Choose one of three approaches. The steps for that option appear below.
Choose an option above. On a phone, swipe the tabs left or right to see all options →
You check sources, rules and decisions. ChatGPT drafts only from the checked material.
Prepare the checked inputs
Prepare numbered criteria list
Label leads against written criteria
Check labels and record verdicts
Draft the screening summary
Approve or hold
Use: You
Have ready: the inbound lead export and the agreed fit criteria.
Use: You
Have ready: agreed fit criteria. Paste the written fit criteria as a numbered list and remove anything not written by sales leadership
Check the list matches the leadership page word for word and nothing new was added.
Use: ChatGPT
Ask ChatGPT to label every lead with fit, numbered-criterion reason and borderline flag.
What to paste into the tool
Task: Labelled table with reason and borderline column. Here is a lead export and numbered fit criteria. For each row give Fit, No fit or Borderline, and a reason that names the criterion number. Use only these criteria. Do not invent criteria. If the row lacks data for a criterion, mark Borderline and say which data is missing.
Use: You
Check each band against the criteria and fill a reviewer verdict column.
Check a sample from Fit, No fit and Borderline bands; fix or relabel any row that breaks the written criteria.
Use: ChatGPT
Ask ChatGPT for counts per label, rejected categories and open questions from the checked table.
What to paste into the tool
Task: Draft screening summary. From this reviewed table, write a short summary: counts per label, the reasons given for No fit grouped by criterion, and open questions. Report only what is in the table.
Use: You
Decide the borderline queue and sign off the list and summary.
Recount the summary numbers from the sheet. The human decision is final. No lead is contacted, rejected or disqualified in any system by this workflow.
You check sources, rules and decisions. Gemini drafts only from the checked material.
Prepare the checked inputs
Break criteria into yes/no questions
Answer each criterion per row
Set and check fit formula
Draft the screening summary
Approve or hold
Use: You
Have ready: the inbound lead export and the agreed fit criteria.
Use: You
Have ready: agreed fit criteria. Break each written criterion into one yes/no question with a clear data column to read
Each question must restate a written criterion only; drop any question that adds a new one.
Use: Gemini
Ask Gemini for yes, no or not enough data in one answer column per criterion.
What to paste into the tool
Task: One answer column per criterion. For this row, answer the question using only the Role, Company and Message cells. Reply Yes, No or Not enough data.
Use: You
Set all yes = Fit, any no = No fit, any not enough data = Borderline; join failed criteria as reasons.
Check the rule on five rows of each outcome and fix the formula if it mislabels. Open a sample of rows and check each yes/no against the row text.
Use: Gemini
Ask Gemini for checked counts per label, borderline rows and reasons, and totals.
What to paste into the tool
Task: Draft screening summary from the checked sheet. Using only this sheet, write a short screening summary: the count for each label, the borderline rows with their reasons, and the totals. Do not change any cell or label.
Use: You
Check the borderline queue, sampled rows, counts and reasons, then sign off the list and summary.
The human decision is final.
You check sources, rules and decisions. Claude drafts only from the checked material.
Prepare the checked inputs
Label batches with criteria, sample
Check sample agreement, each band
Draft the screening summary
Approve or hold
Use: You
Have ready: inbound lead list and agreed criteria. Hand-label about fifteen leads with fit labels and reasons, one per band where possible.
Use only the written criteria when labelling the sample.
Use: Claude
Paste the written fit criteria and your labelled sample into the chat. In the paste box, "The project holds the fit criteria and a labelled sample." becomes "The chat holds the fit criteria and a labelled sample." Label the remaining leads batch by batch and compare with your sample.
What to paste into the tool
Task: Labelled batches with reasons. The project holds the fit criteria and a labelled sample. Label each row in this batch in the same format as the sample, name the criterion in the reason, and mark Borderline when data is missing.
Use: You
Compare sample labels, check each band and record verdicts. If agreement is poor, stop and rewrite instructions before continuing.
Check that the tool reproduces your sample labels before accepting any batch, and spot-check each band.
Use: Claude
Ask Claude for counts per label, rejected categories and open questions from the checked sheet.
What to paste into the tool
Task: Draft summary. Summarise this reviewed sheet: counts per label, rejected categories by criterion and open questions. Use only the sheet.
Use: You
Recount every number, decide borderline leads and sign off.
The human decision is final.
Best when: You want drafting, file analysis and web research together
What it does: ChatGPT drafts; you check sources and decide
Strong at: Writes and summarises text; analyses files and searches the web
Limit: Free plan limits uploads, images, voice and data analysis
Free plan: Yes, limited; uploads and data analysis capped
Best when: You want data work within a spreadsheet
What it does: Gemini drafts; you check sources and decide
Strong at: Analyses spreadsheet data from plain-language requests
Limit: The free plan has usage limits
Free plan: Yes
Best when: You want text drafted and edited from supplied material
What it does: Claude drafts; you check sources and decide
Strong at: Drafts and edits text; summarises documents and analyses images
Limit: Usage capped in 5-hour and weekly windows
Free plan: Yes, limited; usage resets every five hours
Before you paste into any AI tool, remove any confidential information. Put it back at the end.
Deciding between ChatGPT and Claude for this? See the comparison.
Every option keeps the reason next to each lead, uses only the written criteria, and ends with the sales lead deciding the borderline queue.
Check the result against your original inputs. Approve it before sharing or acting.