A persona-level intelligence contact summary is a one-page record for one person that states, in order, the facts recorded about them (name, employer, title, work channel), the persona those facts map to (job function, job level, buying role), the signals modelled onto that persona (importance, fit, engagement) and the next action to take. Every line carries its source, its as-of date and a confidence tier, so a recorded fact is never mistaken for an inference.
Persona-Level Intelligence Contact Summary: The Short Answer
- A persona is a classification, not an observation. In every vendor document read for this guide, a persona is a job function combined with a job level (6sense, Demandbase, Cognism, Clay) or a configured buyer profile (ZoomInfo), produced from a recorded job title. The title is the fact; the persona is a mapping of it.
- Four tiers, four rules. Recorded fields can be verified against a source. Mapped fields can be checked against the title they were derived from. Modelled fields can only be read together with their inputs. Guessed fields are drafting aids. The Four-Tier Persona Ledger below assigns every field in the summary to one tier and states what each tier may be used for.
- The summary is finished when it ends in an action line. A row that lists persona attributes without naming what the seller now does differently is storage, not intelligence.
- Freshness is per field, not per record. Employer, title and channel change on different clocks from function and level, and a modelled importance score changes whenever the model or the CRM behind it changes. The summary records an as-of date on each line rather than one timestamp on the whole record.
What "persona" means in the data that vendors sell
The word is used loosely in sales conversation and precisely in vendor documentation. The table records how five category vendors define a persona in their own published material, how the persona fields are produced, and the date shown on each document. Quotations are verbatim; everything else is a summary of the page.
| Vendor and document | What a persona is made of | How the persona fields are produced, per the document | Date shown on the document |
|---|---|---|---|
| 6sense, Persona Importance Model | "A persona is a function of job function and seniority level." The example given: "Marketing Director would be one persona, while Marketing Manager would be another." | Persona Importance "is calculated using a weighted sum of the three input variables": the average profile-fit score for contacts with that persona, the normalised count of contacts with that persona in the customer's CRM, and the average number of times sales engaged that persona before an opportunity was won. The persona map shows importance as weak, moderate or strong, and admins configure which personas matter for a product. | Published August 14, 2025 |
| 6sense, Contact Profile Fit Model | Not a persona in itself: a score from 0 to 100 indicating similarity to contacts on opened or won opportunities. Listed inputs include lead source, job function, job level, industry and role; "data cleanliness" inputs such as valid email, valid phone and name length; and the normalised frequency of the function-and-level combination. | Calculated from "the demographic and historical CRM opportunity history"; available only with the Predictive add-on. | Updated July 16, 2025 |
| Demandbase, Understanding Buying Groups; Create and Manage Personas; Job Level and Job Function Categorization | A persona is "Title, job function, job level, and custom criteria", built from title keywords, job functions and job levels. A Buying Role is a separate "Pre-defined role in the decision-making process"; the published list includes Decision-Maker, Champion, Influencer, Gatekeeper and Initiator. | Non-English titles are translated into English and mapped to standardised job functions and levels; "Directeur Financier" becomes "Chief Financial Officer", and the translated titles "are not shown to users". A Buying Group Setup Agent generates persona recommendations that admins can regenerate. | Personas article updated March 23, 2026 |
| ZoomInfo, Copilot API reference, List Customer Buyer Personas | "A Buyer Persona represents a profile of a key decision-maker or influencer that a company's sales and marketing teams target." Personas "capture attributes such as the buyer's role, objectives, priorities, and the motivations that influence purchasing decisions." | Personas are configured by the customer and can be connected to products or services. ZoomInfo University's GTM Studio page describes buying-group personas defined by title, department, job function and management level. | API reference updated May 19, 2026; GTM Studio page updated March 9, 2026 |
| Cognism, How to Create and Assign Personas; API Requirements and Entitlements | A persona is a saved set of Job Title, Department, Seniority and Location filters; "Smart Personas" are suggested starting points that can be edited. | Job titles, management level, seniority and job functions are delivered as separate data points on the contact record. | Job title and seniority filter guide dated August 17, 2026 |
| Clay, How to Build a Targeted Prospect List | People are pulled with Find People "by job title, seniority, department, and location" on top of a company list. | Seniority and department are filter values on the people-search record layered on the company list. | Undated; read October 10, 2026 |
Three regularities in the table matter for the summary that follows.
