Sales console — what a company card opens into

Proposed — pending Fede's pick ·14 August 2026 ·Four shapes, one to pick

Click a company on the sales board and a panel opens. This page is about what should be in that panel. Three teams designed it three different ways, each was then torn apart by a reviewer whose only job was to break it, and what survived is below. A fourth shape was then built from Fede's reaction to the first three — it is Option D, and it is the recommendation.

If you want to see the raw material first: everything we know per company — every field the research holds, how many companies have it, and where it lands on the card.

The decision

The board holds 646 companies. Fourteen of them are live conversations. The other 632 are names we have researched to some depth and nobody has contacted yet.

Today every one of those 646 opens the same panel: company description, four facts, a route in, where the buildings are, what happens next, a prep prompt, research notes. It is a research file. It answers "what is this company" well and "what do I do about it" barely.

The first three options each pick a different reader and build the panel for them: the person about to dial, the person working down a long list, and the person chasing a live deal. They are not three coats of paint on one idea — they open on different things and cut different things.

Option D was built afterwards, from Fede's feedback on A and C. It keeps the parts he said he liked, drops the part he didn't, and adds two things nobody had built: a section on how the company actually runs its buildings, and a "don't pitch this" block that changes with the type of customer and cites where each line came from. It is also the only one designed to be read during a call rather than before one. Pick one.

What we checked before designing anything

Every number on this page was measured against the sales repo, not estimated. The reviewers found fabricated numbers in all three original drafts; these are the ones that held up.

646companies on the board — 14 in a live conversation, 632 not yet contacted

288written opening lines sitting in 96 research briefs that nothing on screen shows

567swept companies that each have a real written paragraph — currently hidden behind a fold

8 of 14live deals whose "how we got here" list generates completely empty today

The four options

Option A

The Call Sheet

The panel's job is to get someone ready to say the first two sentences. It opens on the three opening lines a researcher already wrote for this exact company, then the person to dial, then the things we must not claim out loud. Facts move below as backup.

Sections, in order

Who this is — name, kind of company, where the deal stands
Where to focus — the written openers, headline shown, the proof behind a click
Who to call — the named person and why them
Don't claim this — the brief's own list of what to verify first
The four facts — size, software, where, who runs the buildings
Their buildings — folded
How we got here — folded when nothing has happened
Take it with you — prep prompt, link, full brief

What it looks like

Rendered in full below, on KAGE / Urban Phenix — jump to the mockup.

What it cuts vs today

The made-up "route in" sentence, which the app generates from the stage and a name
The four facts stop being the headline and become the backup
Research prose is folded, not a section

What it costs to build

Read and render 96 markdown briefs — the writing already exists
A hand-built list of 18 name aliases, because brief filenames and company files disagree
A build check that fails when any brief points at nothing
No new research, no new data collection

The flaw we could not fix

It is only genuinely good for the 96 companies that have a brief. The other 550 open a call sheet with no call written on it. And the briefs were generated once, on 14 August, by nothing that runs again — so they will quietly go stale with no signal on screen.

Option B

The Triage Strip

The panel is not a page you read, it is a station you step through. Built for working eighty companies in one sitting: a next button inside the panel, the one thing the list didn't already tell you at the top, and everything else collapsed.

Sections, in order

Step bar — previous / next through the list you came from, and a copy link
Who this is — name, kind, stage, how deep the research goes
The new sentence — the researcher's own paragraph, open, at the top
What the list didn't show — only the facts the table didn't already print
Who to call — or an honest "nobody named yet"
Still unknown — the short list of gaps
Everything else — buildings, prep prompt, sources: all folded

What it looks like

Rendered in full below, on the same company — jump to the mockup.

What it cuts vs today

Everything the list already showed — no re-printing the row you just clicked
The made-up "route in" sentence, deleted outright
The buildings section on companies that have no building list

What it costs to build

Panel needs to know the order of the list it opened from — real plumbing, but small
Fix the size line: it currently claims we counted by hand on 270 of 308 swept companies where we didn't
Carry web links through so a source can be clicked

The flaw we could not fix

Stepping is only useful if you can tell where you got to — and nothing anywhere records that a company was looked at. Not the database, not the browser. Someone who steps through sixty companies has no trace of it and no way to resume. That is a data change, not a design change.

