# Measurement - Unit 3 (hosting and targets), final Requirements: NFR-PR-P-01, P-02, S-01; rules BR-HST-21, 23. Neutral reading only; the owner decides on targets and on every exception. No target is changed by this report. **Method**: `lhci collect`, 3 runs per URL, median, same PC and Node (v24.16.0) as units 1 and 2, `config/previewServer.ts` on port 4173 (started by the config with `--audit-route`), configs not edited (`numberOfRuns` is 1 in the files, 3 from the command line `--numberOfRuns=3`). Commit under test: 6d4d012 (clean tree, `pnpm build` run just before). Measured 2026-11-01, mobile 19:32 to 19:40, desktop 19:40 to 19:47. Both folders hold 36 reports (12 URLs x 3). Summary script: `summarize2.mjs` in the scratch folder (medians; LCP element from the median-LCP run). Times in ms, rounded to whole ms (raw medians are in the summary files; TBT is shown as measured). Raw results: `scratch\final\mobile\.lighthouseci`, `scratch\final\desktop\.lighthouseci`; summaries `scratch\final-mobile-summary.txt`, `scratch\final-desktop-summary.txt` (scratch = `C:\Users\Bryan\AppData\Local\Temp\claude\k--Development-SlpSoftware-Projects-AndreaSlopFotografie\bb0a550e-5427-462f-b99b-f9db012a172a\scratch`). **Sources of the other columns**: master and A (unit 1) come from `construction/start-migration/code/measurement-gate.md`; U2 comes from `construction/display-fallbacks/code/measurement-display.md` (which repeats master and A); final is this run. **Notes on the conditions** - These runs have gzip on the preview server; test and production differ until the config is placed (the generated nginx block and the IIS config are not active on this PC). - The not-found page: master measured `/bestaat-niet` (an unknown URL with a real 404 response on master), units 1 and 2 measured `/404.html` (served with status 200 by the old preview rule), this run measures the same page through `/__audit/404` (the opt-in audit route serves it with 200; `/404.html` itself is now a real 404). The master row is a different URL and rule. - The light logo changed after the unit 2 measure: it is lazy only until `load`, then eager (the "Fixed afterwards" item in `measurement-display.md`). U2 figures are from before that fix; the final figures include it. - Lighthouse simulates the connection; observed LCP on localhost is 53 to 110 ms (see the summary files). ## Mobile | URL | Score master / A / U2 / final | LCP master / A / U2 / final | FCP master / A / U2 / final | TBT final | CLS final | Script transfer final (bytes) | LCP element (final) | |---|---|---|---|---|---|---|---| | `/` | 0.75 / 0.85 / 0.85 / **0.85** | 4293 / 4280 / 4356 / **4357** | 2595 / 1358 / 1359 / 1361 | 74.5 | 0.001 | 222,663 | heading `span` in `h1` | | `/over-mij` | 0.86 / 0.90 / 0.92 / **0.92** | 3535 / 3608 / 3233 / **3233** | 2597 / 1357 / 1359 / 1359 | 66.5 | 0.002 | 211,482 | `img` portrait (`andrea-portret`) | | `/portfolio` | 0.90 / 0.94 / 0.93 / **0.93** | 2926 / 3007 / 3232 / **3232** | 2517 / 1358 / 1357 / 1358 | 61.5 | 0.001 | 213,338 | `img` tile 1 (`f41a1916`) | | `/soorten-fotoshoots` | 0.85 / 0.89 / 0.92 / **0.92** | 3750 / 3683 / 3306 / **3306** | 2592 / 1357 / 1357 / 1357 | 62.5 | 0.009 | 213,204 | `img` newborn | | `.../newborn` | 0.85 / 0.90 / 0.91 / **0.91** | 3525 / 3610 / 3456 / **3457** | 2591 / 1358 / 1357 / 1359 | 66.5 | 0.021 | 222,192 | `img` newborn (first story photo) | | `.../zwangerschap` | 0.84 / 0.88 / 0.90 / **0.90** | 3676 / 3834 / 3607 / **3606** | 2591 / 1357 / 1358 / 1357 | 64 | 0.006 | 221,681 | `img` zwangerschap (first story photo) | | `.../gezin` | 0.84 / 0.88 / 0.89 / **0.89** | 3678 / 3908 / 3681 / **3681** | 2592 / 1357 / 1357 / 1357 | 63 | 0.002 | 221,573 | `img` gezin (first story photo) | | `.../loveshoot` | 0.82 / 0.86 / 0.88 / **0.88** | 3768 / 4133 / 3831 / **3831** | 2590 / 1357 / 1357 / 1358 | 67 | 0.002 | 221,599 | `img` loveshoot (first story photo) | | `.../portret` | 0.84 / 0.89 / 0.90 / **0.90** | 3673 / 3759 / 3607 / **3606** | 2590 / 1358 / 1358 / 1358 | 67 | 0.002 | 221,539 | `img` portret (first story photo) | | `/contact` | 0.87 / 0.94 / 0.94 / **0.94** | 3345 / 3007 / 3007 / **3007** | 2725 / 1357 / 1357 / 1357 | 65 | 0.001 | 225,702 | paragraph `p` | | `/privacyverklaring` | 0.90 / 0.96 / 0.96 / **0.96** | 2844 / 2630 / 2780 / **2780** | 2513 / 1207 / 1357 / 1357 | 61.5 | 0.001 | 204,814 | lazy footer logo `img` | | Not found (`/__audit/404`; master `/bestaat-niet`, A and U2 `/404.html`) | 0.91 / 0.97 / 0.97 / **0.97** | 2603 / 2556 / 2629 / **2630** | 2406 / 1208 / 1356 / 1357 | 61 | 0.001 | 202,587 | lazy footer logo `img` | Scores of the three runs per URL are identical or within 0.01 (`/over-mij` 0.92, 0.93, 0.92; `/portfolio` 0.93, 0.91, 0.93; `/contact` 0.94, 0.94, 0.95; `/soorten-fotoshoots/portret` 0.90, 0.89, 0.90; not found 0.97, 0.96, 0.97). ## Desktop | URL | Score master / A / U2 / final | LCP master / A / U2 / final | FCP master / A / U2 / final | TBT | CLS final | LCP element (final) | |---|---|---|---|---|---|---| | `/` | 0.99 / 0.99 / 0.99 / **0.99** | 915 / 866 / 866 / **866** | 579 / 369 / 370 / 369 | 0 | 0.001 | heading `span` | | `/over-mij` | 0.99 / 1.00 / 1.00 / **1.00** | 770 / 752 / 708 / **707** | 579 / 368 / 369 / 368 | 0 | 0.003 | `img` portrait | | `/portfolio` | 0.99 / 0.99 / 0.99 / **0.99** | 868 / 887 / 931 / **931** | 557 / 367 / 368 / 367 | 0 | 0.004 | `img` in `picture` (tile) | | `/soorten-fotoshoots` | 0.99 / 1.00 / 1.00 / **1.00** | 770 / 769 / 706 / **707** | 578 / 367 / 368 / 368 | 0 | 0.014 | `img` newborn | | `.../newborn` | 0.99 / 1.00 / 1.00 / **1.00** | 748 / 809 / 747 / **747** | 577 / 367 / 368 / 368 | 0 | 0.003 | `img` newborn | | `.../zwangerschap` | 0.99 / 1.00 / 1.00 / **1.00** | 769 / 809 / 767 / **766** | 577 / 367 / 368 / 367 | 0 | 0.002 | `img` zwangerschap | | `.../gezin` | 0.99 / 1.00 / 1.00 / **1.00** | 768 / 808 / 767 / **767** | 577 / 367 / 368 / 368 | 0 | 0.003 | `img` gezin | | `.../loveshoot` | 0.99 / 0.99 / 1.00 / **1.00** | 767 / 850 / 786 / **786** | 577 / 367 / 368 / 368 | 0 | 0.003 | `img` loveshoot | | `.../portret` | 0.99 / 1.00 / 1.00 / **1.00** | 767 / 808 / 766 / **766** | 577 / 367 / 367 / 367 | 0 | 0.002 | `img` portret | | `/contact` | 0.99 / 1.00 / 1.00 / **1.00** | 783 / 728 / 708 / **708** | 592 / 367 / 367 / 367 | 0 | 0.002 | `img` in `picture` (aside photo) | | `/privacyverklaring` | 1.00 / 1.00 / 1.00 / **1.00** | 