AI Lead Research: How It Works at Scale

Account 41 gets a worse write-up than account 3. Not because it deserves one, but because by then you have pasted the same prompt forty times and you are tired of reading careers pages. That quality curve, sliding quietly down a list nobody else will ever audit, is what AI lead research actually looks like in most companies: a spreadsheet in one tab, a chat window in the other, and a founder doing both at 11pm.

The thing is, it works. Vendors skip past this and they shouldn't. For fifty accounts, a chat window plus some patience produces research a rep can genuinely use, at no cost beyond a subscription you already pay for.

What breaks is not the quality of any single answer. It's the list. Research done one account at a time never turns into a view of all of them, so you finish with four hundred paragraphs and no answer to the only question that matters on Monday morning: which twelve of these do I work first, and why those twelve. That gap is why people and company search that arrives enriched and scored is the shape we built rather than another research box.

ChatGPT for sales prospecting: where AI lead research breaks

ChatGPT for sales prospecting is the entry point almost everyone uses, and the ways it comes apart are consistent enough to list:

• Nothing accumulates. The answer lives in a conversation rather than on the account record, so the next person to touch that account starts from zero.

• Nothing re-runs. An account that raised a round in March is still described by the paragraph you wrote in January.

• Answers are not comparable. Forty companies produce forty differently shaped paragraphs, because the prompt drifted and so did your attention.

• Sourcing is uneven. Some claims come from a page the model just read. Others come from training data, which is where hallucination in AI systems stops being a research-paper concern and becomes a wrong headcount sitting in your CRM.

Fifty is the number we keep coming back to because it's roughly where a person stops being able to hold the list in their head. Below it, you remember that the fintech one felt promising. Above it, you're looking things up twice and not noticing.

Research agents: the current shape of AI sales research

Every serious tool in the category now ships a research agent, and they are good. You write a prompt once, point it at a domain or a profile, and it runs per account: a few searches, a few page reads, then a structured answer with the sources attached. Better ones retry on a timeout, cache what they already fetched, and say they found nothing instead of filling the gap with a plausible sentence.

Treat that as table stakes. AI sales research at this stage is a genuinely faster version of the chat window, and if you are still doing this by hand, a research agent is the upgrade to make first.

It is also where the category has stopped. The unit of work is one account. Run the agent over four hundred of them and you have four hundred good, well-sourced, structurally identical write-ups, which is a real improvement and still not a ranking.

A small complaint about the genre, since we are here. Nearly every AI research demo picks one beautiful account, usually a Series B company with a tidy website and a blog, and runs the agent live. Account 173 never gets demoed. Account 173 is a holding company with a one-page site and a Gmail address on the contact form, and what a research pass does with account 173 is the only interesting question about it.

AI lead research across every account: score, rank, segment, explain

Ask a different kind of question and the shape of the problem changes. Not "what does this company do," which any agent answers well, but "of these four hundred, which forty match the customers we keep, and which five should someone call today."

A per-account agent can't answer that, and it isn't a model limitation. When the agent runs on account 200, it has never seen accounts 1 through 199. It has no distribution to place that account in. So it hands back an 80, and an 80 means nothing without the other 399 scores next to it.

Reasoning across the list is one pass over everything you know, and it produces four things: a score against your ICP, a rank against the other accounts, a segment or tier the account belongs in, and a written reason. The reason is the part reps care about. "84, paid plan since April, seats climbing, asked about SSO" survives an argument. A bare 84 does not, and a rep who doesn't trust the number goes back to calling whoever signed up last.

The scoring half of this has its own mechanics, from point-based models to what AI changes about them, and we cover those in what lead scoring is and how the models work. The research half is the input side: the pass can only weigh facts that made it onto the record.

What feeds AI account research: providers and your own systems

Which means AI account research is mostly an input problem. The quality of the answer is decided before any model runs, by what you put in front of it, and inputs come in three kinds.

Provider data covers facts about strangers: headcount, industry, funding, a verified work email. No single provider knows every company, so these get chained into a waterfall that tries providers in order until one answers. This part is a lookup, not research, and it's the cheapest evidence you will get.

Web research covers what no database sells. Whether they shipped an AI feature last quarter. Whether the careers page lists three sales roles. What the pricing page says now, rather than in the crawl from eleven months ago. This is where an agent earns its keep, and where sources matter most.

Then there are your own systems, which is the half almost nobody researches. An account that already pays you $740 a month, filed two tickets about API rate limits, and has nine of fifteen seats active is the strongest research subject on your list, and no provider sells any of those facts. They sit in your billing system, your support desk, and your product analytics. A Stripe subscription record knows the plan and whether the last charge failed. Your support desk holds the complaint, in the customer's own words.

Oneprofile treats all 127 integrations as sources and destinations, so billing, support desks, product analytics and a read replica of your own Postgres feed the research pass the same way a data provider does, and the result goes back to the CRM as fields your reps can filter on.

What AI lead research costs per contact

Here is where AI lead research gets compared badly. Per-token AI billing on a meter separate from data billing means nobody can predict the cost of a research run until it has finished, which is a strange thing to accept for a workload you want to run on your whole list every month.

Our prices, so you can do the arithmetic yourself. A credit is $0.02, on every plan and on top-ups alike.

• A built-in research agent: 5 credits an account, so $0.10.

• A person enrichment from an email: 0.8 credits, so $0.016.

• Finding a work email: 2.5 credits. Verifying one: 0.3 credits.

Research plus enrichment on 400 accounts is roughly 2,300 credits, or about $46, and the estimate appears before the run rather than on the invoice. Lookups that find nothing cost nothing. The free plan renews 200 credits a month, which is enough to research a real slice of a list before you decide anything, and Team is $20 for 1,000 credits.

Whether frontier model prices keep falling at the rate they have, we genuinely don't know, and anyone who tells you they've modelled their 2027 AI research budget is guessing. What we'd rather not do is pass that uncertainty to you as a variable bill.

Anyway. If you want to test the premise without touching a tool, take twenty accounts you lost last quarter and write down the five facts that, in hindsight, predicted the loss. Count how many of those five your CRM can currently see. That count is the real ceiling on any AI lead research you run, whichever product does the running.

Then, if you want the version that doesn't stop at one account: Oneprofile researches and enriches every account on your list from data providers and from your own billing, support and product systems, scores each one against your ICP with a written reason, and syncs the whole thing back to your CRM on one credit meter that tells you the price first. Get started free.