Option C

The Case File

For a company we are actually talking to, the panel should open on the deal, not the company: what is owed, what is late, what moved and who moved it. Everything we know about the business drops below that.

Sections, in order

Who this is — name, kind, stage
Where this stands — how long here, last word, what is owed
What we do next — the written next step, and the promises both sides made
How we got here — dated emails and stage moves, from the database
The company in five facts
Where the units are
Before the call — the openers and the prep prompt
Research notes — folded

What it looks like

Rendered in full below, on the same company — jump to the mockup.

What it cuts vs today

Company facts stop being the opening — the deal is
The current "how we got here", which is built by scraping dates out of prose
Promises stop living on a separate tab and land on the company

What it costs to build

Join four database tables onto the panel — emails, stage moves, promises, edits
Retire the prose date-scraper. It is the most expensive line item and it is a deletion
Every cell must come from a row that exists — no cell the database can't fill

The flaw we could not fix

It builds the panel for 14 companies out of 646. The other 98% open a case file with no case in it. And even inside those 14, only eight have any email on record — so half the case blocks are thin on the very deals it exists to serve.

Option D · Recommended

The Hybrid

Built from Fede's feedback on A and C. The facts and the people sit at the top so you always know who you are looking at; then the live deal, then the lines worth saying, then how the company actually runs its buildings, then what we must not pitch to a customer of this type. It is designed to be glanced at mid-call, not read beforehand.

Sections, in order

Who this is — name, kind of company, tier, stage
The facts — size, software, head office, who runs the buildings. At the top, not below the fold
Players — the people on their side, decision-maker first, name and role only
What we do next — from Option C, and moved above the deal block on live deals
Where this stands — from Option C: how long it has sat, who spoke last, why it paused, what each side owes. Replaced by Before you call on a company nobody has contacted
Worth mentioning next time — from Option A: lines you could say out loud, evidence one click away
Don't pitch this — segment by segment, every line citing its source. New
How they operate — structure, who runs what, site teams, staffing ratios, from the operations brief. New
Timeline — what was asked, what came back, what it left us with
Their buildings, research notes, still unknown — folded

What it looks like

Rendered below on three real companies at three stages, behind one toggle — jump to the mockup.

What it cuts vs today

The made-up "route in" sentence, deleted outright
"Don't claim this" is replaced by "Don't pitch this" — the old block was a list of facts to verify; the new one is guidance on what not to sell to this kind of buyer, and it names its source on every line
Nothing empty is drawn. A company with no history shows "nothing yet", not a blank section

What it costs to build

Everything A and C cost, since it uses both: the markdown brief reader plus the database joins
Read the operations briefs too — 96 of them exist and nothing renders them
Somebody has to write the "don't pitch" lines per segment. Owner-operators and portfolio owners have real sourced guidance today; fee managers have almost none, and the panel says so out loud rather than inventing it

The flaw we could not fix

It is the most expensive of the four, because it is the sum of two of them. And the segment-tailored guidance is only as good as the writing behind it: our own research says we have never sold to a third-party fee manager, so that segment's block is honest but nearly empty. Fede's audience-breakout spreadsheet is meant to be the source for these lines, and it could not be read while this was built — Google access had expired — so every line currently comes from the sales playbook, the segment research and the company's own files.

The same panel, drawn four ways

A, B and C are all the real record for KAGE / Urban Phenix — the Denver operator we met on 6 August. Same company, same facts, same colours and type as the live console; only the shape changes. Option D is drawn three times, on three different real companies, because the whole point of it is that the panel changes as a deal moves: a company nobody has called, a company one message has gone out to, and a company mid-conversation with promises outstanding. Every line on every panel is lifted from that company's own files; where we do not have something, the panel says so instead of filling it in. All four live in one mockup page now, not six.

Option A — The Call Sheet

Open full-size ↗

What is different here: it opens on three written lines you could say out loud, each with the evidence one click away; the person to call comes second; and a “don’t claim this” block names the four things we cannot prove yet. The facts strip is pushed below all of that. The frame below scrolls.

Option B — The Triage Strip

Open full-size ↗

What is different here: a previous / next bar inside the panel; the researcher’s own paragraph is the first thing you read; and the facts are rewritten as “what the list didn’t show” — only what the row you clicked did not already print. Everything else, including the history, is folded shut. The frame below scrolls.