641 / 586 / 625 / **625** | 553 / 328 / 367 / 368 | 0 | 0.002 | `h1` | | Not found | 1.00 / 1.00 / 1.00 / **1.00** | 577 / 589 / 626 / **630** | 525 / 327 / 367 / 369 | 0 | 0.001 | `n/a` (the LHR has no LCP-element item) | Desktop is 0.99 or better on all 12 URLs; no URL is below 0.9. ## Accessibility, best practices, SEO Mobile and desktop, all 3 runs of all 12 URLs: accessibility 1 and best practices 1 on every URL. SEO is 1 on the ten indexed pages. Two pages score lower on SEO, identically in every run and both presets: `/privacyverklaring` 0.69 and not found (`/__audit/404`) 0.58. This is the `is-crawlable` audit, which the config lets fail on purpose (its SEO assertions are `document-title`, `link-text`, `crawlable-anchors`, `image-alt`, plus `meta-description` and `canonical` for privacy). No run had a runtime error; every `finalUrl` equals the requested URL. V9 (not-found page, median run): no runtime error, best practices 1, accessibility 1. ## Assertion check against the unchanged targets Command: `pnpm exec lhci assert --config=lighthouserc.cjs` (mobile) and `--config=lighthouserc.desktop.cjs` (desktop) on the collected folders (copied into `.lighthouseci` in `web/` for the check and removed again). The plan names no separate command; this is the assertion step of `pnpm lighthouse`. `lhci assert` aggregates runs with its default (optimistic: best run); `--aggregationMethod` cannot be combined with `assertMatrix` (lhci errors out), so a pessimistic aggregation could not be run. The median figures were compared with the targets by hand below. - Mobile: exit 0, "Checking assertions against 12 URL(s), 36 total run(s) ... All results processed!". All 12 URLs pass. - Desktop: exit 0, same output. All 12 URLs pass. - By hand, median score against the current mobile target: `/` 0.85 vs 0.8, `/over-mij` 0.92 vs 0.85, `/soorten-fotoshoots` 0.92 vs 0.85, newborn 0.91 vs 0.85, zwangerschap 0.90 vs 0.85, gezin 0.89 vs 0.85, loveshoot 0.88 vs 0.85, portret 0.90 vs 0.85, `/portfolio` 0.93 vs 0.9, `/contact` 0.94 vs 0.9, `/privacyverklaring` 0.96 vs 0.9, not found 0.97 vs 0.9: all pass. Desktop: all at or above 0.9. Targets unchanged: `lighthouserc.cjs` SHA-256 `6e146472ed7a3a0ebf69bfc1d357f04e1c0bca36e80a10a13991ee1188703ded` and `lighthouserc.desktop.cjs` `f13d993389babd384177a0a1c601547b356d894ae97ac12177b09229a5318e04`: equal to the values recorded at Step 21 (`6e146472ed7a...` and `f13d993389ba...`). (The `lighthouserc.cjs` checksum in `baseline-checks.md`, `717c80a2...`, is from Step 1, before the Step 11 restructure, and is not the Step 21 value.) `git status --short` after the measurement equals the one before (empty). ## Decision table: mobile pages below 0.90 Current mobile exceptions in `lighthouserc.cjs`: 8 entries, all with the decision text "tijdelijk, tot de eindmeting van prerendering": `/` 0.8, the other seven pages 0.85. | Page | Score (final) | LCP | LCP element | Current exception | Options for the owner | |---|---|---|---|---|---| | `/` | 0.85 | 4357 | heading `span` in `h1` | 0.8 | keep an exception at a stated floor; work on the page; accept a red `pnpm check` | | `/soorten-fotoshoots/gezin` | 0.89 | 3681 | `img` gezin (first story