Understand What's Driving Your ROAS with Spark AI Insights
ROAS changed. Now what? Spark AI Insights surfaces the why behind your performance shifts - so you spend less time investigating and more time acting. This article walks you through how to access, read, and act on your insights.
What Spark AI Insights shows you
Spark AI Insights provides two levels of analysis:
Channel-objective level combines your channel group, platform, and objective (e.g., Paid Social - Meta - Conversion, Paid Shopping - Google - Performance Max).
Campaign level drills into individual campaigns within each channel-objective to show specific performance patterns.
For each level, you'll see:
ROAS change compared to the previous period
Spend and revenue percentage changes
Whether the channel appeared to contribute positively or negatively to overall ROAS
Insights into observed patterns and performance changes
Suggested next steps to consider
Together, these insights help you quickly understand what changed, why it changed, and where to focus next.
How to access Spark AI Insights
Step 1: Navigate to your Home Dashboard
Log in to your Fospha account and select Home from the left-hand navigation.
Step 2: Filter by Channel group, Source or Market
At the top of the dashboard:
Select your date range (and comparison period, if needed).
Apply relevant filters such as Channel Group, Source, Market, Media, or Currency using the filter bar.
Spark AI Insights will reflect whatever view you’ve configured.
Step 3: Open Spark AI Insights
At the top of the dashboard, locate the Spark AI Insights banner.
Click “Find out why” next to the insight summary (e.g. “ROAS decreased by 0.4 points…”).
This opens the Spark AI Insight panel on the right-hand side.
Step 4: Review your insight summary
At the top of the panel, you’ll see:
A clear explanation of what changed (e.g. ROAS movement vs previous period)
A performance comparison
A trend chart
Below this, you’ll find:
Key drivers (channels pulling performance up or down)
Positive and negative contributors
Channel-level performance summaries
Step 4: Explore recommendations and drill deeper
Within each driver card, you can:
Review the AI-generated recommendation
Click “View [Channel] Performance” to navigate directly to that channel
Use the suggested follow-up questions in the “What else can I help you analyse?” section to investigate further drivers
Taking action on Spark's AI insights
Spark Agent provides a starting point for diagnosing performance changes. Always review suggestions against your business context, and validate through Fospha's other tools.
If a channel looks like it has room to scale - use Incremental Forecasting.
Spark may indicate that a channel is contributing positively and performing above your average performance.
Before adjusting budget, Incremental Forecasting can help you identify where additional spend is likely to generate incremental revenue and where saturation means returns may diminish sooner than expected.
Used alongside your Fospha's measurement tools, it gives you a forward-looking view that makes scaling decisions more grounded.
Example: Paid Social – Meta is flagged as a top positive contributor. Use Incremental Forecasting to model the likely incremental return of additional budget before committing spend.
If a campaign appears to be negatively impacting your ROAS - use the Optimization Dashboard.
Spark operates at channel and objective level, and ROAS shifts often originate from one or two campaigns within it.
The Optimization Dashboard lets you move from the channel view down to ad set/ad level granularity - identifying where reallocation may improve overall efficiency.
This is where the insight becomes a decision you can act on.
Example: Google – Performance Max shows a ROAS decline. Use the Optimization Dashboard to identify which campaigns within that objective may be decreasing the average performance and re-allocate spend accordingly.
Spark AI Insights, Incremental Forecasting, the Optimization Dashboard, and Channel Health Check are designed to work together. Each adds a layer - from diagnosis, to planning, to execution - so your measurement informs action at every stage.
Tips
Review insights regularly - Check Spark AI Insights daily to understand performance trends and identify patterns worth investigating.
Use as a starting point - Treat recommendations as hypotheses to validate with your business context, and dive deeper using other tools in Fospha's measurement stack.
Provide feedback - Use the thumbs up/down buttons at the bottom of the panel to help improve the insights over time.
FAQ's
Q: What time period does Spark AI Insights analyze?
A: It analyzes the date range selected in your dashboard filters, comparing it to the equivalent previous period.
Q: Can I analyze specific campaigns?
A: Yes. The insights show both channel-objective level and drill down to individual campaigns within each channel.
Q: How often are insights updated?
A: Insights refresh based on your latest data sync. Return to the dashboard and reopen Spark AI Insights to see updated analysis.
Q: What does "contributed negatively" mean?
A: A channel may show negative contribution when it reduced your overall ROAS. This can happen in several scenarios, such as when you reduce spend on a highly efficient channel, affecting the overall portfolio mix.
Q: Should I follow the recommendations?
A: Recommendations provide potential optimization directions based on your performance data. Review them alongside your business goals, budget constraints, and broader strategy. For significant budget decisions, consider validating through incrementality testing.
Q: Can I export the insights?
A: Yes. Click the "Share" button in the top right to export or share the analysis with your team.