Option C — The Case File

Open full-size ↗

What is different here: it opens on the deal, not the company — how long it has sat, who spoke last, why it paused; a promises block saying what each side owes; and the dated history sits third, above the facts rather than below them. The frame below scrolls.

Option D — The Hybrid, at three stages of a deal

Open full-size ↗

Use the toggle at the top of the frame to switch between not contacted / first conversation / in progress. One panel, three real companies, because the whole point of D is that it changes as a deal moves. The frame below scrolls.

The frame also carries a Full / Lean / Minimal toggle, in the same spot, on all three stages. All three are the same lines, display-only — nothing is a shorter research pass, only a harder cut of what is on screen. Full is everything; Lean cuts the coaching and the padding; Minimal cuts further, to a mid-call glance: one header line, up to three players, the single next action, one line on where the deal stands, two "say" lines, one "don't say" line, and the three most recent timeline rows, with everything else one click away behind a single "Everything else" fold. Minimal is now the recommended density for production — Fede's verdict on Lean was "still way too much information," and Minimal is the response: the whole card readable at a glance, about 800px tall, no scrolling.

Not contacted — Griffis Residential. Tier 1, Greenwood Village, never contacted, no contact name on the list at all. What D borrows: the lines you could say out loud with evidence one click away from Option A, and the written next step from Option C. What is new: the facts and the Players sit at the very top; a “how they operate” section built from their operations brief; and “don’t pitch this”, which replaces A’s “don’t claim this” and is tailored to the type of customer, every line naming its source. With no deal to report there is no “where this stands”, and the timeline says plainly that nothing has happened.

First conversation — Botnick Realty. One message sent on 10 August, nothing back yet. What changes from the cold state: “where this stands” appears, but it is three lines, and it says openly that neither side has promised anything; the timeline has exactly one row, not a padded list. Note how the “don’t pitch this” block differs from the other two companies even though all three are owner-operators — this one carries a warning about their federally subsidised senior building, because their own file carries it.

In progress — KAGE / Urban Phenix. The same company A, B and C are drawn on, so you can compare like for like. What changes again: “what we do next” moves to the top, above the deal block, because there are live action items; “where this stands” is full — how long it has sat, who spoke last, why it paused, and what each side owes; and the timeline runs four substantive rows saying what was asked, what came back and what it left us with.

Straight to one stage, full-size: not contacted, Griffis Residential ↗ · first conversation, Botnick Realty ↗ · in progress, KAGE / Urban Phenix ↗

Everything we know per company (the research inventory)

Every field our research actually holds about a company, what it means in plain words, how many companies have it filled, and where it lands on Option D today. Nothing here is estimated: each count comes from reading the files. Use it to decide what the panel should show that it does not.

The five stacks, and how many companies are in each

There is no single number. The research sits in five layers of decreasing depth, so every count below carries its own denominator.

Layer	Companies	What it is
The board	646	Every company on the ranked list. Name, score, tier and why.
The sweep	567	Machine-enriched: a written paragraph, a segment, usually a software guess and a unit estimate.
Operations briefs	96	A written brief on how the company staffs and runs its buildings.
Deep profiles	96	The full structured research file, plus a hand-written profile page beside it. 79 Tier 1 + 17 Tier 2, merged 14 August.
Actual conversations	7	Companies with an email thread on file. Three of them also have a call transcript.

Counted 15 August 2026: 646 rows in the latest ranking, 567 in the enrichment sweep, 96 operations briefs, 96 deep-profile files (79 Tier 1 + 17 Tier 2, each with a matching written profile), 20 email threads across 7 companies, and 6 call transcripts across 3 of them.

The shape of the gap

Depth falls off a cliff. 96 companies have a full research file and 646 are on the board, so for six companies in seven the panel can only ever show the sweep's paragraph, a segment and a guess at the software. And only 7 companies have any email on file at all — the timeline section is empty for 639 of them, which is why Option D draws "nothing yet" rather than a blank.

