Cloudmaven · review lane · copy directions

Three copy directions for the homepage
and the solutions surface

This page proposes. It does not ship. Every headline and paragraph below is a draft written from named in-repo source material and carries a visible DRAFT — NEEDS VOICE PASS watermark. Nothing here has been accepted by anybody, nothing is deployed, and nothing is merged.

No invented facts. Every pain point and claim below is traced inline to a file in this repository. Where a source carried an unsourced figure, the figure is gone and the claim stays.
No numbers in proposed copy. No watermarked block contains a digit. The two new drafts contain no number words either; the KEEP block reproduces the live hero verbatim and carries the live site's own “one of the leading”, which is flagged where it appears. The homepage's figure bands are discussed here, never reproduced.
Voice is not settled here. These are directions — argument, order and entry point. The sentences are placeholders for a voice pass, not finished lines.

What the site says today

The homepage leads with what Cloudmaven is and what it sells. Its opening line is “We consult your business with ERP & EPM Software and seamless integrations.” and the paragraph under it describes the firm, not the reader's week. Every named difficulty on the page — manual effort, administration, “processes that work automatically for you” — appears after the offer, as a benefit of buying rather than as the reason to read on. That is the single structural choice the three directions below disagree about.

src/data/baseline/pages/index.json src/data/baseline/pages/digital-solutions.json src/data/baseline/title-changes.json

Note on sources: the brief named .baseline-crawl/pages/, which does not exist on main or on any branch in this repository. The committed crawl-derived page data in src/data/baseline/pages/ was used instead, and it is the file the live-copy gate reads. The AI-Specify material and the two lane direction files also live on other branches; the exact commits read are listed in REPORT.md beside this page.

Direction 1 — KEEP

The zero-risk reference. Current copy exactly as it stands, cleaned only of defects that are already written down in this repository. It exists so the other two are judged against something real rather than against a memory of the live site.

Direction 1

Keep — the live copy, defects removed, nothing else touched

Positioning statement

Cloudmaven is a CFO advisory firm that combines ERP, EPM, CRM, HCM, finance operations and AI-supported automation into scalable, integrable solutions.

Assembled from the homepage's own card row and closing offer trio — src/data/baseline/pages/index.json

Example hero block

Draft — needs voice pass · proposed copy, not approved, not live
DRAFT — NEEDS VOICE PASS
We consult your business with ERP & EPM Software and seamless integrations.

Cloudmaven is entrusted by numerous SME customers and is one of the leading CFO advisory firms in the EMEA region. We support your company on its digital transformation journey. With the Cloudmaven strategy “be global, act local” and our international presence, we guarantee fast, flexible and personal support — regardless of where your company operates.

Let's get started

Verbatim from the live homepage. It is reproduced here because this direction's whole proposal is that it stays; even so, it is a proposal on a review page and carries the watermark. It is also the only draft block on this page containing a number word — “one of the leading” — and that word is the live site's, not this lane's. The other two blocks are free of digits and number words alike.

src/data/baseline/pages/index.json — h1 and hero paragraphs

The pain-point entry it leads with

None. This is the honest answer and it is the direction's defining property. The page opens on capability and credential. The nearest thing to a named difficulty — “less manual effort”, “Less administration”, “processes that work automatically for you” — sits well below the fold and is phrased as an outcome of buying, not as a situation the reader recognises. A visitor who arrived because their close is painful has to read past the offer to find out that Cloudmaven knows about it.

src/data/baseline/pages/index.json — “We integrate your systems”, efficiency band, “Start every day…”

What stays from current copy

All of it, in its current order. The only edits are defects that this repository has already recorded against the live site:

