Qualify leads with AI, then check the low ones
Write your question as a column header. Columns answers every row and tells you which answers to check.
Agencies: send us a list, we run up to 1,000 rows free. Results in 2 business days.
The list you have been putting off
You exported 5,000 companies. Half of them are wrong for you, and nothing in the spreadsheet says which half. So someone opens a tab per row, reads a homepage, types "yes", "no" or "maybe", and moves on to the next one. That someone is you, or a VA you are paying by the hour.
It works. It is also slow, it is dull, and the 400th row gets less attention than the first. Most of the cost is not the hard rows. It is reading the easy ones.
How Columns qualifies a list
- 01Upload your list. CSV, with a company name and a website is enough.
- 02Type a question as a column header. For example: "Is this a B2B SaaS company with under 50 employees?"
- 03Every row gets answered. You get a typed answer for each company: yes or no, a pick from your own options, or a score.
- 04Every answer gets a confidence score. Sort by it.
- 05Check the low ones. Read the rows the sheet is unsure about. Leave the high ones alone, or spot check a few.
That last step is the point. Most tools hand you 5,000 answers and ask you to trust them. Columns gives you 5,000 answers and a short list of the ones not to trust. A confidence score is a signal, not a guarantee, so a few high-confidence answers will still be wrong. Spot check them until you are comfortable with how your questions behave.
This is AI lead qualification built for a list you already have, not for chat on your website. If that is what you need, read the overview on the homepage or how confidence scores work.
Qualification questions agencies actually use
Paste these in as column headers and adjust the wording to your offer. Narrow yes/no questions work best.
- Does this company sell to SMBs?
- Is this a B2B SaaS with under 50 employees?
- Do they run paid ads?
- Is this a local service business (plumber, dentist, roofer, etc.)?
- Are they hiring salespeople right now?
- Does this company sell to enterprises?
- Do they work with agencies or list an agency partner program?
- Does the site look like it has an outbound or lead-gen offer of its own?
Several of these exist as ready-made sheets: B2B SaaS checker, agency outbound checker, hiring salespeople checker, local service business checker and sells-to-enterprises checker.
Columns judges each row from the columns in your sheet, such as the company name, website and any description you add. It does not browse the web for you, so the more your sheet says about each company, the better the answer. With only a name and a domain, expect lower confidence on questions about what a company is doing right now, like hiring or running ads.
Tips that save rework: ask one thing per column, make it answerable from what is in your sheet, and say what counts. "Under 50 employees" beats "small". If you want the long version, see how to ask yes or no questions the AI can answer and where to set the confidence threshold.
By hand, with a VA, or with Columns
| Item | By hand | VA | Columns |
|---|---|---|---|
| Cost | Your time | Their hourly rate, plus your time writing instructions and fixing mistakes | Priced per row-question, from free |
| Speed | As fast as you read | As fast as they read | Rows fill in as a batch |
| Consistency | Drifts as you get tired | Depends on the VA and the brief | Same question applied to every row |
| Knowing what to recheck | You remember the shaky ones, maybe | Only if they flag them | Confidence score on every answer |
| Judgment calls | Best | Good | Weakest. This is why you check the low ones |
A VA is still the right call for work that needs a login, a phone call or taste. Columns is for the repetitive question you ask of every row.
And Clay? Clay is a broader data and workflow platform. Its Launch plan, its cheapest paid plan, shows $167/mo on its pricing page, and its FAQ says starting at $185/mo. If you need its data providers and multi-step workflows, it does things Columns does not. If your job is "answer this question for every company on my list and tell me what to verify", Columns does that one job and starts at $0. See Columns vs Clay pricing.
Pricing
- Free: 500 row-questions, no card.
- Starter: $29/mo
- Pro: $99/mo
- Agency: $299/mo
A row-question is one question answered for one row. Runs stop at your limit rather than billing past it, and you are warned at 80%. See every plan.
FAQ
- Can AI really qualify leads from a list?
- It can answer a defined question about each company, such as whether it sells to SMBs, from the columns in your sheet. It will get some wrong. That is why each answer comes with a confidence score and why you check the low ones.
- Are the answers always right?
- No. Treat them as a first pass. Check the low-confidence rows, and spot check some high-confidence ones until you trust your question wording.
- How is this different from asking ChatGPT to go through my CSV?
- Columns runs the same question across every row, gives each answer a confidence score, and keeps it in a sheet you can sort and filter. For a longer take, read can ChatGPT qualify leads from a CSV.
- What do I need to start?
- A CSV with company names and websites, and one question. You can try 50 rows with no signup or start free with 500 row-questions, no card.
- What does the agency trial include?
- Send us a list and we run up to 1,000 rows free. Results come back in 2 business days. See the agencies page.
- What does it cost after the free start?
- Starter is $29/mo and Pro is $99/mo. Agency is $299/mo. Details are on the pricing page.
Bring the list you have been putting off
Write the question. Let it fill. Check the low ones.
Start free, no card