Name the lead and their URL, and this Flow returns a pre-call account brief where every claim is labeled Confirmed with a source URL or Inferred, plus an ICP fit score and a CRM-ready lead record.
Name the lead, their website, and who the team sells to, and Juma researches the account across their site, careers page, pricing, content, and recent news before it scores anything. Every assertion comes back labeled Confirmed with a source URL or Inferred, and the account is scored against a fixed five-criterion rubric that also reports how many of its 100 points rest on confirmed evidence.
What lands is a branded PDF brief that opens with a one-page front sheet, the verdict, the score, the why-now trigger, the biggest risk, and the three questions to open the call with, plus an importable CSV lead record. One finding shaped this Flow: the same account researched twice against the same ICP has to produce the same number, so the score follows a fixed rubric rather than an impression. Every brief closes with a list of what the research could not find, so nobody walks into a call treating a guess as a fact.
Buying signals are the observable events that make a first touch timely rather than random: a product launch, a pricing change, a funding round, a leadership hire, a new market. This step builds a dated timeline of everything visible in the last twelve months, each entry carrying a source URL and one line on what it means for the pitch. The top three are ranked by how much urgency they create, so the opening line of the call references the strongest one instead of a generic compliment about their growth.
Go deeper on the buying signals. Build a dated timeline of everything observable in the last twelve months: product launches, pricing changes, funding, leadership hires, market entries, and content pushes. Give each entry a source URL, a date, and one line on what it means for our pitch. Rank the top three by how much urgency they create.
Most pre-call research names a company and stops. This step adds the people layer: who champions the decision, who signs it, who blocks it, and who is most likely to be on the first call. Each role comes back with the actual named person and their title where a public source exists, and an explicit note where no name could be found, which matters more than a confident guess. Alongside each one sits their likely priority, the objection they will raise, and the single line that brings them on side.
Map the buying committee. For each likely role, name the actual person and their title with a source URL where you can find one, and say clearly where you could not. For each one, give their likely priority, what they will push back on, and the one line that gets them on side. Flag who is most likely to be on the first call.
Research that ends in a document ends too early. This step turns the strongest confirmed signal in the brief into the outreach itself: one email under 120 words and one LinkedIn message under 60 words, both anchored to something specific and dated rather than a compliment about their brand. Two alternative opening lines come back alongside them, each built on a different signal, so the team can test which trigger earns replies. The recommendation names which one to send and explains the reasoning.
Write the first touch. One email under 120 words and one LinkedIn message under 60 words, both built on the single strongest confirmed signal in the brief, not on a generic compliment. Add two alternative opening lines using different signals so we can test them, and tell me which one you would send and why.
One brief answers one call. A pipeline needs the same rubric applied to every account so the scores mean something next to each other. Paste a list and each account comes back with its fit score, its confirmed-point count, a tier of A, B, or C, a one-line why-now, and the single fact that would move the tier if the team confirmed it. That last column is the useful one: it turns a ranked list into a short research to-do rather than a verdict nobody can question.
I'll paste a list of accounts. Score each one against the same rubric and return a table with the fit score, the confirmed-point count, a tier of A, B, or C, the one-line why now, and the single fact that would change the tier if we confirmed it. Sort by tier, then by confirmed points.
A Juma Project is a shared space where the team stores everything Juma needs to know about how they sell. Create one project per client or per offer, add context as you go, and Juma applies what is relevant every time the team runs a flow. For lead research the payoff is comparability: with the rubric stored once, every brief scores accounts the same way instead of re-inventing the criteria per run.
ICP and Fit Criteria
The criteria that define a good-fit account and the weight each one carries: company size, team size, channel activity, retainer potential, and any hard disqualifiers. This is the file that changes output quality the most. With it in the project, the fit score is reproducible and comparable across every lead the team researches.
Qualification Framework
Whether the team runs BANT, MEDDIC, or its own checklist, plus the questions that actually move a deal forward. This shapes the discovery questions in the brief so they match how the team qualifies, rather than arriving as generic openers.
