Score every row
One result for each row.
Use it when every row is its own thing: a lead, a review, a ticket, a customer.
You get: a score and a short reason in new columns on every row.
Sheet Scorer CSV and Excel
Put a score on every row of your spreadsheet.
Upload a file, say what good looks like, and download it back with a score and a short reason on every row. No formulas, and no reading two thousand lines yourself.
Already have an account? Log in
- In: a CSV or Excel file, read in your browser
- You choose: the columns, the scale and what each level means
- Out: your file with the new columns added
See it work
A run, start to finish
An example on made-up data: 40 sales leads from areas of the UK, using the Lead quality recipe. The scores were written by hand to show the format, so this is not output from the AI. It is a replay: it does not use your data or your allowance.
| Company | Lead quality | Lead quality reason |
|---|---|---|
| Ardwick Print Group | High | Director with £40k budget, 1 April deadline and a demo requested for next week. |
| Oldham Logistics | High | Comparing three systems for Q1 and asking for 30-driver pricing. |
| Belmont Joinery | Low | Two-word message with no need, size or timeline. |
| Newton Aycliffe Pest Control | Low | Unrelated promotion for watches; clearly spam. |
| Wolverhampton Windows | Low | Reseller seeking partnership terms, not a customer. |
| Canterbury Cookware | Low | Selling sponsored newsletter posts, not buying. |
| Ashford Auto Parts | Medium | Head office mandate and a January supplier meeting, but no budget stated. |
| Granton Freight | Medium | Planning a trial pending an API answer; no timeline or budget. |
This is a short preview of the results. Your own run lists every row, with search, filters and download.
Sample data, made up for this page. The scores were written by hand to show what results look like. They say nothing about how the AI will score your own file; the test-on-a-few-rows step is there for that.
How it works
Three steps, and you stay in charge
-
Drop in your file
A CSV or Excel file. It is read in your browser and you see your columns straight away. You choose which ones the AI gets to read.
-
Say how to score it
Pick one of 14 ready-made recipes, or describe your own in a few lines. Choose the scale, then test it on a few rows before it uses any of your allowance.
-
Run it and download
Watch the results arrive, cancel whenever you like, and download your file with the new columns. Keep the tab open while it runs.
Three ways to score
Pick the one that matches your question
Do you want an answer for each row, for each group of rows, or both? You can change your mind between files.
Score each group
One result for each group of rows.
Pick a column to group by, such as Sales Area or Customer. The AI reads all the rows in a group together and gives the group a score and a summary.
You get: one row per group. You can also add each group's result back onto every row in it. Very large groups are read from an even sample, and we say so.
Score rows, then roll up
Every row scored, then totals for each group.
Each row gets its own score first. Then each group gets its totals, plus an optional one-paragraph summary. The totals are worked out by code, not guessed by the AI.
You get: a score on every row, plus a table of totals for each group.
Recipes
14 ready-made recipes, or write your own
A recipe is a saved set of instructions for the AI: what you are scoring, what each level means, and which columns matter. You can change any recipe before you run it.
-
Sales
Lead quality
Rate how likely each enquiry is to turn into a paying customer.
LowMediumHigh Every row -
Customer feedback
Review sentiment
Tell whether each review or comment is positive, neutral or negative.
NegativeNeutralPositive Every row -
Support
Support ticket priority
Rank support tickets so the most urgent ones are handled first.
LowMediumHighUrgent Every row -
Customer success
Churn risk
Flag customers who look likely to cancel or leave soon.
LowMediumHigh Every row -
Hiring
CV and resume fit
Score how well each candidate's CV matches a role you describe.
12345 Every row -
Marketing
Content quality
Score articles, product descriptions or posts for clarity and usefulness.
12345 Every row -
Risk
Spam or fraud likelihood
Spot entries that look like spam, scams or fake submissions.
LowMediumHigh Every row -
Data
Data quality check
Flag records with missing, inconsistent or implausible values.
NoYes Every row -
Customer feedback
Review helpfulness
Score how useful each product review is to someone deciding what to buy.
12345 Every row -
Customer feedback
NPS follow-up urgency
Decide which survey respondents deserve a personal follow-up first.
RoutineSoonUrgent Every row -
Sales
Sales area health
Summarise each sales area from its leads and say how healthy it looks.
At riskSteadyStrong Each group -
Customer success
Account health by customer
Read all notes for each customer account and rate how the relationship looks.
At riskWatchHealthy Each group -
Customer feedback
Comment themes to sentiment
Score every comment, then see sentiment totals and a summary for each theme.
NegativeNeutralPositive Rows + roll-up -
Customer feedback
Product reviews: score and average
Score each review 1 to 5, then get the average and a summary for each product.
12345 Rows + roll-up -
Your own
Write your own
Say what you are scoring and what each level means, in your own words. Test it on a few rows, tighten it, and run it. Basic can save 10 of your rubrics to reuse, and Pro 50.