Field	What it means	Filled	Where it lands on Option D
Identity & facts
`company`	The company name	96/96 · 543/567	The title of the card
`resolvedCompany`	The name after we worked out which company the list actually meant	517/567	The title, when it differs from what the list said
`companyNameNote`	Why the name is written this way — trading names, parent companies	95/96	Research notes
`identity.status`, `identity.note`	How sure we are this is the right company at all	96/96	Research notes and the footer line
`identity.candidates[]`	The other companies with the same name, and why they were ruled out	15/96	Not shown yet
`note` (sweep)	The written paragraph describing the company	567/567	The description under the title
“The short version” (profile page)	The hand-written summary at the top of the profile	76/80 files	The description under the title
`segment`	What kind of buyer: owns and runs, owns and hires out, or manages for a fee	96/96 · 567/567	The type pill beside the title — and it picks which “don’t pitch this” list is highlighted
`sean_tier`, `tier`, `score`, `tierReason`	How high a priority, and why	646/646 ranked · 96/580 hand-tiered	The tier pill. The reason behind it is not shown yet
`unitCount`, `unitsEstimate`	How many apartments they have	88/96 · 308/567	Units, first fact
`unitCountNote`, `unitsNote`, `unitsBasis`	Where the number came from, and which competing figures exist	96/96 · 539/567	Units, the small print under it
`unitCountConfidence`, `unitsCountedFloor`, `unitsClaimed`	Counted by us, claimed by them, or a floor we can prove	88/96 · 85 counted · 178 claimed	The ✓ counted tick beside the number
`hqCityState`, `hq`	Where the head office is	93/96 · 514/567	Head office, third fact
`founded`	What year they started	84/96	Not shown yet
`aum`	How much money they manage	57/96	Not shown yet
`website`	Their site	480/567	Not shown yet
`multifamily`	Whether they are actually in apartments	96/96	Not shown yet — used to filter the board
People
`people[].name`, `.title`, `.note`	Who works there, what they do, why they matter	96/96	Players
`people[].verified`, `.sourceUrl`	Whether the company published that title or a third party did	96/96 · 95/96	The caveat line in Players and the card footer
`decisionMaker.personId`	Which of them actually signs	88/96	The “Aim here” badge in Players
`decisionMaker.canDecideAlone`	Whether that person can decide without a committee	96/96	Players, in the reasoning under the name
`inputContact` (sweep)	The one contact name the original list carried	171/567 · 193/580 in the sheet	Players — for most of the board it is the only person we have
`peopleRuledOut[]`	People we checked and decided were the wrong door	1/96	Not shown yet
`warmPath`, `warmPathHops`	Who we know who could introduce us, and how many hops away	96/96 · 56/96	Feeds What we do next. The hop count is not shown yet
How they operate
`pmModel`	Whether they run their own buildings or hire someone	96/96	Who runs them, fourth fact
`managedBy.name`, `.confidence`, `.note`	The outside manager, when there is one	71/96 · 77/567	Who runs them, and the small print under it
`employsOnSiteStaff`	Whether there are people at the building	96/96	Not shown as its own line — folded into the prose
`alsoFeeManages`, `backsOperatorsOnly`	Whether they also manage for other owners, or only put up money	96/96	Not shown yet — both change who the buyer is
“Company Snapshot” (ops brief)	What the company is and how it is put together	96/96 briefs	How they operate — Structure and Span
“Staffing Model” (ops brief)	Which roles exist at head office and at the buildings, and which do not	91/96 briefs	How they operate — Site teams, Not staffed, Headcount
“Key Ratios” (ops brief)	Apartments per employee, per manager, per maintenance tech, against the industry norm	94/96 briefs	The folded staffing ratios table inside How they operate
“Why This Matters for Clara” (ops brief)	The brief's own read on whether they are a good fit and why	83/96 briefs	Not shown yet
`corporateStructureNotes`	Holding companies, funds, who owns whom	2/96	Not shown yet
Software
`pmsVendor`, `properties[].pms`	Which property-management software they run	374/567 · 78/84 with a property list	Software, second fact
`pmsConfidence`	How firmly that was established	551/567	Software, the small print under it
