# Measurement at the Decision Gate - Unit 1 Compares master (c4c2f83, see [measurement-baseline.md](measurement-baseline.md)) with unit 1 (8dc120e). Requirements: NFR-PR-P-03, P-04, Q-04; rules BR-MEA-02 to 05. **Method**: `lhci collect`, 3 runs per URL, median, same PC, Node 24.16.0 and summary script. Measured on 2026-10-31. **Differences in setup**: - Baseline ran on `vite preview`, unit 1 on `config/previewServer.ts`. Both compress text responses. - The not-found URL is `/bestaat-niet` in the baseline and `/404.html` in A and B, because Lighthouse refuses a page with status 404. - **B** forces `Reveal` visible through a temporary, uncommitted change to `useReveal.ts`. It is only an indication of unit 2 (BR-MEA-05). It has no photo preload and is not unit 2's real behaviour. ## Mobile | URL | Score base | Score A | Score B | FCP base | FCP A | LCP base | LCP A | LCP B | TBT base | TBT A | CLS base | CLS A | |---|---|---|---|---|---|---|---|---|---|---|---|---| | `/` | 0.75 | 0.85 | 0.85 | 2595 | 1358 | 4293 | 4280 | 4281 | 293 | 67 | 0.0011 | 0.0012 | | `/over-mij` | 0.86 | 0.90 | 0.92 | 2597 | 1357 | 3535 | 3608 | 3306 | 132 | 65 | 0.0002 | 0.0002 | | `/portfolio` | 0.90 | 0.94 | 0.90 | 2517 | 1358 | 2926 | 3007 | 3607 | 126 | 64 | 0.0012 | 0.0012 | | `/soorten-fotoshoots` | 0.85 | 0.89 | 0.92 | 2592 | 1357 | 3750 | 3683 | 3383 | 127 | 63 | 0.0070 | 0.0070 | | `.../newborn` | 0.85 | 0.90 | 0.91 | 2591 | 1358 | 3525 | 3610 | 3457 | 151 | 65 | 0.0205 | 0.0205 | | `.../zwangerschap` | 0.84 | 0.88 | 0.89 | 2591 | 1357 | 3676 | 3834 | 3682 | 142 | 62 | 0.0056 | 0.0056 | | `.../gezin` | 0.84 | 0.88 | 0.89 | 2592 | 1357 | 3678 | 3908 | 3758 | 148 | 61 | 0.0023 | 0.0023 | | `.../loveshoot` | 0.82 | 0.86 | 0.88 | 2590 | 1357 | 3768 | 4133 | 3907 | 188 | 61 | 0.0021 | 0.0021 | | `.../portret` | 0.84 | 0.89 | 0.89 | 2590 | 1358 | 3673 | 3759 | 3682 | 140 | 65 | 0.0024 | 0.0024 | | `/contact` | 0.87 | 0.94 | 0.94 | 2725 | 1357 | 3345 | 3007 | 3006 | 114 | 64 | 0.0009 | 0.0012 | | `/privacyverklaring` | 0.90 | 0.96 | 0.97 | 2513 | 1207 | 2844 | 2630 | 2630 | 164 | 66 | 0.0011 | 0.0008 | | Not found | 0.91 | 0.97 | 0.97 | 2406 | 1208 | 2603 | 2556 | 2556 | 171 | 62 | 0.0014 | 0.0009 | Times in ms. Reading of base against A: - **Score**: better on all 12 URLs. - **FCP and TBT**: better on all 12 URLs. FCP drops from about 2.6 s to 1.36 s; TBT roughly halves or better. - **LCP**: better on 5 URLs (`/`, by 13 ms, `/soorten-fotoshoots`, `/contact`, `/privacyverklaring`, not found) and worse on 7: `/over-mij` +73, `/portfolio` +81, newborn +85, zwangerschap +158, gezin +230, loveshoot +365, portret +86. - **Against 0.9**: six URLs reach it in A (`/over-mij`, `/portfolio`, newborn, `/contact`, `/privacyverklaring`, not found). Below 0.9: `/`, `/soorten-fotoshoots`, zwangerschap, gezin, loveshoot, portret. ## Desktop | URL | Score base | Score A | Score B | FCP base | FCP A | LCP base | LCP A | |---|---|---|---|---|---|---|---| | `/` | 0.99 | 0.99 | 0.99 | 579 | 369 | 915 | 866 | | `/over-mij` | 0.99 | 1.00 | 1.00 | 579 | 368 | 770 | 752 | | `/portfolio` | 0.99 | 0.99 | 0.99 | 557 | 367 | 868 | 887 | | `/soorten-fotoshoots` | 0.99 | 1.00 | 1.00 | 578 | 367 | 770 | 769 | | `.../newborn` | 0.99 | 1.00 | 1.00 | 577 | 367 | 748 | 809 | | `.../zwangerschap` | 0.99 | 1.00 | 1.00 | 577 | 367 | 769 | 809 | | `.../gezin` | 0.99 | 1.00 | 1.00 | 577 | 367 | 768 | 808 | | `.../loveshoot` | 0.99 | 0.99 | 1.00 | 577 | 367 | 767 | 850 | | `.../portret` | 0.99 | 1.00 | 1.00 | 577 | 367 | 767 | 808 | | `/contact` | 0.99 | 1.00 | 1.00 | 592 | 367 | 783 | 728 | | `/privacyverklaring` | 1.00 | 1.00 | 1.00 | 553 | 328 | 641 | 586 | | Not found | 1.00 | 1.00 | 1.00 | 525 | 327 | 577 | 589 | Scores are equal or better; FCP is better everywhere. LCP is at most 83 ms worse, on loveshoot. ## JavaScript | Measure (gzip) | Base | A | Difference | |---|---|---|---| | Entry JavaScript on `/` | 181,161 bytes | 206,550 bytes | +25,389 bytes (+14%) | | All JavaScript | 284,705 bytes | 305,234 bytes | +20,529 bytes (+7%) | | Moved inline scripts, all 16 files | none | 9,531 bytes | Included in all JavaScript | Script transfer on mobile rises by about 7 to 10 kB per page, for example `/` from 214 kB to 221 kB and `/over-mij` from 202 kB to 210 kB. ## Moved scripts (BR-MEA-03) Three per page, five on `/contact`. Single mobile runs of `/` and `/over-mij` show them finished at about 20 to 45 ms of observed time, before FCP (about 1357 ms) and LCP. No sign of delayed first paint, and FCP is better than the baseline on every page (NFR-PR-P-04). ## CLS (BR-MEA-04, V10) Below 0.01 on `/` and `/contact`, mobile and desktop (`/` 0.0012 and 0.0009; `/contact` 0.0012 and 0.0019). The e-mail placeholders pass. Outside this rule, the existing values on newborn (0.0205 mobile) and `/soorten-fotoshoots` (0.0125 desktop) equal the baseline. ## Not found page `/404.html` scores 0.97 and 1.00 and gave no errors in the Lighthouse logs. Whether the Lighthouse assertions on best practices pass follows from `pnpm lighthouse`, which `collect` does not assert. ## State after the measurement `useReveal.ts` restored and checksum verified (015643dc...), `git status` clean, `dist` rebuilt from the committed code.