Ask the Community: What Do You Do When You Inherit a Giant Product Backlog?

Teresa Torres and the Continuous Discovery Habits community explore what a product manager should do after inheriting a large, aging product backlog of feature requests and bugs. The discussion rejects both extremes of throwing the backlog away and treating it as a to-do list. Community members stress anchoring the backlog to current context, defining the outcome first, and using customer requests as invitations for further conversation. Teresa separates keeping the backlog as a historical record from using it to choose what to build next. She recommends searching the archive during discovery and, where permitted, using generative AI tools to surface themes and duplicates. The advice matters because it offers a practical middle path for PMs taking over legacy work.

01Key takeaways

• Do not discard an inherited backlog automatically; treat it as a possible source of customer insight.
• Define your product outcome before deciding which old backlog items matter.
• Use the backlog as a historical record to search during discovery rather than as a prioritized to-do list.
• Old feature requests lose value quickly, so weigh their age and context before relying on them.
• If using generative AI, use a business account and strip personally identifying information first.
• Iterate on AI prompts by giving feedback, since the first response is often imperfect.

02Key sections

Why backlogs can be valuable

Old backlogs can hold either stale internal ideas or valuable customer feedback, and the only way to know is to review them. The article frames this as a judgment call worth making early.

Community perspectives

Community members advise anchoring the backlog to current context, asking predecessors and stakeholders for their views, and recognizing that old requests lose value quickly. Even weak requests can prompt useful customer conversations.

Start with the outcome

Defining a product outcome first makes it easier to tell which backlog items are relevant. Without that anchor, sorting the list becomes arbitrary.

Using AI to summarize themes

Tools like Dovetail or ChatGPT can summarize high-level themes quickly, provided data privacy is respected. Expect broad themes rather than sharp insights.

Teresa's take: archive versus idea source

Keep the backlog as a historical archive, but do not use it to decide what to build next except for quick wins. Consult it during discovery once outcomes and opportunities are chosen.

Prompting and iterating with AI

Teresa suggests exporting the backlog and prompting an AI assistant to find themes, changes over time, and duplicates, using an enterprise account and removing personal data first. Treat the model like an intern that improves with feedback.

03From the post

“Old product backlogs can be the place where ideas go to die. But they can also be a treasure trove of feedback from your customers. And in some cases, they might be a mix of both. The trouble is, you might not know until you spend some time reviewing your”

04Frameworks mentioned

Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.