Boldstream can pull your connected numbers and answer questions about them in plain English. The quality of the answer depends almost entirely on the quality of the question. A few habits make the difference between a figure you glance at and a figure you can act on.
The five-part query formula
Outcome + timeframe + comparison + segment + decision
"Show [outcome] for [timeframe], compared with [baseline], split by [one available segment]. Highlight [decision or action]."
A few worked examples:
"Show total sales for the last 90 days, compared with the previous 90 days, split by product. Highlight any product that moved materially."
"Show new sign-ups for completed months this year, split by acquisition source. Highlight the month with the largest change and what drove it."
"Show average order value for new customers versus returning customers in Q2. Highlight which group is growing faster."
When comparing your data, it is better to query one period at a time and line them up afterwards.
You don't need all five parts every time. But a query missing a timeframe or a baseline will almost always produce an answer you can't act on — a number with nothing to compare it to.
Before you ask: a short checklist
Name the decision. What will you do differently if the answer isn't what you expect? If nothing changes either way, the query probably isn't worth running.
Use a fixed period. Last week, last 30 days, month to date, a quarter, or a named campaign window. Vague periods produce vague answers.
Stick to complete periods. For clean comparisons, ask for completed months or weeks — for example, "completed months only." That stops a half-finished period being mixed in with full ones and making a trend look like a collapse.
Add a benchmark. Previous period, year on year, a target, or a clearly available comparison group. A number on its own isn't insight.
Choose one sensible breakdown. Ask narrow, then follow up. "Sales by month for the last year" is clean. "Sales by month by product by region by customer type" produces too many rows and a stitched-together answer that's hard to read and harder to trust.
Ask for the output you want. A trend, a comparison, the drivers of a change, a forecast range, or a recommended action. Say which.
Keep comparisons like-for-like. Match periods, markets, audiences and attribution rules before you draw a conclusion. Most misleading charts are a definition mismatch, not a data error.
Ask it to show its working
If a figure matters - if it's going in a board deck, a client report or a budget decision - ask Boldstream to show how it got there. Prompts like "show me how you got that" or "include the query you ran" will lay out what was pulled and how, so you can check the logic rather than take the number on faith.
This is also the fastest way to catch a definition mismatch. If the number looks wrong, it is likely that the query was answering a slightly different question than the one in your head.
For the definition of any metric you see, check the Glossary.
When the answer isn't there
Boldstream will flag when something isn't in your connected data, rather than guessing or filling the gap with a plausible-looking number.
Some gaps show up repeatedly across organisations:
No consistent customer identifier. Without a way to link one person's activity and transactions over time, questions about retention, repeat rate, churn and lifetime value can't be answered - no matter how much transaction data is connected.
No channel or attribution labels on conversions. If sales or sign-ups arrive without a channel, device or campaign label, they can't be connected back to media activity. Marketing-to-outcome questions stay out of reach.
No cost data aligned to activity. Efficiency, ROI and cost-per-outcome questions need spend mapped to the relevant media, partnership or operational activity. Revenue alone can't produce them.