- The persona is produced from the title. Function and level are classifications of a title string by a taxonomy, a translation layer or a model. None of the documents describes function or level as something observed directly.
- The buying role is configured or generated, not recorded. Demandbase separates it from the persona and calls it pre-defined; 6sense's "importance" is a model output; ZoomInfo's persona is a profile the customer writes. A seat on a buying committee is a hypothesis until the person or the company states it.
- The ranking layer describes the seller's past, not the contact. 6sense's published inputs are the customer's own CRM counts and pre-win engagement history. On a plain reading of those inputs, a persona scores as important because people like them were engaged before this seller's previous wins, and a new product line or a new segment with no history gives the model nothing to weigh.
The pages that Google returned for the exact phrase on October 10, 2026 treat the term as four different objects: a CRM activity digest (HighLevel's Contact Summary Agent consolidates "communications, payments, appointments, and tasks" into one summary), a dialling-block grouping (Nooks' Persona Intelligence, announced August 23, 2024, groups prospects "by their job roles, seniority, and more"), an executive research brief (Draup's Persona Analysis Agent, with sections running from "Career & Leadership Profile" to "Engagement Strategy") and a real-estate archetype model (Conversion Realtor). None of the four states which lines in its output were recorded and which were inferred. That distinction is the subject of this page.
The Four-Tier Persona Ledger: Recorded, Mapped, Modelled, Guessed
The ledger is a sort. Every field that can appear in a contact summary belongs to exactly one tier, the tier is written next to the value, and each tier carries a rule about what the value may be used for.
| Tier | What it is | Fields that belong here | How it is produced | Where it breaks | What it may be used for |
|---|---|---|---|---|---|
| Recorded | A value copied from a source that states it | Name, employer, verbatim title, location, work email, direct dial, profile URL, public statements, dated company events | Retrieval from a company page, profile, filing, register or press item, plus a direct check for channels | Stale copy; a namesake; an ambiguous title string; a channel that passed the check and bounced anyway | May be stated as fact in a message once it reaches the Verified or Corroborated rung of the prospect dossier template confidence ladder |
| Mapped | A classification of a Recorded value by a taxonomy | Job function, job level or seniority, department, persona name, headcount band, region bucket | A vendor taxonomy, a translation step, keyword rules or a classifier applied to the title | Level words that mean different things by industry and company size; compound titles; translation loss; one vendor's level scale silently differing from another's | Routing, list building, sequence choice and which message variant to send; stated to the buyer only as the underlying title |
| Modelled | A number or rank computed from Mapped fields plus history or activity | Persona importance, contact profile fit, engagement score, generated buying-role assignment | Weighted sums over the seller's CRM history, engagement counts, agent generation | Encodes who the seller happened to engage before past wins; small samples; CRM hygiene drives the score; the model date is not on the record | Ordering work. Never a reason given to the buyer, and never, on its own, grounds to drop a person from a committee map |
| Guessed | A trait or motive attributed to the person | Communication-style read (DISC-style), stated-nowhere priorities, likely objections, archetype | A person's or a model's reading of what the contact has said and done in public | The research on predicting traits from public footprints supports modest group-level correlations and no per-person claim; see the Said, Did, Guessed sort | Choosing between two drafts and preparing for a call. Never written to the buyer, never used for scoring, qualification or exclusion |
Three rules follow from the sort.
- A field's tier never rises by being copied. A Mapped level written into a CRM "Seniority" field is still Mapped. The ledger travels with the value or the value loses its meaning at the first export.
- Promotion needs a source, not a better model. A Guessed priority becomes Recorded when the person states it in public or on a call. A Mapped buying role becomes Recorded when a procurement notice names the evaluation contact or the person says "I run the evaluation".
- Each tier has a job in the first touch. Recorded facts are what the message may cite. Mapped and Modelled fields decide who is contacted and in what order. Guessed fields decide how a draft is phrased. Crossing those lines is where the congratulations-on-the-role-you-left-in-March message comes from.