photo) | 0.85 | same three options | | `/soorten-fotoshoots/loveshoot` | 0.88 | 3831 | `img` loveshoot (first story photo) | 0.85 | same three options | Number of pages below 0.90 on mobile: 3. Two pages sit exactly at 0.90 (`/soorten-fotoshoots/zwangerschap` 0.90, 0.90, 0.90; `/soorten-fotoshoots/portret` 0.90, 0.89, 0.90): with `numberOfRuns: 1` they can land on either side of 0.9 by chance. `/soorten-fotoshoots/newborn` (0.91, 0.91, 0.91) is one point above. ## Pages at 0.90 or higher that still carry an exception (these would lose it) | Page | Score (final) | Current exception | |---|---|---| | `/over-mij` | 0.92 | 0.85 | | `/soorten-fotoshoots` | 0.92 | 0.85 | | `/soorten-fotoshoots/newborn` | 0.91 | 0.85 | | `/soorten-fotoshoots/zwangerschap` | 0.90 | 0.85 | | `/soorten-fotoshoots/portret` | 0.90 | 0.85 | Pages without an exception today (`/portfolio` 0.93, `/contact` 0.94, `/privacyverklaring` 0.96, not found 0.97) are unaffected. ## JavaScript (gzip, `dist/client`, commit 6d4d012; script `gzip-moved.mjs`) | Measure | Master | A | U2 | Final | |---|---|---|---|---| | Entry JavaScript on `/` | 181,161 | 206,550 | 208,100 | 208,103 | | All JavaScript | 284,705 | 305,234 | 306,472 | 306,477 | | Moved inline scripts, all 16 files | none | 9,531 | not recorded | 9,543 | Script transfer on mobile (median, bytes): `/` 222,663 (U2 222,660), `/over-mij` 211,482 (U2 211,506), `/contact` 225,702 (U2 225,724). ## CLS All mobile and desktop CLS values are below 0.01 except: `/soorten-fotoshoots/newborn` mobile 0.021 (equal to master and A) and `/soorten-fotoshoots` desktop 0.014. Mobile `/soorten-fotoshoots` is 0.009. ## State after the measurement Port 4173 was free before and after; no process left running; `.lighthouseci` removed from `web/`; `dist` built from the committed code just before the runs and not changed since; `git status --short` empty, as before. No git write command was used. ## Owner decisions Recorded 2026-11-01 by the owner, after this report. Applied in `web/lighthouserc.cjs` (`mobileExceptions`); `lighthouserc.desktop.cjs` and `numberOfRuns: 1` are unchanged. | Page | Score (final) | Decision | Mobile target now | |---|---|---|---| | `/over-mij` | 0.92 | no exception | 0.9 | | `/soorten-fotoshoots` | 0.92 | no exception | 0.9 | | `/soorten-fotoshoots/newborn` | 0.91 | no exception | 0.9 | | `/soorten-fotoshoots/zwangerschap` | 0.90 | exception stays at 0.85: "scoort rond 0,90; met 1 run onbetrouwbaar als doel (eigenaar, 2026-11-01)" | 0.85 | | `/soorten-fotoshoots/portret` | 0.90 | same as zwangerschap | 0.85 | | `/` | 0.85 | floor raised from 0.8 to 0.85: "bij de huidige score vastgelegd, tot responsive foto's (eigenaar, 2026-11-01)" | 0.85 | | `/soorten-fotoshoots/gezin` | 0.89 | stays at 0.85, same text as `/` | 0.85 | | `/soorten-fotoshoots/loveshoot` | 0.88 | stays at 0.85, same text as `/` | 0.85 | `/portfolio`, `/contact`, `/privacyverklaring` and the not-found page keep 0.9. The check above was repeated with the new config: `lhci assert` on the same 36 mobile and 36 desktop reports: all 12 URLs pass in both presets (no new collect). The median figures are all at or above the new targets (lowest margin: `/` 0.85 at floor 0.85).