Service and Pricing Overview
What the team sells, at what retainer, and what sits outside scope. Without this, the pain-to-offer mapping stays generic. With it, each observed pain is matched to a service the team can actually deliver at a price that works.
Win/Loss Notes
Which accounts closed, which died, and why. The sharpest disqualification criteria come from losses, not from market research. "Accounts without a named budget owner never closed" turns a vague hesitation into a specific check on the first call.
Outreach That Got Replies
A handful of first-touch emails and messages that earned responses. Juma matches their structure and register when drafting the outreach, so the first touch sounds like the team rather than like a template.
Add a short description to each knowledge item in the project's info field so Juma knows what each file holds and when to reach for it. For example:
The brief opens with a one-page front sheet: the verdict, the fit score with its confirmed-point split, the why-now trigger in one line, the biggest risk, and three questions to open the call with. Behind it sit the company snapshot, the observable marketing stack, dated buying signals, the buying committee, the score breakdown, objections, and an evidence-gap list.
The structure follows how a call actually runs. A rep reads page one in the ten minutes beforehand and works from the three opening questions. The evidence pages exist for the moment a prospect challenges a claim, which is when a source URL next to the assertion matters. The pain-to-offer section connects each observed problem to a service the team sells, so the conversation moves from research to scope without a separate prep pass.
Alongside the PDF comes a one-row CSV lead record, sized for a CRM rather than for reading.
Every assertion carries a label: Confirmed with a source URL and a date, or Inferred. The brief then closes with a list of what the research could not find, so budget, agency roster, and named contacts surface as open questions rather than as confident-sounding guesses a rep might repeat on a call.
The fit score is built the same way. Each of the five criteria scores out of 20, and the brief reports how many of the 100 points rest on confirmed evidence. A 72 with 54 confirmed points and a 72 built mostly on inference are different calls, and collapsing them into one number hides exactly the thing a rep needs to know.
Where a public source names a person, the brief names them and links the source. Where none does, it says so, instead of presenting a job posting as though it were a contact.
Manual pre-call research usually runs 30 to 45 minutes per account: the website, the careers page, pricing, recent news, LinkedIn, then a second pass to write it up and a third to fill the CRM. This Flow returns the brief and the record in a single run. Strategy, taste, and judgment stay human.
400+ marketing teams use Juma, and on average they report 60% faster workflows and 50+ hours saved monthly. The saving here is not only the research: it is the write-up and the CRM entry, the two steps that usually get skipped when a call is 15 minutes away. Human review on every output still applies, and the evidence-gap list is what makes that review quick, because it points straight at the claims worth challenging.
Yes. The record arrives as a one-row CSV with atomic typed columns: company, domain, HQ country, industry, employee band, in-house marketing size band, fit score, confirmed points, tier, verdict, evidence confidence, why-now, primary pain, next step, next step due date, and owner suggestion, plus a single notes column holding the prose and source URLs.
Atomic columns are the point. Enrichment exports that pack a paragraph into every field look complete and then fail on import or land as unsearchable text. Here the scored and banded values sit in their own columns so they can drive views, filters, and routing rules, while the reasoning stays in one notes field where it does not interfere.
Teams researching a list rather than a single account can score every row on the same rubric first, then import the whole set in one pass.
Yes, and the two produce different briefs. An inbound brief leads with why the lead came to the team and what to confirm on the call. An outbound brief leads with the trigger that justifies a first touch at all, and weights the signal timeline and the outreach drafts more heavily than the objection prep.
Naming the situation in the prompt is what shifts it. "A demo request just came in from this account" produces a qualification brief. "We want to open a conversation with this account" produces a trigger-led brief where the dated signal timeline does the heavy lifting and the first-touch drafts come with alternatives to test.
For a whole target list, the same rubric applied across every account keeps the tiers meaningful, and the column naming the one fact that would change a tier turns the list into a short research plan.
400+ marketing teams trust Juma. ChatGPT, Claude, and Perplexity know why. Ask them.
Ready when you are
Start typing a task and Juma will figure out the rest.