What goes into a persona-level intelligence contact summary, field by field
The table is the evidence layer of the summary: for each field, the tier, the source to record, the event that makes it stale, the check that verifies it and the minimum confidence rung before it may drive an action. Confidence rungs are the five from the dossier template: Verified, Corroborated, Reported, Inferred, Unknown.
| Field | Tier | Source to record, with its link | Goes stale when | How to verify | Minimum rung before it drives an action |
|---|---|---|---|---|---|
| Full name | Recorded | Employer page, the person's own profile, a register or filing | A name change, or when a namesake has been merged in | Match two further attributes (employer and role) across two sources | Corroborated |
| Current employer | Recorded | Employer's own site, an announcement, a filing | A move; the field with the shortest shelf life in the record | Employer's own page or announcement, read on a stated date | Verified |
| Title, verbatim | Recorded | Same as employer; copied exactly, including punctuation | Promotion, reorganisation or a retitling | Copy the string as published, with its as-of date; never store only the normalised form | Verified or Corroborated |
| Work email | Recorded | A direct check, logged with its date and result | Domain change, departure, mailbox policy change | The mailbox checks and their limits are the subject of the B2B contact verification methods guide | Recorded as "not refused at check time", never as "deliverable" |
| Direct dial, profile URL | Recorded | Public listing or profile, with the date seen | Departure, number reassignment | Direct check; profile URL opened and matched to employer and title | Verified |
| Job function | Mapped | The taxonomy or rule that produced it, named, plus the title it was applied to | The title changes, or the taxonomy changes | Compare with the department or team the employer's own page lists; re-run the mapping on the current title | Corroborated (title and employer page agree) |
| Job level or seniority | Mapped | Same, plus the vendor's level scale | Same as function | Check a public reporting line or leadership page where one exists | Corroborated |
| Persona name (function and level) | Mapped | Derived from the two rows above | Whenever either changes | Recompute; it inherits the lower rung of its two parts | Inherits |
| Buying role or seat | Mapped when assigned by a title rule; Modelled when generated; Recorded only when stated | The rule, the model or the statement, with date | Every purchase; a seat is per decision, not per person | Tested on the call with the buying committee mapping Yes-No-Live Test | Inferred until confirmed; then Verified |
| Persona importance, profile fit | Modelled | The vendor model, its listed inputs and the date the score was read | The seller's CRM history or the model changes | Read the inputs; ask the vendor for the sample behind the persona | Reported; sets order only |
| Engagement with your company | Recorded per event; Modelled as a roll-up | Your own systems, with timestamps | Continuously | Event log | Verified per event; the roll-up is Reported |
| Account context (industry, headcount band, dated public events) | Recorded per event; Mapped for bands | Company page, filings, press, postings, each with its URL and date | Per event | Link each event to its source | Verified or Corroborated |
| Public statements (what the person has said) | Recorded | Talk, post, article or interview, with URL and date | Does not age, but can lose relevance | Quote verbatim | Verified |
| Communication-style read | Guessed | The artifacts it was drawn from, listed | New artifacts appear | Cannot be verified; can be contradicted on the call | Inferred; drafting only |
| Action line | Output | Cites only Recorded rows | Any cited row changes | Reviewed before send | Every cited row Verified or Corroborated |
When a vendor's contact record is being compared on these fields rather than on raw counts, the workable-contact-cost method in the best B2B contact database guide is the pricing lens; this page supplies the field list to compare against.
How Mapped fields are produced, and the three places they break
A Mapped field is a taxonomy's opinion about a title. Demandbase's documentation is the clearest description of the step: titles in other languages are translated to English, the translation is mapped to a standardised function and level, and the translated titles "are not shown to users". 6sense defines the persona as the function-and-level pair and gives the example that a Marketing Director and a Marketing Manager are two personas. Cognism and Clay expose seniority and department as filter values on the record. In each case the summary should store three things next to the Mapped value: the verbatim title it came from, the name of the taxonomy or vendor that produced it, and the date it was mapped. Without the first, the mapping cannot be audited; without the second, two vendors' "Director" buckets get merged as if they were one scale; without the third, nobody knows whether the level reflects this year's title or the one before.
The step breaks in three recognisable places.
- Level words that change meaning by context. "Vice President" at a large bank and "Vice President" at a forty-person software company sit at different distances from the budget, and "Head of" is used for everything from a team lead to a board-level officer. A level taxonomy applies one rule to all of them. The summary's defence is the employer's own leadership page or a public reporting line, recorded as a source.
