Impact Cards & wins
You optimized a page or published a new article. What did tracking observe afterward? Impact Cards compare fixed before and after cohorts and show the receipts without claiming the publish caused the change.
What is an Impact Card?
An Impact Card answers a narrow, checkable question: what did tracking observe before and after this live publish? It compares your page’s matched prompts — the tracked questions that page is supposed to answer — across two real tracking batches before and two after the live change, and reports what moved. When that evidence is not available, it says Inconclusive instead of guessing.
Impact Card
A cached, before/after observation report attached to a single publish. It answers three questions on that publish’s matched prompts: did you gain a citation you didn’t have before, did your mention rate change, and did your visibility score on those prompts change. Once computed, it’s stored permanently on that publish — it’s not recalculated on every page load. The target prompts and two baseline batch IDs are frozen when publishing begins, so later match changes cannot rewrite the baseline.
This is deliberately narrow. An Impact Card doesn’t claim your whole project got better, and it does not isolate the publish as the cause. It reports something specific and checkable: on the exact prompt-platform cells frozen for this page, here is what tracking observed before and after publishing.
The Verification state
Publishing is what starts the clock. When you push an optimized page or a newly generated article live — see Publishing back — three things happen at once: the current draft is snapshotted for one-click rollback, a publish record is created with a timestamp and a count of the accepted fixes that just went live, and the page’s status flips to published. That publish record is what opens a Verification window.
You publish
The page’s content is snapshotted. For a live WordPress update, the current prompt set and the last two qualifying real tracking batches are frozen before the remote write; the successful publish timestamp starts the after window.
The page enters Verification
The first 2 qualifying tracking batches that start after the live publish count toward closing the window. A newer publish caps the older window so its results cannot leak backward into the previous change.
Tracking runs advance it
A batch counts only when it finishes completed or partial, has a succeeded run and a present answer on a frozen prompt, and the adapter explicitly attests that the answer is real. Failed, running, simulated, and unknown-provenance data do not count.
The card becomes ready
Once two comparable after batches land, the card is computed and cached. If 14 days pass, a republish closes the window, or the evidence cannot be compared, the card resolves as Inconclusive with the reason and evidence progress.
Note
Local snapshots and WordPress drafts are not verified live changes, so they never receive a positive or negative impact verdict. A live publish with no frozen prompts or fewer than two baseline batches also resolves immediately as Inconclusive instead of a misleading “No change.”
How before and after are measured
An Impact Card compares two fixed cohorts of trusted, already-captured tracking measurements — nothing is re-fetched or re-run to build the card itself:
- Before — exactly the last two qualifying batches available before the remote live update began.
- After — the first two qualifying batches started after the successful publish, within 14 days and before any later publish of the same page.
Metrics are computed inside each batch first, then the two batch values are averaged. A large batch therefore cannot overpower a smaller batch. The same frozen prompt-platform cell must have a trusted measurement row in all four batches. A legitimate no-answer measurement stays in the cohort as zero instead of disappearing; this prevents a later favorable answer from manufacturing improvement through survivorship. If no exact cells can be compared, the result is inconclusive.
From those two windows, three signals are computed:
| Signal | What it compares |
|---|---|
| Citation gained | Was this exact page absent from both before batches and present in both after batches? A one-batch citation is not called a gain. |
| Mention rate | What share of the frozen prompt-platform cells mention your brand, before vs. after? Trusted no-answer cells count as zero. |
| Prompt visibility score | The mean per-prompt visibility score across matched prompts, before vs. after. |
The per-prompt visibility score is a 0–100 read on how present you are in a single prompt’s measurements: it uses the same mention and citation weights as the Optimization Queue but keeps the fixed cohort denominator. The live queue ignores platforms that did not answer; an Impact Card retains a trusted no-answer as zero so the before and after population cannot change underneath the result.