What changes in your file
The same file, with answers beside every row
| Your file, as you uploaded it (more columns hidden here) | Added by Sheet Scorer | |||
|---|---|---|---|---|
| Company | Lead quality | Lead quality reason | ||
| Ardwick Print Group | High | Director with £40k budget, 1 April deadline and a demo requested for next week. | ||
| Framwellgate Dental | Medium | Real need, but 'not urgent' and 'nothing is decided', with no timeline. | ||
| Digbeth Vintage | Low | Free trial request from a 2-person shop who is 'not sure I'll need it'. | ||
| Medway Marine Services | High | MD with an £8k budget, summer deadline and a call requested for Tuesday. | ||
- New columns go on the right of your own, and you choose what they are called.
- A short reason beside each score, so you can see why and disagree.
- Group modes add group columns, or give you one row per group, whichever you prefer.
Sample data, made up for this page, with scores written by hand to show the format.
Limits and pricing
Honest limits, by plan
This table is generated from the same limits the tool enforces, so what you read here is what you get. You can watch the example run without an account; scoring your own file needs a free one.
Feature
Free
$0
Basic
$9.99 /month
Pro
$29.99 /month
Rows scored per month
Resets on the 1st of each month, UK time.
150
10,000
60,000
Rows in one file
The most one file can hold.
100
2,500
20,000
Test runs per day
Try a recipe on a few rows first. These do not use your monthly rows.
10
40
100
Saved rubrics
Keep your own recipes to reuse.
Not included
10
50
Batches scored at once
More at once means a big file finishes sooner.
2
4
6
Runs open at once
1
2
3
Results kept for
Then deleted automatically. You can delete sooner.
7 days
30 days
90 days
- All three modes and every recipe are on every plan, including Free.
- CSV and Excel download on every plan.
- Rows that fail are not counted, and you can re-run them on any plan.
- In group modes, rows are counted by the rows we read, not by the number of groups.
Guests can watch the example run; running your own file needs an account.
Your first month of Pro is free with code FREE at checkout. See Pro
Your data
What happens to your file, in plain English
Read in your browser
Your file is opened on your own computer. Only the rows and columns you choose are sent to be scored.
Scored by Claude
Those rows go to Anthropic's AI service (Claude), and the results are stored on The Prompt Index while you work.
Deleted on a timer
We keep your file and results for 7 days on Free, 30 on Basic and 90 on Pro, then delete them automatically.
Delete it yourself
Every run has a delete button, and it really deletes: the rows and the results.
We do not use your file to train models. Please do not upload data you are not allowed to process, such as other people's sensitive personal data. You will be asked to agree to this before your first run. Read the privacy policy.
Questions
Things people ask
What files can I use, and how big can they be?
CSV and Excel files. The number of rows in one file depends on your plan: 100 rows on Free, 2,500 on Basic and 20,000 on Pro. Your file is read in your browser, and you choose which columns the AI gets to see.
What do I get back?
Your own columns, followed by new ones: a score, and if you want them a short reason and a confidence level. You decide the scale (Low, Medium, High; Yes or No; True or False; 0 to 10; 1 to 5; or your own labels) and what the new columns are called. Download the result as CSV or Excel.
How accurate is it?
It is an AI reading your text, so it will sometimes disagree with you. That is why you can test a recipe on a few rows first, see a short reason beside every score, and re-run with a sharper description of what each level means. Use the scores as a fast first pass for sorting and prioritising, and check the rows that matter.
How are rows counted against my allowance?
One row read is one row used. In the group modes we count the rows we read, not the number of groups. Rows that are completely blank, rows that fail, and rows you cancel before they are scored are not counted. Your allowance is 150 rows a month on Free, 10,000 on Basic and 60,000 on Pro, and it resets at the start of each calendar month, UK time. Testing a recipe on a few rows does not use your monthly allowance.
What if a row fails, or I close the tab?
Scoring is driven from your open browser tab, so keep it open while a run is going. If you close it, the run stops where it was and you can pick it up again from your history; everything already scored is kept. Any rows that fail can be re-run on every plan, and failed rows are not counted.
Is my data kept, and who sees it?
The rows and columns you choose are sent to Anthropic's AI service (Claude) to be scored, and the results are stored on The Prompt Index while you work. We keep your file and results for 7 days on Free, 30 on Basic and 90 on Pro, then delete them automatically, and you can delete any run immediately. We do not use your file to train models. Please do not upload data you are not allowed to process, such as other people's sensitive personal data.
Can text inside a cell tell the AI what score to give?
It is not meant to. Everything in your cells is treated as data to be scored, not as instructions, and if a cell looks like it is trying to instruct the AI the row is flagged so you can see it. No AI system is perfectly immune to this, so look again at any row where a score surprises you.
Is the downloaded file safe to open in Excel?
Yes. If a cell starts with =, +, - or @ (which a spreadsheet could try to run as a formula), the download puts an apostrophe in front of it so it stays plain text. Long cells are cut to a generous number of characters for the AI, and when that happens the row is marked "truncated" so it is never silent.
Ready to try it on your own file?
Make a free account, drop in a CSV or Excel file, and test a recipe on a few rows before you run anything.