Users with access to ARI Insights can now chat directly with their advertising performance data using ARI Assistant Chat.

ARI Assistant Chat runs on a multi-agent architecture, which allows it to support multiple data sources at once. This includes Sponsored Ads, Walmart DSP, and Amazon DSP data agents, so users can ask about their full-funnel advertising performance — spanning sponsored ads and DSP — all in a single conversation.

The platform is designed to continue expanding to additional data sources over time, so you can rely on ARI Assistant Chat as a single place to ask questions across your advertising data.
​

Click on "ARI Assistant Chat" from the ARI Insights left nav to get started.

To get the most out of your advertising data, ask clear, specific questions about your campaigns, products, ad groups, and performance across sponsored ads and DSP. Follow these best practices to receive accurate insights.

When asking general questions, be specific and provide context. Instead of broad questions, narrow down the scope by including relevant details such as date ranges, specific campaigns, products, data source (sponsored ads, Walmart DSP, Amazon DSP), or performance metrics.

Example

❌ Original question: "How are my campaigns performing?"

✅ Refined question: "Show me the total ad spend, ACOS, and conversion rate for all active Sponsored Products campaigns in the last 30 days. Break down the results by campaign type."

Why this works: The refined question specifies the time period, desired metrics (ad spend, ACOS, conversion rate), and the grouping dimension (campaign type), making it easy for ARI Assistant Chat to provide actionable insights.

For complex analyses involving multiple metrics or dimensions, break your question into separate parts. This helps ARI Assistant Chat process each component accurately.

Example

❌ Original question: "What's the average CPC and CTR for each ad group across my top-performing campaigns, which are driving the most conversions, and how much budget should I allocate to each?"

✅ Refined question, part 1: "What is the average CPC and CTR for each ad group in campaigns with ACOS below 25% between January 1, 2026 and January 12, 2026?"

✅ Refined question, part 2: "Which keywords have generated the most conversions in these campaigns during the same period?"

✅ Refined question, part 3: "What is the total ad spend for each of these campaigns, and what percentage of my overall budget does each represent?"

Why this works: Asking questions separately allows ARI Assistant Chat to address each specific metric (CPC, CTR, conversions, budget allocation) with the appropriate filters for more accurate results.

When asking questions with specific conditions, clearly state all necessary filters, including exact dates, performance thresholds, campaign names, product ASINs, data source, and limiting criteria.

Examples

✅ "Show me all Sponsored Products campaigns with ad spend greater than $500 and ACOS above 30% between December 1, 2025 and December 31, 2025."

✅ "List the top 10 products by total sales, along with their ACOS, ad spend, and conversion rate for Q4 2025. Sort by total sales descending."

✅ "Provide the campaign name, total impressions, CTR, and ad spend for ad groups that have received fewer than 100 clicks in the last 7 days."

✅ "Which keywords in my 'Winter Collection' campaign have a CPC higher than $2.00 and fewer than 5 conversions in January 2026?"

✅ "Show me total spend and ROAS for my Amazon DSP campaigns in the last 30 days."

✅ "Compare my Walmart DSP impressions and click-through rate to my Sponsored Products performance for the same period."

Why this works: Each question clearly specifies the filters (date ranges, spend thresholds, campaign names, data source), metrics needed, and any sorting or limiting requirements.

When you ask for aggregated data, please specify the type of aggregation (sum, average, count, etc.) and the dimensions you want to group by (campaign name, product ASIN, ad group, keyword match type, data source, etc.).

Examples

✅ "Provide the sum of ad spend, average ACOS, and total conversions grouped by campaign type for the last 30 days."

✅ "Show me the average CPC, total clicks, and sum of sales for each product category in my active Sponsored Products campaigns."

✅ "What is the total number of impressions and average CTR for each ad group in my 'Best Sellers' campaign between January 1–12, 2026?"

✅ "What is my total full-funnel ad spend across Sponsored Products, Amazon DSP, and Walmart DSP for Q1 2026?"

Why this works: These questions clearly specify the aggregation type (sum, average, count) and the grouping dimension (campaign type, product category, ad group, data source), ensuring accurate results.

When asking follow-up questions, reference the previously established context but be explicit about any changes. If you're adjusting date ranges, performance thresholds, data source, or limits, clearly state the new parameters.

Examples

Original question: "Show me the top 5 campaigns by ad spend with ACOS above 35% in the last 30 days."

✅ Follow-up question: "Show me the same metrics but for campaigns with ACOS above 50% and expand the results to the top 10 campaigns."

✅ Follow-up question: "Now show me the top 5 performing ad groups within these campaigns."

✅ Follow-up question: "What were the results for the same campaigns in the previous 30-day period for comparison?"

✅ Follow-up question: "Now show me the same comparison but include my Amazon DSP spend as well."

Why this works: Each follow-up clearly specifies what changed (ACOS threshold, result limit, scope to ad groups, time period, data source) while maintaining the context from the original question.

Here are examples of well-formatted questions for typical use cases:

Campaign Performance

Product Analysis

Keyword Performance

Ad Group Insights

Full-Funnel & DSP Performance

Budget Analysis

ARI AI typically takes between 40-50 seconds to respond to any questions asked. The AI is analyzing all of your search performance data so the more complex your query the longer it may take. This response time will only improve overtime.