Why does Sherpa matter?
Most ecommerce brands don't have a data problem. They have a translation problem. The CMO wants to know what's actually making money. The CFO wants a product-level P&L before quarter close. The buying team wants a reorder list. None of them write SQL. None of them want to wait a week for an analyst to hand back a static chart that's out of date by the time it arrives.
Sherpa removes that translation step. Compare asking a librarian to find you a book with asking a research assistant who already knows your business and just gives you the answer.
What is Sherpa?
Sherpa is a multi-agent AI system built into ProfitPeak. "Multi-agent" means it's made of specialist agents that work together on a single question. There are separate agents for attribution, product, inventory, customer, and creative data. This is how one prompt can connect ad performance to margin to stock levels.
Sherpa works directly on your brand's own finance, product, inventory, and marketing data. It isn't a generic chatbot with your numbers added on the side. Your data is the foundation it answers from.
Sherpa is less like a search bar and more like an analyst who has already been onboarded. It knows your SKUs, your margin structure, and your ad accounts, so you never have to explain context before you ask.
What can you do with Sherpa?
Capability | What it means in practice |
Plain-language Q&A | Ask a real question and get answers and dashboards in minutes, without long wait times. |
Cross-functional answers | One prompt can combine attribution, inventory, and margin. For example: "Which Meta ads are driving contribution profit, and do we have stock to scale them?" (Contribution profit is what's left after all variable costs, including shipping and ad spend.) |
Widget creation | Ask Sherpa to build a chart, table, or metric block. It becomes a reusable widget you can pin for later. |
Scheduled reports | Turn any prompt into a recurring email to any distribution list, on any schedule. It works like a standing trading brief that runs without anyone lifting a finger. |
External context | Sherpa can factor in outside data like seasonality or interest rates. It can also analyze influencer or competitor content to help explain a soft or strong period. |
How do you chat with Sherpa?
1. Start from the module closest to your question
Sherpa lives inside every core module, such as Attribution Vertex and Product Intelligence. It also has its own space in the top left of ProfitPeak. If you open it from the most relevant module, it starts with the right context already loaded.
2. Ask like you'd ask a colleague, not a search engine
Be specific about what you want to see and what a good answer looks like. Here are some prompts taken from real merchant requests:
"Give me a 30-day media buying roadmap for Google Ads."
"Split Google Ads into brand vs. non-brand traffic."
"Build a bundle analysis. What do customers who buy [Product X] also purchase?"
"Which Meta ads are driving contribution profit, and do we have stock to scale them?"
"Which products should I exclude from my Google feed this week?"
3. Follow the thread instead of starting over
Every answer ends with a suggested next prompt. Even someone newer to the data can keep digging without knowing exactly what to ask. Follow that thread rather than starting a fresh query each time.
4. Correct it in the conversation
If a result uses the wrong attribution model, date range, or product set, just tell Sherpa what to change. It updates the answer in place, so you don't have to rebuild the query from scratch.
5. Pin anything you'll want again
When an answer is something you'll come back to, use the three dots menu to pin it as a widget to a Peak. A Peak is a saved dashboard inside a workspace. Pinning turns a one-off question into a permanent view you can reuse. See the Peaks article for the full workflow.
6. Schedule anything that should be recurring
Any prompt can become a standing email report, sent to any distribution list on any schedule. This works well for weekly trading briefs, buying briefs, or leadership summaries that shouldn't depend on someone remembering to run them.
Why might the same prompt give a different answer later?
Sherpa generates each answer fresh from your live data. If the underlying data changes, a saved prompt can return a slightly different result than it did the first time.
For that reason, pinning matters. Check a query's logic once, and when it's right, pin it as a widget. From then on it's a locked, trusted view you can open without re-checking it every time.