At a glance
Per-template performance score plotted over time, shows whether each template (homepage, PDP, collection, cart, checkout, search, about) is improving, stable, or regressing across audit cycles. The trend reveals which templates are quietly sliding while another template grabs headline attention. A homepage that holds at 75 while collection drops from 65 → 41 over 60 days reveals where engineering effort is needed even though the headline aggregate score may have moved less than expected.
| What it counts | Per-template Lighthouse Performance Score plotted across audit cycles (typically weekly), one line per template. Reveals direction of change and rate of change for each template independently. |
| Sample type | Lab data from per-template Lighthouse audits, plotted over the trend window. |
| Why per-template trend matters | The site-wide aggregate score is an average; one template’s rapid decline may be masked by another’s improvement. Per-template trend exposes hidden regressions and confirms wins at the template granularity, which is where ship decisions actually happen. |
| Reading the trend | (1) Identify templates with downward slopes, these are the regressions to investigate. (2) Identify templates with upward slopes, these confirm previous fixes are paying off. (3) Look for inflection points, release dates, theme changes, app installs that align with score breaks. (4) Compare slope magnitudes, a 20-point drop on collection over 30 days needs urgent attention; a 3-point drop is noise. |
| Currency | n/a, score 0-100 over time. |
| Time window | 90D rolling, daily granularity (or weekly depending on audit cadence). |
| Alert trigger | any-template-score-drop > 15 points in 30 days (rapid template-level regression detected). |
| Sentiment key | null (line chart; sentiment driven by direction). |
| Roles | owner, operations |
Calculation
Calculated automatically from your Website Performance (PageSpeed + CrUX) data. See the At a glance summary above for what the metric tracks and the worked example below for a typical reading.Worked example
A UK-based BigCommerce fashion store, per-template trend across 90 days, Wednesday 15 May 26.| Template | 90 days ago | 60 days ago | 30 days ago | Today | Net change | Slope |
|---|---|---|---|---|---|---|
| Homepage | 76 | 75 | 74 | 72 | -4 | mild downward |
| Product detail page | 70 | 69 | 68 | 68 | -2 | flat |
| Collection page | 65 | 58 | 49 | 41 | -24 | steep downward |
| Cart | 78 | 78 | 77 | 78 | 0 | flat |
| Checkout | 86 | 86 | 86 | 86 | 0 | stable |
| Search | 64 | 62 | 64 | 64 | 0 | flat |
| About / contact | 80 | 81 | 82 | 82 | +2 | mild upward |
- Collection page is in steep decline. -24 points in 90 days. This is the most important signal in the entire view. No other template is moving at this rate. Trend reveals what a snapshot can’t: the bottleneck is collection and it has been worsening for the full window.
- Inflection points to investigate. Score break around the 60-day mark (65 → 58) suggests a release or theme change at that date. Second break around the 30-day mark (49 → 41) suggests a follow-on regression. Bisect commits or app installs on those dates, most template-level regressions trace to a specific change rather than gradual decay.
- Other templates are essentially flat. Homepage drifts mildly down (-4 points) which is within noise; cart, checkout, search are stable; about/contact is mildly improving. Collection is the only template requiring urgent action.
- Why this is surfacing now. The aggregate site-wide score has moved less dramatically because checkout (86) and cart (78) have anchored the average. The collection regression has been hiding in the aggregate, only visible when the trend is decomposed per template. This is the diagnostic value of the card.
- What changed on the 60-day mark, candidate causes: (a) theme update, check theme version history; (b) new app install (filter widget, image lazy-loader, review widget); (c) image library refresh, bulk upload of higher-resolution product imagery without size optimisation; (d) framework update, Stencil/Catalyst version bump. The timestamp gives the bisect window; engineering can pull the changelog and identify the candidate.
-
Recommended response:
- Hour 1: Bisect changes between 60 days ago and today. Identify candidate root cause.
- Day 1-3: Roll back or fix the regressing change. Confirm score lifts back toward 65+.
- Day 4-7: Re-audit collection daily to confirm trend reverses.
- Day 14: Confirm 30-day trend slope has flattened or reversed.
- What to do if the trend doesn’t reverse after rollback. It means the regression is in another layer (CDN config, server-side change, third-party dependency). Escalate to Vortex Mind investigation with the trend timestamps as the bisect anchor.
- Identify steepest downward slope. That’s the highest-priority investigation.
- Locate inflection points within the slope (release dates, app installs).
- Bisect changes at each inflection point.