- Compound and non-standard titles. "Director of Revenue and People", "Chief of Staff" and "GM, Platform" each straddle two functions or hide the function entirely. A keyword rule picks one; a classifier picks the likelier one. Either way the record should show the verbatim string so a reader can see what was decided.
- Scale mismatch across vendors. One vendor's level set may distinguish "C-Level", "VP", "Director", "Manager"; another may fold owners and partners into the top bucket or split "Senior Director" out. When records from two providers are merged, the Mapped level should carry its provider name or be re-mapped under one rule. The B2B data enrichment guide covers the write-back discipline that keeps a CRM from accumulating three incompatible seniority scales.
Modelled fields inherit all three breakages and add their own: the inputs are the seller's own past behaviour. 6sense's three published inputs are all drawn from the customer's CRM, and publishing them is what makes the limitation readable. A summary that shows "importance: strong" without the model's inputs and date is asking the rep to trust a number whose meaning changed the last time someone cleaned the CRM.
From a stored row to a sales action: the Action Line
A contact summary earns the word intelligence at its last line. The format used here is one sentence with five slots:
Action: [verb] [person] on [channel] about [a Recorded, dated event or statement]; ask [the question that tests the seat]; hold if [the named field] is older than [n] days.
The slots enforce the tier rules without a checklist. The event or statement must be Recorded, so the message cites a fact. The question tests the Mapped or Modelled seat rather than assuming it. The hold clause names the field with the shortest shelf life, which for a contact record is the employer line, and pins a re-check to it.
The table below shows how each kind of persona field changes the first touch. The middle column is framed as a hypothesis to test, not a finding, because every one of these fields is Mapped, Modelled or Guessed.
| Persona field on the summary | Hypothesis to test | What changes in the first touch |
|---|---|---|
| Mapped level is executive, under the taxonomy named | This person can approve spend in this category | Open with the outcome and the dated public event; ask who runs the evaluation rather than assuming they do |
| Mapped function is the one that operates what you sell | This person will live with the result | Open with the workflow change; ask what they measure today |
| Mapped function is security, legal or procurement | This person can say no but not yes | Send after the operator and the sponsor, with the documents those seats will ask for |
| Modelled importance is "strong" | People in this seat were engaged before this seller's past wins | Earlier in the sequence; nothing changes in the message itself |
| Guessed style reads as data-first | Evidence will land better than narrative | Choose the draft that opens with the dated event; the read itself is never mentioned |
| Recorded public statement on a relevant topic | The person holds a stated priority | Reference the statement by name and date |
Two outcomes are deliberately absent from the right-hand column: dropping a person because a model scored them low, and personalising on a trait. The first turns a Modelled field into a decision about a human being on the seller's own historical data; the second writes a Guess to the buyer. Both are the failure modes the ledger exists to prevent.
A worked example (invented)
The company, the person and every value below are invented to show the format; nothing here describes a real account.
Halvern Logistics Software (invented), 320 employees, mid-market freight-visibility SaaS. The seller offers a sales-intelligence platform. The summary is for one contact.
| Line | Value | Tier | Source type | As-of | Rung |
|---|---|---|---|---|---|
| Name | Dana Okafor | Recorded | Employer leadership page; own profile | 2026-10-08 | Corroborated |
| Employer | Halvern Logistics Software | Recorded | Employer leadership page | 2026-10-08 | Verified |
| Title, verbatim | Director, Revenue Operations | Recorded | Employer leadership page | 2026-10-08 | Verified |
| Work email | d.okafor@[domain] | Recorded | Direct check, not refused | 2026-10-08 | Recorded, not refused at check time |
| Job function (vendor A taxonomy) | Operations | Mapped | Vendor A record, mapped from the title above | 2026-09-30 | Reported |
| Job function (corrected) | Sales / Revenue Operations | Mapped | Employer page lists the role under "Sales" | 2026-10-08 | Corroborated |
| Job level (vendor A scale) | Director | Mapped | Vendor A record | 2026-09-30 | Corroborated with leadership page |
| Persona | Revenue Operations, Director | Mapped | Derived | 2026-10-08 | Corroborated |
| Buying seat | Operator, candidate champion | Mapped by rule | Title rule; untested | 2026-10-08 | Inferred |
| Persona importance (vendor A) | Strong | Modelled | Vendor model; inputs are the seller's CRM history | 2026-09-30 | Reported |
| Account event | Posted two SDR roles and one RevOps analyst role | Recorded | Employer careers page, URLs stored | 2026-10-06 | Verified |
| Public statement | Conference talk on routing inbound leads by territory | Recorded | Event programme and recording, URL stored | 2026-10-01 | Verified |
| Communication-style read | Conscientious-leaning | Guessed | Drawn from the talk and two posts | 2026-10-08 | Inferred |
Action: email Dana Okafor on the work address about the three roles posted on October 6 and the territory-routing talk on October 1; ask whether the new SDR seats will source their own accounts or receive routed ones, which tests the operator seat; hold if the employer line is older than 30 days at send time.
