Prioritize with AI

Learn how to use ProductLift's two AI-driven prioritization tools to score individual posts against a framework or surface the top five posts to focus on next.

ProductLift ships two separate AI-driven prioritization tools. They live in different screens, produce different output, and are useful in different situations.

Both tools use the same underlying signal: your Product Vision plus engagement data on the post. They cost 1 AI credit per run.

Prerequisite: a Product Vision

Both tools rely on your Product Vision. If it is vague or empty, the AI has nothing strategic to align to and its output will feel generic.

How to get there: Click Prioritize in the sidebar → Product Vision.

At minimum, fill in the Vision statement, Target group, and Business goals. See Create a product vision for details.

AI Suggest: score individual posts

Use AI Suggest when you already picked a framework (RICE, ICE, I/E, or MoSCoW) and want the AI to fill in the numeric scores for a batch of unscored posts, or to sanity-check the ones you already scored.

How to get there: Click Prioritize in the sidebar → Score-Based Prioritization → AI Suggest button in the toolbar.

What the AI sees

For every post in the batch, the AI receives:

• Your Product Vision (Vision, Target group, Needs, Product, Business goals)
• Up to three posts you have already prioritized manually, used as calibration reference
• The post title, description, vote count, and comment count
• The framework you selected, with its scoring rules

The AI processes up to 20 posts per run.

How each factor is scored

Every factor is on a 1-10 scale. The composite score is calculated with the same formula the manual scoring uses, so an AI-suggested score and a manually-typed score are directly comparable.

For RICE and I/E, Effort is scored so that 10 means "heavy lift" and 1 means "trivial". For ICE, Ease is scored the opposite way, 10 means "very easy". Switching between ICE and RICE/IE mid-project will trigger a warning because the same stored value flips meaning.

How to influence the AI's suggestions

There is no per-portal prompt override today. What you can tune:

• Sharpen your Product Vision. This has the biggest impact. The AI leans on it heavily for Reach and Impact scores. A vague vision produces vague scores.
• Manually prioritize 3-5 posts first. The AI picks up to three of your existing scored posts as reference and uses them to calibrate the rest of the batch. In effect it learns your team's sense of scale, a "7 Impact" for your team becomes a "7 Impact" for the AI.
• Pick the framework that matches how you actually decide. ICE tends to feel more intuitive when Effort estimates are noisy, RICE gives more granularity when Reach varies a lot between features.

Applying suggestions

After the AI returns, you see a modal listing each suggestion with the suggested values and a one-sentence reasoning.

• Apply on a single row writes those values to that post and removes the row from the modal.
• Apply All writes every suggestion in one go and closes the modal.

The numeric scores are saved to the post. The reasoning text is shown in the modal but is not persisted, if you want to keep it, copy it into the post description before closing.

AI Prioritization: surface the top five posts

Use AI Prioritization when you have a large backlog and want a "what should we focus on this quarter?" shortlist rather than per-post scores.

How to get there: Click Prioritize in the sidebar → AI Prioritization.

What the AI sees

• Your Product Vision (same fields as above)
• The top 50 posts by engagement from the statuses you configured for prioritization
• Vote count and comment count per post

What it returns

A ranked list of five posts, each with:

• A Bootstrap icon representing the theme
• A brief explanation of why it was selected

Unlike AI Suggest, this does not write anything to the posts, it is a read-only recommendation view. Use it to decide what to move into your next planning cycle, then score those posts individually with AI Suggest or manually.

When to use which

• Grooming a backlog for a planning session → AI Suggest, so every post has comparable numbers.
• Answering "what should we work on next?" → AI Prioritization, for a focused shortlist.
• Onboarding a new team member → AI Prioritization first, to align on strategic direction, then AI Suggest to teach them the scoring framework.

Limitations

• The AI cannot estimate technical effort or feasibility. Its Effort/Ease scores are best-guess based on post description length and complexity signals, not on engineering knowledge.
• It does not weight by customer MRR, enterprise vs. self-serve, or contractual commitments.
• It does not know about political context, executive mandates, or dependencies on other features.
• Prompt customisation per portal is not available today. If this would be useful for your team, mention it to support.

Treat AI output as a starting point that a human product manager then adjusts, not as the final answer.

Troubleshooting

AI suggestions feel generic or off-strategy. Your Product Vision is probably too vague. Rewrite it to be specific about who you serve and what you are trying to achieve.

AI keeps giving similar scores to every post. Manually score 3-5 posts across the range (some 3s, some 8s) so the AI has calibration reference. Then re-run AI Suggest.

Composite score looks wrong. Confirm the factors are in the expected 1-10 range on the post detail. If you see values above 10, contact support with the post URL.

Related articles

• Create a product vision
• Use prioritization methods (RICE, ICE, MoSCoW, I/E)
• User segments