# Lesson 55: ASO roadmap: prioritize your optimization backlog

Select a feasible first batch of work and explain what the team will defer.

Prerequisite: Your audit, keyword strategy, experiment history and available team capacity.

Teaching examples are fictional unless explicitly identified as platform definitions.

## Foundations

### Start with the objective

A backlog is a set of choices under constraints. Connect each item to the useful customer or business outcome from the project brief. Separate verified corrections, evidence collection and experiments because they have different completion conditions. A large list is not a strategy until the team decides what should happen first and why.

### Compare evidence and dependencies

Estimate the audience affected, plausible value, confidence and effort using explicit reasoning. Product truth and urgent defects should not be traded away by a convenient numerical score. Keep dependencies visible: an asset test may need approved copy, while localization may need a product release and native reviewer.

### Limit work in progress

Choose a batch the available people can finish and verify. Reserve capacity for review and analysis rather than allocating every hour to production. Keep deferred items with a reason and a condition for reconsideration. Revisit priorities when evidence changes, not simply when a stakeholder repeats a request more loudly.

## Apply the method

### Make the unit of work small enough to finish

Split vague items such as improve screenshots into a brief, production pack, review, experiment setup and readout when those have different owners or dependencies. Preserve the parent objective so the pieces do not become disconnected tasks. Each item needs a completion condition that can be inspected. “Research localization” is incomplete until it specifies the markets, questions and decision output. Small, well-defined work makes capacity and blocked dependencies visible before the schedule becomes unrealistic.

### Use scoring as a conversation aid

A weighted rubric can expose assumptions, but its output is only as credible as the judgments entered. Keep evidence confidence, effort and dependencies visible beside any score. Do not multiply guessed impact by guessed reach and present the result as a revenue forecast. Compare scenarios when uncertainty could reverse the order. Mandatory accuracy corrections and critical product defects may need a separate lane because their priority is not adequately represented by a speculative growth score.

### Choose what stops or waits

Every new priority consumes capacity that could serve another task. Record the deferred item and the condition that would bring it back. If a design resource can complete only one treatment this cycle, selecting three creative experiments is not a plan. Reserve time for approval, publication and interpretation. At the review, update priorities based on new evidence and actual capacity rather than treating the original order as a commitment that cannot change.

## Procedure

1. Classify the backlog into corrections, investigations and experiments.
2. Connect each item to an outcome and supporting evidence.
3. Record effort, dependencies and an owner with realistic capacity.
4. Choose the first batch and document why other items wait.

## Worked example

Trail Notes corrects an outdated offer, validates activation tracking and briefs a clearer opening screenshot. A second-market rollout waits for product-language readiness. The team does not claim that these three tasks have the highest numerical return; it explains their evidence and dependencies.

| Backlog item | Evidence and dependency | Decision |
| --- | --- | --- |
| Correct outdated offer | Verified mismatch; copy approval needed | Do first |
| Validate activation event | Baseline depends on reliable tracking | Investigate now |
| Launch new language | Product and reviewer unavailable | Defer with readiness gate |

The first two items enable trustworthy acquisition work. The localization idea may have high potential but is not executable under current conditions. The decision record should state what would change that: a supported local experience, reviewer capacity and a clear pilot objective. Deferral is not rejection. It protects the team from starting work that cannot reach a valid release or produce interpretable learning.

## Your ASO assignment

1. Create six backlog items with classification and evidence.
2. Choose three for the first batch and identify the limiting resource.
3. Write the condition that would reopen two deferred items.

You have one designer and six proposed tests. Select one treatment using evidence and dependencies, then explain which research can proceed without that designer and which work should wait.

## Handover

### Backlog definition

Give each item an outcome, classification and inspectable completion condition.

### Priority evidence

Record affected audience, confidence, effort and dependencies without false numerical precision.

### Capacity plan

Name available contributors and include review, release and analysis time.

### Deferrals

Explain what waits and the specific condition for reconsideration.

### Review decision

Keep changes in priority with their evidence and accountable owner.

## Check your work

- Priorities serve a defined outcome.
- Uncertain impact estimates remain labeled as judgments.
- The first batch fits real production and review capacity.

## Common mistakes

- Using a precise score to hide unsupported assumptions.
- Starting everything and leaving no capacity to review results.

## Knowledge check

Should a speculative high-impact idea outrank a verified misleading offer?

Correct the inaccurate offer through the appropriate process. Then compare the speculative idea with other learning opportunities using evidence, effort and dependencies. A score should not make a known false promise acceptable.

## Sources

- [Apple — Creating your product page](https://developer.apple.com/app-store/product-page/) — Product-page fields and presentation guidance.
- [Google Play — Metadata policy](https://support.google.com/googleplay/android-developer/answer/9898842?hl=en) — Accurate, relevant metadata and prohibited presentation.

Platform references reviewed 29 September 2026.
www.asoagency.com