Two things in the example are the point. Vendor A's taxonomy mapped the title to "Operations", which would have routed Dana into an operations sequence about fleet uptime; the employer's own page put the role under Sales and corrected the function. And the "strong" importance score did not change the message at all; it moved Dana up the week's order. The communication-style read chose the data-first draft and appears nowhere in it.
Persona mapping, profiling and what the GDPR text asks
Mapping a title to a function and level, scoring a person's fit and inferring a communication style are automated processing of personal data to evaluate aspects of a person. The GDPR's definition is wide: profiling "means any form of automated processing of personal data consisting of the use of personal data to evaluate certain personal aspects relating to a natural person, in particular to analyse or predict aspects concerning that natural person's performance at work, economic situation, health, personal preferences, interests, reliability, behaviour, location or movements" (Article 4(4)). Two provisions attach to it in the outreach context. Article 14 requires that a person whose data were not collected from them is told, among other things, the source, and where the data are used for communication, "at the latest at the time of the first communication". Article 21(2) gives the person the right to object at any time to processing for direct marketing, "which includes profiling to the extent that it is related to such direct marketing", and once they object, "the personal data shall no longer be processed for such purposes". For the summary, that means the source column is not only a quality control; it is the material the Article 14 notice is built from, and an objection has to stop the Modelled and Guessed processing as well as the sending.
Where Lead Seeker fits
Lead Seeker's per-person record is the Prospect Dossier, which its product page describes as "a verified contact, the public signal that surfaced the person, a short personality read, and a suggested first line tuned to the buyer". Against the ledger, its six published fields sort as follows: retrieved contact data (work email, LinkedIn URL and direct dial when public, "with the freshness date stamped on the record"), company context and source-backed research are Recorded, with "every claim in the dossier" linking "to the public source where available"; the DISC-style communication guidance is Guessed, and the product page labels it "a directional read, intended to help a rep choose tone and structure for the first message, not a clinical assessment"; the outreach recommendation is "an editable opening line tuned to the DISC read and the surfacing signal", the dossier's equivalent of the action line. Upstream of the dossier, Lead Compass "uses your company website, ICP, and intent themes to surface opportunity picks, cite public sources where available, and generate ready-to-run search prompts", and does not consume Lead Units until a user runs a search. The how Lead Seeker works page shows the six stages from persona definition to booked meeting. For the account-level question that precedes any contact summary, which accounts are in market this quarter, Percepture's guide to using intent data to identify sales qualified leads covers the signal side.
Frequently Asked Questions
What is a persona-level intelligence contact summary?
It is a one-page record for one person that states the recorded facts about them (name, employer, verbatim title, work channel), the persona those facts map to (job function, job level and, where assigned, buying role), the signals modelled onto that persona (importance, fit, engagement) and one action line, with a source, an as-of date and a confidence tier on every line. The persona-level intelligence contact summary differs from a plain contact record in the tier column: it says which lines were recorded, which were mapped from a title, which were modelled and which were guessed.
How is a persona different from a job title?
The title is a string published by the employer or the person; the persona is a classification of that string. 6sense defines a persona as "a function of job function and seniority level"; Demandbase builds one from "Title, job function, job level, and custom criteria"; Cognism saves one as a set of job title, department, seniority and location filters. Two people with different titles can share a persona, and one title can be mapped to different personas by different vendors' taxonomies, which is why the summary stores the verbatim title next to every Mapped field.
How is job level or seniority assigned in B2B contact data?
