The benchmark pages are 302 of the 756 URLs in plotly.com's main sitemap and the main proof that Plotly Studio works. The index says "Every day, we run a CI benchmark" and "Our 99.1% success rate", but its own summary shows 92.5% and "302 total attempts" beside "302 datasets × 3 attempts each". Each dataset page carries about 1,800 characters of text, "1 / 1 attempts passed" and "Generated On December 20, 2025". A skeptical analyst sees stale, inconsistent numbers, and search engines see 302 near-empty pages.
First action
Pick one success-rate number, date it, and fix the "302 total attempts" line on the index.
How to measure it
Studio downloads from benchmark pages
- Now
- –
- Target
- 3x
- By
- 2 quarters
Steps
Agree the numbers
One success rate, its date and method, from the CI run.
Product Manager, Plotly Studio 2 days
Connect the data
Pages pull each dataset's latest run and all attempts.
Front-End Web Developer 1 week
Design the page
Generated dashboard, chart choices, code excerpt, CTA.
Marketing Manager, Content & Web 1 week
Capture the dashboards
Screenshot or embed from each run.
Developer Experience Engineer 1 week
Write titles and intros
Search-led titles and two-paragraph intros, AI-drafted and edited.
Marketing Manager, Content & Web 2 weeks
Prune weak pages
Merge test files and duplicates into category pages.
Marketing Manager, Content & Web 3 days
Add the CTA
"Build this with your data" opens Studio with the dataset.
Product Manager, Plotly Studio 1 week
Measure
Indexed pages, search clicks, downloads per page.
Marketing Operations Manager weekly
Why it works
Feed the pages from the live daily run, so the index shows one success rate with its date and method, and every dataset page shows its latest run date and all three attempts. Add what a searcher came for: a screenshot or embed of the generated dashboard, the chart types Studio chose and why, a short code excerpt, and a "Build this with your data" button that opens Studio with the dataset. Title each page for the search ("Supermarket sales dashboard in Python, built by AI"), and fold the weakest test files into category pages.
Data people search for dashboards for specific datasets and trust measured results. Fresh, consistent numbers turn the benchmark into proof, and real content turns 302 thin pages into 302 entry points to Studio.
Ready to use
ChecklistBenchmark page template
[ ] One success rate, dated, with method link [ ] Latest run date and all three attempts [ ] Screenshot or embed of the generated dashboard [ ] Chart types chosen and why [ ] Short code excerpt [ ] "Build this with your data" button [ ] Search-led title and two-paragraph intro [ ] Dataset licence noted
Risks and how to handle them
A bad day lowers the published rate
Show a rolling 30-day rate with the date and method.
Dataset licences
Keep only datasets whose licences allow redistribution.
Evidence
- Your websitePlotly Studio Benchmarks page: "Every day, we run a CI benchmark" across real-world datasets, "Each dataset receives three generation attempts", and "Our 99.1% success rate means that you can trust Plotly Studio"; the summary on the same page shows "Success Rate 92.5%", "302 Apps Generated Daily (302 datasets × 3 attempts each)" next to "302 total attempts", and a median generation time of 98 s.Source
- Your websiteThe 302 individual benchmark pages (for example Supermarket Sales, NASA Global Landslide Catalog, Tesla Stock Price, Diabetes Health Indicators, San Francisco 311) each have about 1,800 characters of text: a one-line description, "1 / 1 ATTEMPTS PASSED", outline and generation times, "Generated On December 20, 2025" and "Download Dataset". Checked on five pages on October 23, 2026. They are 302 of the 756 URLs in plotly.com's main sitemap.Source
- HiringMarketing Manager, Content & Web (remote Canada, posted October 8, 2026): owns "end-to-end website copy, including homepages, product pages, solution pages, landing pages, blog posts"; will "Manage site-wide heading architecture ... to optimize SEO performance" and keep "a strong SEO site architecture"; runs the blog program and "daily social media content and posting cadence"; creates collateral from templates; supports "SEO/AEO execution by optimizing metadata, on-page content, and heading structures"; reviews AI-generated content "to match Plotly's tone and voice"; works with Marketing Operations, Front-End Web Development and Product.Source
- Your websitePlotly Studio page: "AI-Native Data Analytics, Visualization, and App Generation", a desktop app for macOS and Windows or "embedded in your Dash app"; connects to Snowflake, Databricks, BigQuery, Postgres, MySQL, SQL Server, MotherDuck, Oracle and CSV; "Publish directly to Plotly Cloud or Dash Enterprise with a single click".Source