Automate a weekly PageSpeed report for your key pages
Site speed drifts as pages gain scripts and images, and nobody tests again until it hurts. A scheduled agent runs PageSpeed Insights weekly on your key pages, compares scores with last week, and drafts a report you approve.
What you’ll set up
Connections
Works from public sources: Google PageSpeed Insights for the public URLs you list
Upgrades
Delivery (channel or inbox)
Delivers the approved output to your channel or inbox.
Inputs
Pages to test · Required
List, no default
Device · Required
Choice of mobile or desktop, default mobile
Recent releases or changes shipped · Optional
Text, no default
Ready by · Required
Time, default 09:00
Brief · Required
Text, prefilled and editable during setup.
Runs
Runs every Monday at 08:00
Done by Research analyst
You approve before anything goes out.
You approve every report · Free to start
How it runs
One agent does the work. Its department head reviews it, then you approve.
- 1Research analyst
Runs every Monday at 08:00 and prepares the report from your connected sources.
- 2CTO · Review
The department head reviews the report before it reaches you.
- 3You
You approve before anything goes out.
What you get
Sample, not live data. URLs and scores are invented.
Page speed, week of 3 Nov (sample, mobile)
/ (home) Performance 91 (was 93). LCP 2.1 s, CLS 0.02, INP 160 ms. /pricing Performance 72 (was 84). LCP 3.9 s. Worse: new chat widget script (from your notes). /blog Performance 88 (was 88).
Changes within 3 points are treated as noise.
Needs your approval before it is sent.
Why speed drifts
Each added tracking script, hero image or font costs milliseconds, and no single change looks harmful. PageSpeed Insights reports both field data, from real Chrome users over a trailing 28 days when enough exists, and lab data from Lighthouse in a simulated load. Running it once after a redesign tells you little; running it weekly shows the trend and which release caused a drop.
How the scheduled job works
You list the pages to test and choose mobile or desktop. Each Monday a scheduled agent calls the PageSpeed Insights API for each URL and records the Lighthouse category scores, plus the Core Web Vitals field data when Google has it: Largest Contentful Paint, Cumulative Layout Shift and Interaction to Next Paint. It compares with last week's report and lists the pages that got worse, by how much; if that report is not available it records a baseline. Lab scores vary between runs, so small moves are marked as noise. The agent reads public pages and changes nothing on your site.
Review and approval
The CTO agent, as department head, reviews the report before it reaches you: is a drop larger than normal run-to-run variation, is field data confused with lab data, is a page missing because a test failed. Then you approve, edit or skip it, and nothing is sent until you do. Lighthouse is a simulation, so treat the score as a trend and use field data when it exists.
Acting on the numbers
Pick one regression a week and fix it, usually an oversized image or a new third-party script. Write shipped changes in the releases field so the report can link a drop to a change. Test the pages that earn traffic, not every URL.
Questions
Related jobs
Works from public sources: Your public sitemap.xml and the pages it lists
Automate a weekly broken link check for your website
A weekly list of broken and redirecting links on your site, with the page each one sits on.
Connects to: Search Console read access (granted by you) (required)
Automate a weekly Google indexing check for your key pages
A weekly check of whether your important pages are indexed, with Google's stated reason when they are not.
Works from public sources: GitHub REST API: public repository star counts and activity
Automate a weekly GitHub stars and activity report for your repos
A weekly report of star growth and recent activity for the public repositories you list.