By applying a taxonomy, a set of keyword rules or a classifier to the title. Demandbase's reference describes translating non-English titles into English and mapping them to standardised functions and levels, with the translated titles not shown to users; Cognism lists management level, seniority and job function as separate data points on the record; Clay's prospect-list guide filters people by job title, seniority, department and location. In every case the level is Mapped, so it should carry the name of the taxonomy that produced it and be checked against a public leadership page where one exists.
What is the difference between a contact summary and a prospect dossier?
A dossier is the full research record for a person and account; the summary is the one page a rep reads before the first touch. The prospect dossier template on this site holds the complete field set with sources and shelf lives; the contact summary selects from it the recorded facts the message may cite, the persona fields that set the order and the variant, the guessed read that picks the draft, and the action line. A summary without a dossier behind it has no source column to point to.
When should a persona-level contact summary be refreshed?
Per field, on the event that makes the field stale, rather than on a calendar for the whole record. The employer line has the shortest shelf life and is the one the action line's hold clause names; title and channel follow it; function and level only change when the title or the taxonomy does; a modelled importance score changes whenever the model or the CRM behind it changes, so it should be re-read, with its date, each time it is used to order work. The B2B contact data decay guide on this site covers the field-level clocks in detail.
Is mapping a contact to a persona "profiling" under the GDPR?
The definition in Article 4(4) covers "any form of automated processing of personal data" used "to evaluate certain personal aspects relating to a natural person", including aspects of their performance at work, interests and behaviour, which is broad enough to include automated title mapping, fit scoring and style inference. Article 21(2) lets the person object to direct-marketing processing "which includes profiling to the extent that it is related to such direct marketing", and Article 14 requires the source to be disclosed no later than the first communication where the data were not collected from the person. How those provisions apply to a particular programme is a question for the team's counsel; the summary's source column is the material that answer is built from.
Sources
- 6sense, Persona Importance Model, published August 14, 2025: persona definition, the three model inputs, the weighted-sum calculation and the weak, moderate, strong scale. Read October 10, 2026.
- 6sense, Contact Profile Fit Model, updated July 16, 2025: the listed inputs, the 0 to 100 score and the Predictive add-on prerequisite. Read October 10, 2026.
- Demandbase, Understanding Buying Groups, Create and Manage Personas for Buying Groups (updated March 23, 2026), Buying Roles for Buying Groups and Job Level and Job Function Categorization. Read October 10, 2026.
- ZoomInfo, List Customer Buyer Personas, Copilot API reference updated May 19, 2026, and ZoomInfo University, Setting Up Buying Groups in GTM Studio, updated March 9, 2026. Read October 10, 2026.
- Cognism, How to Create and Assign Personas, How to Use Job Title and Seniority Filters (August 17, 2026) and API Requirements and Entitlements. Read October 10, 2026.
- Clay, How to Build a Targeted Prospect List. Read October 10, 2026.
- HighLevel, How to Generate a Contact's Summary Using Ask AI, modified July 1, 2026; Nooks, Focus on Great Conversations with New Persona Intelligence, August 23, 2024; Draup, Persona Analysis AI Agent; Conversion Realtor, Persona-Level Intelligence: Contact Analysis for Real Estate. The pages returned by Google for the exact phrase on October 10, 2026; summarised, not audited. Read October 10, 2026.
- GDPR, Article 4, Definitions, Article 14, Information to be provided where personal data have not been obtained from the data subject and Article 21, Right to object. Read October 10, 2026.
- Lead Seeker product pages: Prospect Dossier, Lead Compass, How it works and Try free, quoted as published on October 10, 2026.
About the Author
Bob Generale is President of Percepture. He works across SEO, AI search, digital PR, sales intelligence and AI-powered revenue systems, with a focus on connecting visibility, buyer intent and sales action.
Next Steps
Take ten records from this week's sequence and write the tier next to every field: Recorded, Mapped, Modelled or Guessed. Any message that cites a Mapped or Guessed line gets rewritten before it goes out, and any persona-level intelligence contact summary that cannot end in an action line with a Recorded event in it is a research task, not a lead. To see the account side of the same discipline, read how Lead Compass turns market signals into prospecting direction, see how Lead Seeker works end-to-end, then claim 5 free verified leads and sort their lines with the same ledger; the source and as-of date are already on each record.