Content

Convert a spreadsheet of data into a narrative blog post

Finds the actual story inside a raw benchmark spreadsheet instead of narrating every row, while flagging outlier data points and sample-size limitations before they get written up as settled fact.

Works with Claude / GPT850 uses 4.3

The prompt

content-spreadsheet-data-to-narrative-post
You write for a WordPress hosting review site. Your team ran a benchmark, loading
the same test site from 8 hosting providers across 3 regions, and logged the
results in a spreadsheet. Now it needs to become a readable post, not a re-typed
table.

SPREADSHEET DATA: [
Host A: US 1.2s, EU 1.4s, Asia 2.1s
Host B: US 1.1s, EU 1.3s, Asia 1.9s
Host C: US 2.8s, EU 3.0s, Asia 45.2s
Host D: US 0.9s, EU 1.0s, Asia 1.5s
(4 more hosts omitted for brevity, similar pattern)]
TEST METHODOLOGY: [single test site, tested once per region at one time of day, not
averaged across multiple runs]
GOAL: [a post ranking the hosts, readable by non-technical store owners]

Work through this before drafting:
1. Find the actual story instead of narrating every cell: the headline pattern here
   is that Host D is consistently fastest across all three regions, and Host C is
   fine in the US/EU but has a number that doesn't belong (45.2s in Asia against
   everything else under 3s).
2. Edge case: that Host C Asia number looks like a data error or a one-off server
   issue, not a real, representative result, a 15-20x jump from every other data
   point is a red flag. Don't write it up as "Host C is unusable in Asia" without
   flagging it, note that it should be re-tested before being stated as fact, and
   either exclude it or clearly caveat it in the post.
3. Given the methodology note (single test per region, not averaged), don't state
   results with false precision ("Host D is exactly 23% faster than Host B"). Use
   ranged, honest language ("noticeably faster," "roughly a third quicker") and
   disclose the single-run limitation near the methodology section rather than
   burying it in a footnote.
4. Decide what belongs in a table (the raw region-by-region numbers, for readers who
   want to check a specific host) versus what belongs in prose (the overall ranking
   story and the practical recommendation), don't force every number into narrative
   sentences.
5. Close with a practical recommendation matched to the audience (which host to pick
   for a store whose customers are mostly in a specific region), not just a ranked
   list with no interpretation.

Draft the full post: intro stating what was tested and why it matters, the
narrative ranking section, a results table, a methodology/limitations note, and a
closing recommendation.

Replace the bracketed data, methodology, and goal above with your own.

Raw benchmark data almost always contains one number that doesn’t belong, and the instinct under deadline is to write around it rather than flag it. Catching the outlier before publish protects the whole post’s credibility, since one obviously wrong stat makes readers doubt every other number in the table too.