- Roll back or fix and confirm the trend reverses.
| Time horizon | Action |
|---|---|
| First 1 hour | Identify steepest downward template. Locate inflection points. |
| First day | Bisect the changes within the inflection window. |
| First week | Roll back or fix the offending change. |
| Re-audit | Confirm 30-day slope has reversed. |
Sibling cards merchants should reference together
| Card | Why merchants reach for it |
|---|---|
psi_score_by_template | Per-template snapshot; this card adds the trend dimension. |
psi_slowest_template | Single-line slowest template; trend confirms whether the slowest is degrading or stable. |
psi_lcp_by_template | LCP per template; identifies which CWV drives the score change. |
psi_score_trend | Site-wide score trend; this card decomposes by template. |
psi_biggest_regression | Biggest URL-level regression; this card shows template-level regression. |
crux_regression_timeline | Field-data regression timeline; complements lab-data per-template trend. |
Reconciling against the vendor’s own dashboard
Where to look:- PageSpeed Insights, single-URL audits per template, captured manually over time (no built-in trend view).
- Lighthouse CI, if the merchant runs Lighthouse CI, per-template trends may already exist in their CI dashboard.
- Google Search Console, Page Experience report shows page-group experience over time; template-adjacent.
| Reason | Direction | What to do |
|---|---|---|
| Audit cadence. Vortex IQ audits per template on a fixed schedule (typically weekly). Manual one-off audits on different days will show different points. | Variable | Use the Vortex IQ trend as the canonical record; manual audits sanity-check direction not absolute. |
| Run-to-run variance. Lighthouse score variance is ±5 points per run for typical pages. Single-run readings may diverge from the trend line. | Either direction | Look at the slope across multiple points, not single-point comparison. |
| Template classification. If a template’s URL pattern changed during the window, the trend may include URLs from different templates over time. | Variable | Confirm template-to-URL mapping has been stable across the window. |
| Audit URL changes. If the representative URL changed during the window (PDP product retired, collection renamed), the trend may have a discontinuity. | Variable | Check audit URL stability per template. |
psi_score_by_template, psi_score_trend, crux_regression_timeline).
Quick rule for support tickets: when a merchant disputes a trend, check (1) audit URL stability, (2) template classification stability, (3) Lighthouse run-to-run variance. Most disputes resolve within those three checks.
Known limitations / merchant FAQs
Q: Our site-wide score trend looks flat but this card shows collection dropping 24 points. Which is right? Both. The site-wide score is an average across all templates; if collection drops 24 points but checkout is anchored at 86 and homepage holds at 72, the average barely moves. The per-template trend is the more useful diagnostic when you need to know where to act. The aggregate trend is the headline; the per-template trend is the root-cause story. Q: How do we tell a real regression from Lighthouse run-to-run variance? Slope and persistence. A 3-5 point fluctuation week-to-week is normal variance. A 10+ point drop sustained across 3+ audit cycles is a real regression. Use the 30-day-net-change column as the primary signal, single-week jitters get smoothed out across 30 days. Q: A template’s score jumped up 15 points after we shipped a fix. How long until we trust it? Wait one to two more audit cycles. A single uplift could be variance; two consecutive cycles of held gain is a confirmed win. The trend chart will show whether the new level is stable or whether the score regresses back. Plan a “win confirmation” milestone two audit cycles after every ship. Q: Our cart template has been flat at 78 for 90 days. Is that good or bad? Context-dependent. Flat at 78 is fine for cart pages, they’re typically lean and stable. Flat at 41 would be bad. Use the score level alongside the trend: flat-and-healthy is the goal; flat-and-failing means you’ve stagnated and need a refactor sprint. Q: One of our templates shows a 20-point drop on a single day. Is that real? Almost always traces to a specific change at that timestamp. Check (1) the deploy log for that day, (2) any new app installs, (3) any theme version changes, (4) any third-party script changes (GTM tags). Single-day cliffs are diagnostic gold, they give you the bisect window. Q: Why isn’t checkout in this trend? It only shows other templates. Likely a sample-size or stable-score reason. Some configurations exclude templates that consistently score above 90 or whose score variance is below ±2 points (because the trend would be a flat line at the top of the chart). Check the audit configuration if a template you care about is missing. Q: How does this card differ frompsi_score_trend?
psi_score_trend is a single line, site-wide aggregate score over time. This card is multiple lines, one per template, and reveals which template is driving the aggregate change. Use psi_score_trend for the executive view; use this card for the engineering diagnostic.
Q: Our trend shows two of our templates moving in opposite directions. How do we read that?
That’s a healthy diagnostic outcome, you’re seeing differential ship effects. Templates moving up are benefiting from your recent work; templates moving down need attention. The differential exposes where to invest next. A site where every template moves the same direction usually indicates a site-wide change (CDN, theme bundle); differential movement indicates template-specific causes.