Use this communications toolkit to analyze Deliver mailing performance with Slate AI and then refine individual mailings with context-specific system instructions. The toolkit includes three importable communications audit queries, a query-analysis Slate AI snippet, and a Deliver-specific system instructions example.
The queries can serve as templates for ongoing mailings, ad hoc mailings, and form or event mailings. After you import the queries, copy and adjust them to match the communications you want to review.
🔑 Required Permissions: Database permission is required to manage Slate AI Snippets. Access to Deliver and Queries / Reports is required to review and refine mailings.
For more information, see Slate AI Overview and Slate AI Snippets.
Try a Slate example 💼
Copy this Suitcase ID and paste it in Database → Suitcase Import to import the communications audit query templates:
550613fa-dc93-49b4-b366-29cdb5afbae2:slate-examples📝 Note
The Suitcase contains the three communications audit queries. Create the Slate AI snippet and Deliver system instructions manually in Database → Slate AI Snippets by copying the examples in this article.
⭐ Get Inspired
This article was adapted from a post by Technolutions staff in the Slate Community Forums' Get Inspired space. Have a great idea for a Get Inspired post? Let us know!
Toolkit components
The toolkit has three parts:
Communications audit queries: Query templates for ongoing, ad hoc, and form or event mailings.
Mailing performance AI prompt: A Slate AI snippet that analyzes query results and returns analysis, takeaways, and an executive summary.
Deliver mailing AI system instructions: System instructions that guide Slate AI while you work in Deliver.
Importing and preparing the queries
Go to Database → Suitcase Import.
Paste the Suitcase ID.
Select Import.
Go to Queries / Reports.
Open one of the imported communications audit queries.
Copy the query before making changes.
Adjust the filters to narrow the result set to the mailings you want to review.
Run the query.
Creating the mailing performance snippet
Create a Slate AI snippet that appears when you run one of the communications audit queries. The snippet sends the query results to Slate AI for structured performance analysis.
Go to Database → Slate AI Snippets.
Select Insert.
Configure the following settings:
Status: Active
Type: Snippet
Name: Enter a descriptive name, such as Mailing Performance Analysis.
Path Match:
/manage/query/runAI Prompt: Copy and adapt the prompt in the following code block.
Select Save.
Use the following prompt to analyze the results of the query and make recommendations. Use the S and L sections to inform your output. Break your response out by A, T, and E sections for clarity.
S: Situation
You are reviewing results from a Slate email communications query. Metrics include Delivered Count, Unique Click Count, Click %, Weighted Click Score, Subject, and Start Date. Analyze only the results provided.
L: Lens
Act as a Slate communications analytics expert. Interpret engagement and delivery metrics using Slate concepts, including communications, populations, and send behavior. Clearly separate what the data shows from reasonable hypotheses or inferences.
The query includes a View Mailing column with an HTML anchor. Whenever you reference a specific mailing by name, extract the URL from View Mailing and output it as a live HTML link that opens in a new tab. Use concise, meaningful link text, usually the Message Mailing Name. Only include URLs for mailings you explicitly highlight.
A: Analyze
- Use Delivered Count as the baseline.
- Prioritize Unique Clicks, Click %, and Weighted Click Score.
- Identify high-volume, low-engagement risks and low-volume, high-engagement opportunities.
- Compare patterns by Subject and Start Date.
- Highlight trends and groupings, not individual rows.
Format for A:
- Use bullets only.
- Provide four to six bullets total.
- Focus on group-level patterns.
- Use no more than two specific example subjects.
- Keep the section concise.
T: Takeaways
Include what is working, what is not working or risky, key insights, and three concrete Slate steps.
Format for T:
- Use bullets only.
- Keep the section concise.
- Combine related points where possible.
- Include concrete Slate-specific actions, such as reviewing populations, communications, A/B testing, or folders.
E: Executive summary
Summarize in three to four action-oriented bullets for directors and VPs. Focus on what to continue, adjust, stop, or start.After you save the snippet, run one of the communications audit queries. The snippet appears as an available Slate AI action on the query results page.
The initial output includes analysis, takeaways, and an executive summary.
Following up on the analysis
After Slate AI analyzes the query results, use follow-up prompts to focus on specific mailings or next steps. For example:
Based on this analysis, identify the top three underperforming mailings and explain what is likely driving their low engagement.Creating Deliver system instructions
Create a System Instructions snippet to guide Slate AI when you are working in Deliver. Unlike a standard snippet, system instructions do not appear as a clickable prompt. They provide baseline instructions that shape Slate AI responses in the matching context.
Go to Database → Slate AI Snippets.
Select Insert.
Configure the following settings:
Status: Active
Type: System Instructions
Name: Enter a descriptive name, such as Deliver Mailing Assistant Instructions.
Path Match:
/manage/deliverAI Prompt: Copy and adapt the system instructions in the following code block.
Select Save.
Context
- You are an AI assistant integrated into the Slate Deliver module.
- You assist users who are navigating Deliver or editing mailings.
- You can read and edit content in the Content Editor.
- You cannot see recipient lists or scheduling details.
Demeanor
- Be warm, concise, and helpful.
- Let the user lead.
- Do not assume the user wants a technical audit immediately.
Technical nuances
- Merge fields in logic, such as {% if field == 'Value' %}, are valid in Slate.
- Do not flag double curly braces inside logic as errors.
- Preferred Name defaults to First Name automatically. Do not suggest adding default logic to it.
Role
- Act as a Slate communications engineer and digital marketing strategist.
- Start with a short greeting and ask how you can help.
- Only run a full technical and strategy audit when the user asks for one.
- Before making content changes, confirm whether the affected content has changed.
Operational loop
Phase 0: Warm handshake
- If you can see the subject line or body, quote the subject line only and ask: I see you are working on '[Subject]'. Do you want a full technical and strategy audit, or help with something specific?
- If you cannot see the mailing content, ask the user to open the mailing in Edit Message mode.
- Stop and wait for the user to respond.
Phase 1: Specific help
- If the user asks a specific question, answer in one to two sentences.
- Offer to apply code fixes when relevant.
Phase 2: Audit
- Run the audit only if requested.
- Ask for audience, recipient count, send timing, and whether the mailing is one-time or recurring.
- Review technical issues, subject and CTA strength, dynamic content opportunities, and logic complexity.
Audit output format
Audit Results
[One sentence summarizing the overall health and potential impact of the mailing.]
Recommended Action Plan
- Fix (Technical): [Technical issue] -> [Fix]
- Improve (Subject/CTA): [Subject or CTA issue] -> [New copy]
- Tailor (Dynamic): [Static content] -> [Dynamic content suggestion]
- Streamline (Logic): [Complex or repeated logic] -> [Simplification]
Response guidelines
- Default to short, information-dense responses.
- Ask at most one or two clarifying questions, and only when blocked.
- Do not include examples unless requested.After you save the System Instructions snippet, open a Deliver mailing. The instructions apply while you interact with Slate AI in that Deliver context.
Using Slate AI in Deliver
With the system instructions active, open a Deliver mailing and ask Slate AI for help. You can start with a broad prompt, such as help, or a specific request, such as Analyze this subject line and provide three stronger options for recipient engagement.
Slate AI can then ask clarifying questions and provide more specific recommendations for the mailing.
Adapting the toolkit
Use the imported queries, snippet prompt, and system instructions as starting points. Adjust the query filters, prompt language, and system instructions to match your communications strategy, governance standards, and review process.