. > *"Create three BFCM scenarios: conservative (20% lift), moderate (40% lift), and aggressive (60% lift)."* ### Always verify before acting Mo can occasionally produce confident-sounding answers that are wrong. Build the habit of checking: * Spot-check at least one number against the underlying SKU card or a Moselle report * Review all line items before confirming a PO * For large or high-stakes decisions, have a second person review before committing *** ## Part 3 — Prompt Examples by Use Case These are starting points — swap in your own SKUs, channels, dates, and targets. **Sell-through & trends** > *"How has sell-through for \[collection] trended over the last \[X] days, by channel — as a bar chart?"* **At-risk SKUs & stockouts** > *"Which SKUs are at risk of stocking out before \[date/event]? Include projected stock-out date and current WOS."* **Weeks of supply** > *"How many weeks of cover do we have on \[SKUs / category] based on current forecast?"* **Reorder points** > *"Which SKUs are currently below their reorder point, and what do I need to order to get back on track?"* **Purchase orders** > *"Create a PO for \[X units] of \[SKU] from \[supplier], deliver by \[date], in \[currency]."*\ > \&#xNAN;*"I have \[$X] to spend with \[supplier] — what should I order to maximize coverage on at-risk SKUs?"* **Promotions** > *"We have a \[X%] off promotion on \[SKUs] starting \[date] — do we have enough inventory to support a \[X]x demand spike?"* **Seasonality & event planning** > *"Adjust my \[season] forecast to account for \[event/factor] and tell me what I need to order and by when."* **Scenarios** > *"What happens to my replenishment plan if my supplier delays by \[X] weeks?"* **Forecast accuracy** > *"How accurate has my forecast been for \[category] over the last \[X] months? Use MAPE and flag anything above 30%."* {% hint style="info" %} **MAPE benchmarks:** Under 10% = highly accurate | 10–20% = good | 20–30% = reasonable | Over 30% = needs attention {% endhint %} *** ## Part 4 — Common Mistakes to Avoid | Mistake | Why it matters | What to do instead | | --------------------------------------- | ---------------------------------------------- | -------------------------------------- | | Prompt is too vague | Mo guesses at scope and gives a generic answer | Name SKUs, channels, and a time window | | Asking multiple questions at once | Shallow answers across the board | Break into follow-up prompts | | Restarting when something's off | You lose all the context Mo built | Refine in the same conversation | | Not checking the output | Mo can produce wrong numbers confidently | Verify key figures before acting | | Re-typing the same prompt every session | Wastes time, inconsistent results | Save it as a Favourite | | Adding jargon to sound more specific | Overcomplicating actually hurts results | Clear, plain instructions work best | | Uploading sensitive or personal data | Data privacy risk | Anonymize files before uploading | *** ## Part 5 — Building Good Habits ### If you're just getting started * Use the prompt template every time until it's second nature * Start with three workflows and get comfortable: a weekly sell-through check, an at-risk SKU review, and a PO draft * Verify Mo's numbers against one source for your first few weeks — build trust through evidence * Save your three most-used prompts as Favourites right away ### If you're a regular Mo user * Build a prompt library of 10–20 prompts that cover your monthly planning cadence * Use file uploads to bring in context from outside Moselle — promo calendars, supplier updates, retail timelines * Default to asking for step-by-step reasoning on any multi-step question * Always run scenarios in threes before making a big inventory call * Define your own business terms at the start of complex conversations ### When to pause and double-check * Mo flags a number as anomalous → investigate before acting * A recommended PO is above your internal review threshold → get a second sign-off * Forecast MAPE is above 30% for a priority SKU → investigate inputs before locking * Mo's recommendation contradicts your instinct → ask it to explain step by step; if the logic doesn't hold, trust your judgment *** ## Why This Matters Getting good at working with Mo is a skill with a real payoff. McKinsey research on AI-driven supply chain forecasting shows companies using AI effectively can reduce forecast errors by 20–50%, cut stockouts by up to 65%, and lower planning administration costs by 25–40%. Gartner predicts 70% of large organizations will adopt AI-based forecasting by 2030. The brands that pull ahead won't just be the ones with the best tools — they'll be the ones whose teams know how to use them. *** ## Related Guides {% content-ref url="/pages/DAsnRIRTbbjM6bgR0RLM" %} [Save Your Favourite Prompts](/mo/tips/save-favourite-prompts.md) {% endcontent-ref %} {% content-ref url="/pages/bhnbOdyzHJU16Tc1oHKc" %} [Upload Your Files to Mo](/mo/tips/upload-your-files-to-mo.md) {% endcontent-ref %} --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://learn.moselle.io/mo/tips/tips-for-better-prompts.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.