  • The shared page title. The live site puts one title on many pages, the homepage among them; the port already replaces it with the page's own H1. src/data/baseline/title-changes.json
  • The heading-level skip. The homepage goes from its H1 straight to the H3 card row, which reparents everything under it for a screen reader and for every content extractor. The port records this pattern across the site. docs/baseline-port-2026-08-11.md
  • Brand casing. “Cloudmaven”, one capital, everywhere — a build gate fails on the alternative. docs/lab-aispecify (src/pages/lab/aispecify/_README.md)
  • The two figure bands are a decision, not a defect. The homepage carries a “Cloudmaven in Figures” band and an efficiency band. Whether each figure is sourced is a separate question this lane cannot answer, and the standing rule against unsourced figures is explicit in the e-flow source: “No outcome figure appears above — the published case studies carry none, and we do not invent them.” Under this page's own no-numbers rule the bands are not reproduced here. KEEP means keep them; it does not mean this lane has verified them. docs/design/2026-08-12-claude-design/e-flow.html

What to judge — the one/three/thirty-second rule

  • At one secondThe reader can name the category: enterprise finance software and consulting. They cannot yet tell whether it is for them.
  • At three secondsWeakest point. There is no verb and no consequence — nothing has broken, nothing has been repaired. The sentence describes a seller.
  • At thirty secondsStrongest point. The scope is complete and legible: solutions, integrations, service, proof, people, a next step.

The rule is Christoph's, relayed as binding in review/codex-homepage-x1-x3/DIRECTION.md where it governs illustration. Applying it to copy is this lane's own extension of it, and should be read as a proposal too.

Direction 2 — PAIN-FIRST

Process-first framing. The page opens on the reader's week, not on Cloudmaven's portfolio. Every difficulty named is one the source material already names.

Direction 2

Pain-first — the spreadsheet in the middle, the manual gates, the re-keying

Positioning statement

The systems are not the problem — the space between them is, and Cloudmaven works on the handoffs where finance stalls.

docs/design/2026-08-12-claude-design/e-flow.html — “the systems are fine, and the space between them is where the month goes”

Example hero block

Draft — needs voice pass · proposed copy, not approved, not live
DRAFT — NEEDS VOICE PASS
Your close is not slow because your people are.
It jams in the space between your systems.

Invoices arrive by email. Payment approvals happen on a board outside the ledger. Bank reconciliation happens in a spreadsheet. Each of those handoffs is a person re-keying, exporting or emailing a file so that the systems agree — and the file in the middle has quietly become the system of record. Cloudmaven works on the handoffs.

Find your version of it

Source sentence carried a count of the jam points; the count is removed under this page's no-numbers rule and the claim is left standing. Same treatment the AI-Specify port applied to its own unsourced frequency.

e-flow.html — “Your close is not slow because your people are”; “a person re-keying, exporting, or emailing a file so that two systems can agree”; the reconciliation file labelled “The actual system of record” docs/lab-aispecify variant A — “Invoices arrive by email. Payment approvals happen in Monday.com. Bank reconciliation happens in Excel.”

Variant A names the two third-party tools. The draft above generalises them, because a named board vendor on Cloudmaven's own homepage is a positioning decision the lane has no source for. Restoring the names is a legitimate alternative and would be faithful to the source.

The pain-point entry it leads with

The spreadsheet in the middle of the company. One reconciliation file that every system defers to, kept alive by hand, held by somebody who cannot take holiday in the first week of the month. It is the entry because it is the most physical of the sourced pain points and the one a finance operator recognises without being told what it means.

Behind it, in order, the material supplies the rest of the ladder: manual gates — the points where somebody has to touch it before the data can move; re-keying and CSV handoffs — CRM data crossing into the ERP as a file; a recurring report rebuilt by hand — the board pack assembled from exports, where two people produce two versions of the same number and both are defensible; and approval chains living outside the ledger — the invoice approved on a board, then entered into the core system a second time.

e-flow.html — “Manual gate — someone has to touch it”; “a spreadsheet in the middle of the company”; “The board pack is rebuilt from exports”; “Two people produce two versions of the same number, both defensible”; “somebody who could not go on holiday in the first week of the month” docs/lab-aispecify variants A and C — the shared invoice inbox, the approval board, the Excel reconciliation, the CRM CSV export, the recurring report rebuilt in Excel