`pmsEvidenceUrl`, `properties[].evidenceUrl`	The page we traced it on	375/567 · 83/84	Not shown as a link — the panel describes the evidence in words
`incumbentVendor.name`, `.contractStatus`, `.note`	Whoever already sells them something like us, and whether they are locked in	38/96	Not shown yet — arguably the single most useful missing field
Buildings
`markets`	The cities and states they operate in	96/96	Where the units are
`properties[].name`, `.city`, `.state`, `.units`	Building by building, where and how big	84/96 have a list · 80/84 with unit counts	The folded building table and the spread bar
`properties[].class`, `.affordability`	How upmarket, and whether rent is subsidised	84/84 of those with a list	Don't pitch this, where subsidised housing changes the pitch
`properties[].portfolioStatus`, `.propertyRegion`, `.note`	Still owned or sold, which region, and anything odd about it	73/84 · 84/84 · 84/84	The folded building table's note column
`propertyCount`, `propertiesNote`	A stated number of buildings, and caveats on the list	1/96 · 5/96	Not shown as a field — the panel says it in prose
Deal state
`pipeline.stage`, `stage`	How far along we are, from never contacted to closed	96/96 · every row on the board	The stage chip beside the title — and it decides which shape D draws
`pipeline.nextStep`	The written next move	83/96	What we do next
`pipeline.basis`	What that next move is based on	96/96	The “Basis:” line under it
`promises` (database)	Who owes whom what, by when, and whether it was kept	7 companies with email to detect from	Where this stands — the open-promises block
`verbal_target` (database)	What they said out loud they wanted	an editable field on every company	Not shown yet
`note` (database)	A human's free-text note on the company	an editable field on every company	Not shown yet
`rejection.reason`, `.byWhom`, `.date`	Why a company was passed on, and by whom	1/96 · 35/567 carry a status	Not shown yet
`questionForSean`	The one thing we need Sean to answer before moving	26/96	Not shown yet
`terms`, `gates`, `timing`, `deliverNow`	The ranking's read on what it would take and whether we could deliver today	646/646	Not shown yet
Timeline
`emails` (database)	Every message in and out, with the date and who sent it	20 threads across 7 companies	Timeline
Call transcripts	What was said on the phone	6 transcripts across 3 companies	Timeline, as the substance of a row
`stage_moves` (database)	When the deal moved, who moved it and why	recorded from now on, per company	Timeline
`researchedAt`, `recheckedAt`	When we last looked	96/96 · 1/96	The card footer
`prospect_edits` (database)	Who changed which field, and what it was before	recorded from now on, per company	Not shown yet
Pitch guidance
“Where to Focus: Top 3 Positioning Plays” (ops brief)	Three written lines you could say out loud about this exact company	96/96 briefs	Before you call / Worth mentioning next time — the numbered lines with their evidence folded under
`verbalTarget`	The angle the research thinks will land	96/96	Feeds those lines; not shown as its own field
“Caveats” (ops brief)	What the brief says not to assume about this company	96/96 briefs	Don't pitch this, and Still unknown
Sales playbook and segment research	What not to sell to each type of buyer, written once, not per company	3 segments — fee managers nearly empty	Don't pitch this, source named on every line
“Public Footprint” (ops brief)	What they say about themselves publicly	8/96 briefs	Not shown yet
Provenance & unknowns
`sources[].url`, `.what`	Every page the research was built from	96/96	Research notes, and the source line under each caution
`sourcesUnreachable[]`	Pages we tried and could not read	1/96	Not shown yet — but the panel says it in prose when it matters
`openQuestions`	What we still do not know about them	94/96	Still unknown, folded at the bottom
`dataGaps`	The ranking's own list of what is missing for this company	642/646	Not shown yet — the only gap list that exists for the other 567
`confidence` (sweep)	How much to trust the swept record as a whole	564/567	Not shown yet
`risks`	What could go wrong with this one	1/96	Not shown yet
`distanceToDenverMiles`, `deliveryLoadEstimate`	Two fields the schema defines and nothing fills	0/96 and 1/96	Dead — delete rather than design for

