Hello Reader
A quick note: Next week, you’ll notice a new sender email address: tahnee@perryco.ai. It’s still me and your weekly AI at Work. Add it to your contacts so you never miss an issue!
~
Performance reviews are coming up, and AI can make the preparation considerably easier. It can gather project records, summarize feedback and produce a polished draft in minutes.
That’s appealing when reviews have long been criticized for taking too much time and doing too little to improve performance.
The risk is that both of you arrive with an AI-written account of the year and very little to say about it.
Jake Milstein, Head of Healthcare Marketing at Zscaler shared an experience he had when he was running marketing at Contrast Security.
“You know the employee deserves specific feedback, but I can’t always remember what happened six months ago,” Jake says. “Well, AI to the rescue!”
He used Claude to search old Slack messages, emails, and HR system notes (in this case HiBob), and it pulled specific moments in the employee’s history.
“I was able to provide the most detailed, and honest performance review I’ve ever delivered,” he says. “Claude even surfaced arguments in Slack that I hadn’t seen, so we could discuss some of the employee’s co-worker relationships as part of the review."
The employee told Jake it was the most comprehensive review of their career.
"Because it was so specific, I saw demonstrable change immediately after.”
If you want the best outcome for your performance reviews, then use the time AI saves to think more carefully about each person.
- Where have they grown?
- What have you asked them to take on?
- Have you given them enough support to do it well?
Evaluate AI using real work
So you've got your AI supported analysis.
What next?
Start assessing the person's performance, keeping in mind the responsibilities and goals you agreed they'd accomplish for the year.
- How did they apply AI to their work?
- What impact did it have on the business?
- Where did they exercise good judgment?
- Where do they still need to close the gap?
For your team, that means examining how someone used AI to make progress toward the agreed goal, including the decisions they made along the way.
Turn the review into a development agreement
Once you’ve assessed how each person is using AI and the results they’re getting, you can discuss what they’re ready to take on next.
A product marketer who spends less time summarizing customer feedback could take on more responsibility for interpreting the feedback and shaping positioning.
To do that well, they might need coaching in research methods, more time with customers or practice turning an analysis into a recommendation.
These are the capabilities you need in 2027
Across your marketing team, five capabilities deserve attention in the coming year:
Make each development goal specific enough that you can easily track progress.
The product marketer who’s using AI to analyze customer feedback might next develop an AI agent that tracks recurring pain points across sales calls and support tickets, flags changes and links each finding to the original customer comments.
Heidi Bullock, CMO at Tealium says her responsibility is to make employees “more marketable than when they first joined the company.”
Only 36% of respondents in BCG’s research felt they’d received adequate training. Your development agreement needs to spell out what you’ll provide, including opportunities to learn from your own experiments.
Lauren McHugh, Head of Marketing at Docket says if you want your team using AI, you have to be the first one in the sandbox.
"I build things in Claude myself every week: skills, agents, whole workflows," she says. "That does more than any mandate could, because people adopt what they watch their leaders actually use. Show your work, share what broke and make experimenting feel safe.”
Agree on the coaching, training and access each person needs to develop their chosen capability.
Make room for practice by removing or reassigning work where necessary. Then set a date to review progress together.
Enjoying AI at Work? Share with a friend of a colleague!
Prepare for the conversation
AI can help you gather project records, organize feedback and draft a review, but you’re responsible for interpreting that evidence, reflecting on each person’s progress and deciding what feedback they need to hear.
If you use Claude or ChatGPT for the analysis, make sure you dedicate time to reviewing the output and adding your own thoughts. Ask yourself:
- Where does the assessment need more context?
- Can you explain the feedback with specific examples?
- What do I need to hear from this person before I finalize my assessment?
When you sit down together, give your direct report room to challenge your assessment and share work you may have missed. Their perspective may change your understanding of their contribution and the development goals you agree on.
Prep for your performance review with this prompt
This prompt creates a brief for each person that includes their contributions, documented results and questions to discuss.
It helps you review work from across the year and spot gaps in the evidence before you assess someone’s progress.
Before you prompt:
Upload project exports, campaign reports, work samples or notes from one-to-ones to your company-approved AI tool. Include agreed goals and responsibilities.
The prompt:
Help me prepare for my marketing team’s performance reviews and identify development priorities for 2027.
Review period: [dates]
Team members and roles: [paste]
Agreed goals and expectations: [paste]
2027 priorities: [paste]
Review the uploaded records for missing periods, duplicate entries and gaps in results.
Summarize team progress against our goals. For each person, create a table of documented contributions, results, source references, questions to discuss and space for manager and employee observations.
Identify AI use or skill growth only with explicit evidence.
Ask up to five questions about missing contributions, changed priorities, support provided, strengths and development interests. Wait for my answers.
Then suggest one or two development goals per person. Consider these capabilities: AI fluency; workflow design and orchestration; data interpretation and experimentation; customer and creative judgment; cross-functional execution.
Specify the capability, practice opportunity, evidence of progress, training, manager support, time required and review date. Distinguish documented evidence, my observations and your suggestions.
Next week, I’m covering how to budget for AI in 2027.
Tell me where you expect to spend more, and I’ll share the results.
| Where do you expect the biggest increase in your team’s AI spending in 2027? |
|
|
|
|
|
|
|
What’s making your budget difficult to plan this year? Hit reply and let me know.
I’ll address reader questions in next week’s edition.