Reading the verdict
Once a card is ready, the three signals produce one plain-language label. Continuous deltas are observations, not calibrated judgments of whether the publish caused an improvement:
| Verdict | When it applies |
|---|---|
| Citation gained | This exact page was cited in neither before batch and in both after batches. This strict repeated transition is the only on-site Impact outcome promoted to Wins. |
| Observed change | Citation consistency, mention rate, or prompt visibility moved, but the strict citation-gain rule did not fire. The movement may be positive, negative, or mixed; there is no significance threshold. |
| No change | All three stored measures were exactly unchanged. |
| Pending | The verification window hasn’t closed yet. |
| Inconclusive | No defensible comparison can be made — for example the publish was not live, evidence was insufficient, a newer publish closed the window, or a legacy row has no frozen cohort. |
It won't show you an early verdict
The editor can compute a live preview of a card’s numbers at any time, but until the verification window actually closes, it only ever shows “Verifying” — never a sneak peek at a verdict that might change with the next batch. A card is cached and shown as final exactly once, the moment it’s truly ready. That’s a deliberate integrity check: no celebrating (or worrying) early.
Where you see it
A page’s current publish state and Impact Card live right in its editor, and a fresh card also surfaces as a standalone Impact card in the Action Center for 14 days after it’s computed. Results are newest first regardless of verdict; an inconclusive card is labeled as inconclusive evidence, never a win or loss. While an eligible live page is still waiting on batches, the Action Center instead shows a single Tracking card counting every page that’s awaiting re-measurement, with a direct link to run tracking now.
Example: a page after publishing
Impact2 before + 2 after batches
‘Best CRM for Startups’ was cited after publishing
Citation gained (0/2 before → 2/2 after) · Mention rate +18 pts · Visibility +9 pts · observed on the same prompt-platform cells across 2 before + 2 after batches
The Wins feed
Wins is a running list of recent workflow milestones and qualified receipts, shown alongside the Action Center feed. A milestone means work moved forward; it does not by itself prove visibility improved. The feed draws from four source families:
- Product milestones — a content gap that stopped being detected, a page that got published, a draft that finished generating.
- On-site citation gains — a cached Impact Card enters Wins only when the page moved from 0/2 cited before batches to 2/2 cited after batches. An Observed change card does not enter Wins, even when one continuous delta is positive.
- Earned-play receipts — completed off-site playbooks whose own measured window qualifies as a win.
- Win alerts — the moment a tracking run detects you newly won a prompt, newly earned a citation, or overtook a competitor, it lands here directly. See Alerts.
The eight most recent wins are shown, newest first, each linking back to the page or gap that earned it.
Measuring what happened next
This is the measurement step that follows everything else in these docs. Monitor measures where you’re losing. The Action Center turns that into evidence-specific lanes, sending you to Optimize to fix an existing page or Generate to create a missing one. You publish. Verification opens. Tracking runs land. And an Impact Card tells you, with specific numbers on specific prompt-platform cells, what changed afterward — not what caused it. Those observations feed back into Monitor’s next measurement and the Action Center’s next feed.
Start here
What is OptimizeCamp?
The whole measure–fix–create–prove loop, in plain language, from the top.
Prove
The Action Center
How verification and observed results join separate lanes for ready work, opportunities, and measurement.
Fix
Publishing back
What actually happens the moment you click publish — the step that opens a verification window.
Measure
The Visibility dashboard
The tracking runs that advance every open verification window.
Key takeaways
- A live publish freezes its prompt scope and last 2 qualifying baseline batches before WordPress changes; local snapshots and drafts make no impact claim.
- The first 2 qualifying after batches are compared per batch, within 14 days and before a republish; failed, simulated, and unknown-provenance data never count.
- A citation gain requires 0/2 cited before and 2/2 cited after. A verdict is never shown early.
- Missing or non-comparable evidence resolves as Inconclusive with a reason. Observed change and Inconclusive cards do not enter Wins.
Next step
New to OptimizeCamp entirely? What is OptimizeCamp? walks the full measure–fix–create–prove loop from the very beginning. Ready to see today’s work and evidence? Head to the Action Center.