What stays from current copy

  • “Start every day with business processes that work automatically for you.” It survives intact and moves up — it is the current page's only line that already speaks about the reader's day.
  • “Which systems would you like to connect with each other?” Kept as written. It is a question addressed to the reader, which is exactly this direction's grammar.
  • The named pre-built integrations and the middleware sentence. Specific, checkable, and the concrete answer to the jam the hero just described.
  • The customer testimonials, verbatim. They are the only quoted material on the page and they already speak in pain-and-relief terms.
  • The closing offer trio and the contact block. Unchanged; after a pain-led opening the reader needs the full scope somewhere, and this is where the current page already puts it.

src/data/baseline/pages/index.json

What to judge — the one/three/thirty-second rule

  • At one secondStrongest point. The reader recognises a situation before they know who is speaking. Risk: they may not yet know this is enterprise software at all.
  • At three secondsVerb and consequence are present and explicit — it jams, a person re-keys, the file becomes the record. This is what the rule asks for.
  • At thirty secondsWeakest point. Scope has to be rebuilt after the pain — does the reader still learn that ERP, EPM, CRM, HCM, integrations and ongoing service are all here?
  • Also judgeWhether a CFO reads a named jam as insight or as an accusation about their own house. The material assumes insight; nothing in the sources tests it.

Direction 3 — ADVISORY-DEPTH

Knowledge depth. The claim is not what Cloudmaven sells but what the people answering the phone have already read: the whole surface a CFO's numbers cross.

Direction 3

Advisory-depth — the buyer is talking to people who have read the whole landscape

Positioning statement

Cloudmaven answers on the whole surface a CFO's numbers cross — ERP, EPM, CRM and HCM — rather than on the one system in front of the question.

src/data/baseline/pages/index.json — “Our focus is not on individual tools, but on the overall picture of the financial architecture”; “Cloudmaven combines ERP, EPM, CRM, HCM, finance operations and AI-supported automation”

Example hero block

Draft — needs voice pass · proposed copy, not approved, not live
DRAFT — NEEDS VOICE PASS
The question is rarely which system.
It is who has read the whole landscape before answering.

A finance question almost never stays inside a single system. It crosses the ledger, the plan, the customer contract and the payroll input before it becomes an answer you can defend. Cloudmaven works across ERP, EPM, CRM and HCM, in the order of the process rather than the order of the licence — process first, then tool, then automation, then AI, and never skipping a stage.

Ask us the hard questions first

Depth is claimed only as system surface and method. No promise about who stays on an engagement, no statement about how many people there are, and the word “consultants” does not appear — each of those is a hard rule for this direction.

e-flow.html — the EPM / ERP / CRM / HCM layers; “Process, tool, automation, AI. We do not skip one”; “The damage happens when a stage gets skipped” src/data/baseline/pages/index.json — “the overall picture of the financial architecture”

The pain-point entry it leads with

Answers the system cannot give. Not a broken process but an unanswerable question: the board pack rebuilt from exports every quarter, two defensible versions of the same number, planning happening outside the system in a workbook. The reader's problem here is that nobody they have asked can see far enough across the landscape to settle it — which is precisely the gap this direction claims to fill.

The supporting move is the questions the material says are worth asking any partner in the room: who writes the process map, and when — because if the answer is “during the project”, the price in front of the buyer is an estimate of work nobody has scoped yet; and when did you last recommend a rival system — because every partner calls its recommendation independent and far fewer can date the last time being right about it cost them the licence revenue.