Counts are of files on disk on 15 August 2026. “96” is the deep-profile set (79 Tier 1 + 17 Tier 2) and equally the operations briefs, “567” the enrichment sweep and “646” the ranked board — a field can only be filled for companies in its own layer. Database fields have no per-company count here because they are filled as the work happens rather than by a research pass.

Recommendation

Recommended Option D — The Hybrid

Plainly: it is built from Fede's own feedback on A and C. He liked the deal block and the next step from the Case File, wanted the history rewritten as a timeline, liked the lines to say before a call from the Call Sheet, and did not like "don't claim this" — so that block is now "don't pitch this", tailored to the type of customer and sourced from our own positioning material rather than invented. Two things were added on top: the people and the facts sit at the very top of the card, and a "how they operate" section reads the operations brief nothing on screen currently shows.

The design test it is built against is a mid-call glance. Sean is on the phone; the pitch lines, the basic facts, the people and the things not to say all have to be findable without reading. That is why the dense material is at the top, every line has a bolded lead-in, and all the prose is folded. The cost is honest: D is the sum of A and C, so it is the most expensive of the four to build, and the fee-manager guidance behind "don't pitch this" is nearly empty because we have never sold to one.

Fixes that ship no matter which option wins

The reviewers found two things that are wrong on screen right now. Neither is a design question.

What's wrong	How often	The fix
The size line says "counted by hand, building by building" on companies where the number came from a public filing. The tick beside it means "we counted these ourselves."	270 of 308 swept companies that have a size	The web link is stripped out before the sentence is built, so the sentence falls through to a canned line. Keep the source in words, or say nothing.
"The way in" is a sentence the app writes from the stage and a name. On institutional owners it advises a direct call the scoring rules forbid, and it renders two people as one name.	Every one of the 646	Delete it. Fede already ruled the same pattern out one function away in the same file: a summary appears only when a person wrote one.

What the adversarial panel taught us

Each proposal was handed to a reviewer told to break it. All three broke. The failures rhymed.

Every mockup had invented numbers in it

All three. One put a company's headquarters in the wrong state, another invented a unit count for a company that has five conflicting ones on file, a third named the wrong person as the decision maker two lines under a sentence naming the right one. Design mockups are where fabrication hides, because nobody checks a picture.

"The field exists" is not "the field is useful"

One design gave a permanent row to "can this person sign alone." It is filled in on a third of companies — but only 11 of 96 say yes. The row would read "nobody has asked" or "no" on nine cards in ten. Coverage percentages were quoted as if they were usefulness percentages, three separate times.

The join between the briefs and the companies is broken

Two proposals assumed a research brief and a company file with matching names are the same company. Measured: 78 of 96 briefs match, 18 don't — including CONAM, our furthest-along deal. Causes are typos in filenames and inconsistent name endings. Fuzzy matching is the dangerous fix: attach the wrong brief and someone reads another company's pitch as fact.

The "how we got here" list is mostly empty or wrong

It is built by scraping dates out of research prose. Run against the 14 live deals it produces nothing at all for 8 of them — including our own first customer — and for our furthest-along deal it prints a future plan as though it already happened, cut off mid-sentence. No design should promote it until it runs on real dated records.

The thin company is the normal case, not the edge case

Two designs were shown off on the richest company in the set. 632 of 646 companies have never been contacted and 550 have no brief. A design judged on its best case is a design that will be met with an empty screen nine times out of ten.

The best sentence we own was already hidden

Every one of the 567 swept companies carries a real written paragraph — a proper description, median 518 characters. Today it is trimmed and buried behind a fold labelled "research notes." On a company we know nothing else about, that paragraph is the record.

Your pick

Which shape should the company panel take?

One choice. The two fixes in the table above happen either way.

D — The Hybrid Facts and people on top, then the deal, the lines to say, and what not to pitch to this kind of buyer. Built from your feedback on A and C. Recommended. A — The Call Sheet Opens on the three written opening lines. Best for making the call. B — The Triage Strip Opens on the one new sentence, with a next button. Best for working a long list. C — The Case File Opens on the deal: what's owed, what's late. Best for chasing live conversations. A now, C later Build the call sheet first; add the deal block once more companies are in play. D, but in two goes Ship the facts, people, next step and "don't pitch this" first; add the deal block and the timeline once the database work lands.

In plain terms

Sean has 646 companies to work through and has only started talking to 14 of them. When he clicks one, what should he see first? Option A shows him what to say — three lines somebody already wrote about that exact company, plus the things we can't prove yet. Option B shows him the one fact the list didn't already tell him, and a button to move to the next company. Option C shows him what's owed and overdue on a deal already in motion. A helps him start conversations, B helps him get through the list faster, C helps him close the ones he's in. Option D does all three from one card and adds the two things none of them had: how the company actually runs its buildings, and a short list of things not to say to a buyer of that type — with a note under each line saying where it came from. It is also the only one built to be glanced at while he is on the phone. That is why D is the recommendation.

PropFlow Docs