Perform post-launch evaluations with Spark

Jake Sawyer

Updated June 04, 2026 14:18

Note: Productboard Spark is currently in open beta and available to all customers. The beta is a standalone experience that doesn't integrate with existing Productboard workspaces.

If you're on an Enterprise plan and workspace integration is important to you, contact your Productboard representative to discuss options. Otherwise, you can join the Spark beta here.

Shipping a feature doesn't tell you whether it worked. Spark's post-launch evaluation skill guides you through a structured impact review that combines your original goals, analytics data, and customer feedback to produce a verdict and a recommended next action.

What post-launch evaluation does

The post-launch evaluation skill guides you through a structured review of a shipped initiative or feature. Spark reads your entity details, spec, and original goals, then compares them against post-launch analytics and customer feedback to produce an evaluation document.

At the end of the flow, Spark creates the document in your personal section and links it to the evaluated entity. The document includes:

An executive summary with an impact rating and confidence level.
A breakdown of target outcomes vs. actual performance.
Analytics highlights showing weekly metric trends and segment breakdowns.
Feedback highlights comparing pre-launch and post-launch customer signals.
Recommendations and a suggested next action.

Spark assigns one of four impact ratings:

Rating	What it means
🟢 On track	Primary metric is meeting or exceeding target; qualitative signal confirms value.
🟡 Mixed signals	Some metrics are off-target or qualitative concerns are present.
🔴 Off track	Primary metric is significantly below target; qualitative signal shows the original pain is persisting.
◽ Pending	Insufficient data to make a determination. Spark lists what's missing and how to get it.

Before you begin

To run a post-launch evaluation, you need:

A Productboard space with Spark enabled.
A shipped (or recently shipped) initiative or feature.
Success metrics defined in the initiative or feature's spec, if available. Spark measures performance against these original targets.

Note: An analytics integration is optional. If you connect Amplitude, Pendo, or Hex via MCP, Spark queries them directly. If not, Spark walks you through a manual data input flow using a CSV upload or exported report. More on analytics integrations below.

Running a post-launch evaluation

Step 1: Launching the skill

Launch the post-launch evaluation skill in any of these ways:

Type /post-launch-evaluation in any Spark chat.
Click the skill on Spark Home.
Select it from the Skills page.

Step 2: Identifying the initiative or feature

Tell Spark which initiative or feature to evaluate. You can:

@mention the initiative or feature directly.
Type its name (for example, “Smart Insights Digest”).
Describe it (for example, “the sprint planning feature we shipped in February”).

Spark confirms the entity name, release date, and original goals before continuing. If the release date is ambiguous, Spark asks you to confirm it rather than guessing.

Step 3: Confirming your target metrics

Spark checks whether the initiative has clearly defined success metrics in its original spec or brief. If metrics exist, Spark confirms them with you and uses them as the evaluation baseline.

If no metrics were defined at spec time, Spark flags this as a gap and asks what outcome you were expecting. Any metrics you add at this stage are noted as post-hoc in the final document.

Note: Spark always evaluates against your original targets, not metrics added after the fact.

Step 4: Confirming instrumentation events

Before pulling data, Spark identifies the analytics events that map to your success criteria (for example, plan_applied for an adoption metric). It shows you each event name alongside the success criterion it covers and asks you to confirm or correct the list before proceeding.

Step 5: Gathering quantitative data

If you have an analytics tool connected via MCP, Spark queries it directly and retrieves:

Primary metric performance week by week (weeks 1, 2, 4, 8, and 12).
Secondary metric performance, if defined.
Segment breakdowns (target segment vs. general population).
Benchmarks such as rolling three-month averages.
Top engaged users and accounts.

If no integration is available, Spark generates a copy-paste prompt you can run in your analytics tool and return the results as a CSV or link.

Note: If data is unavailable due to an instrumentation gap or because not enough time has elapsed, Spark marks those cells as "Not available" and records them as open items. Spark never estimates or extrapolates.

Step 6: Reviewing qualitative feedback

Spark searches your Productboard notes for customer feedback related to the evaluated functionality. It splits feedback into pre-launch and post-launch windows, then compares:

What customers said before the launch.
What customers are saying after the launch.
Whether the original pain appears resolved, reduced, persisting, or evolving.

Spark surfaces representative quotes from both windows, positive and negative.

Step 7: Reviewing the verdict and output document

Spark synthesizes the quantitative and qualitative evidence, assigns an impact rating and confidence level, and creates an evaluation document in your personal section. The document includes a recommended next action:

Iterate: The feature is working but has clear improvement opportunities.
Close: The feature succeeded; no further investment is needed.
Expand: The feature exceeded expectations; invest more.
Investigate further: Impact can't be determined; resolve data gaps first.

Understanding the evaluation document

The evaluation document follows a fixed structure. Here's what each section contains:

Section	Contents
Executive summary	Impact rating, confidence level, release date, and a brief summary of the available signal.
Target outcomes	Primary and secondary metrics with their definitions.
Analytics highlights	Weekly metric table, benchmark comparisons, segment insights, and a link to the source dashboard.
Feedback highlights	Pre- and post-launch signal, target segment check, and representative customer quotes.
Recommendations & next steps	Synthesis, suggested next action, and open items for any unresolved data gaps.

Tip: If any data is missing when Spark generates the document, it lists each gap as an open item under “Recommendations & next steps” with a specific suggested action so you know exactly what to resolve before re-evaluating.

Connecting an analytics tool

Spark supports several analytics tools like Amplitude, Pendo, and Hex via MCP. Connecting one of these tools lets Spark query your analytics data directly instead of relying on manual input.

To learn how to connect an analytics tool, see Connect external tools to Productboard Spark.

Once your analytics tool is connected, return to Spark and run /post-launch-evaluation. Spark detects the connected integration and queries it automatically.

Note: If you don't use Amplitude, Pendo, or Hex, you can still complete an evaluation by providing data manually via CSV upload or a link to an exported report.