The source lists a third question, about which named person is still on the engagement late in the project. It is deliberately not used: it is a consultant-permanence question, and permanence promises are out of bounds for this direction.

e-flow.html — “Answers the system cannot produce”; “The board pack is rebuilt from exports every quarter”; “Planning happens outside the system, in a workbook”; “Most ERP projects do not fail on the software”; the two questions quoted above

What stays from current copy

  • “Your CFO Advisor for Finance, ERP, EPM & AI-supported automation” and the paragraph under it. This is the current page's own advisory-depth section; the direction is largely an argument for promoting it to the top.
  • “Our focus is not on individual tools, but on the overall picture of the financial architecture.” Kept verbatim. It is the thesis.
  • The card row's system list — ERP, EPM, CRM, HCM, finance operations, AI-supported automation. Kept as the surface claim, moved up to sit under the hero.
  • The named case studies as the depth evidence, with the source's own restraint: the published case studies carry no outcome figures, so none are added.
  • The contact block with a named person. Kept — a page arguing that the reader should talk to people who have read the landscape has to show one.

src/data/baseline/pages/index.json e-flow.html — the five described projects and “we do not invent them”

What to judge — the one/three/thirty-second rule

  • At one secondWeakest point. Nothing physical has happened. The reader meets a claim about judgement, which is slower to land than a jammed close.
  • At three secondsThe verb is “crosses” and the consequence is “an answer you can defend”. Judge whether that is concrete enough to count under the rule, or whether it is still abstract.
  • At thirty secondsStrongest point. Surface, method and the questions to ask any partner are all present, and the reader leaves with something usable even if they do not call.
  • Hard rules to check on every lineNo promise about who stays; no claim about how many people; the word “consultants” absent. Any redraft is failed on any one of the three.

Section by section — what each direction does with the homepage as it stands

The rows are the current homepage in its current order, read from src/data/baseline/pages/index.json. Verdicts are proposals.

Current homepage section Direction 1 — Keep Direction 2 — Pain-first Direction 3 — Advisory-depth
Hero — “We consult your business with ERP & EPM Software and seamless integrations”, plus the SME/EMEA and “be global, act local” paragraphs KeepVerbatim. Title defect fixed, wording untouched. ReplaceThe jam replaces the credential. The EMEA and “be global, act local” material moves down to the scope section rather than disappearing. ReplaceReplaced by the landscape claim. “Be global, act local” survives lower as reach, not as the opening argument.
Card row — Digital solutions · Integrations & Middleware · Service KeepVerbatim, with the heading level corrected so it does not hang off the H1. RewordSame three cards, each opening on the difficulty it removes instead of on what it is. KeepKept and promoted directly under the hero — it is the surface claim the direction rests on.
“Start every day with business processes that work automatically for you” KeepVerbatim, in place. KeepKept word for word and moved up. The one existing line already written from the reader's day. RewordReframed from automation as outcome to automation as the third rung of process → tool → automation → AI.
“Which systems would you like to connect with each other?” — pre-built integrations, middleware, end-to-end landscapes KeepVerbatim, including the named integrations. KeepVerbatim and promoted. It is the concrete answer to the jam described in the hero. RewordSame content, recast as one inspectable boundary between systems rather than as a catalogue of connectors.
“Cloudmaven in Figures” — locations, countries, staff, customers, projects KeepKept as the zero-risk reference. This lane has not verified the figures and does not reproduce them. DropDropped from the homepage. It answers a question the pain-led reader has not asked yet; reach belongs on the about surface. DropMust go — the staff count is a team-size claim, which is a hard rule for this direction. Reach without headcount can be argued elsewhere.
Logo wall — “Companies that already rely on our portfolio” KeepVerbatim. KeepKept, moved below the pain ladder so it reads as “these companies had this problem”. KeepKept as depth evidence, sitting beside the named case studies.
“Your CFO Advisor for Finance, ERP, EPM & AI-supported automation” + the two advisory calls to action KeepVerbatim. RewordKept in place, reworded so it answers the jam rather than introducing the firm a second time. KeepKept verbatim and promoted to the top. It is already this direction's thesis.
Efficiency band — the four statistics and “Cloudmaven increases your efficiency – and your profitability” KeepKept as the reference. Unverified here, and not reproduced under the no-numbers rule. ReplaceReplaced by the named jams and what each one costs in effort, stated without figures. ReplaceReplaced by the questions worth asking any partner — the strongest advisory-depth material in the sources.
Customer testimonials — the four quoted customers KeepVerbatim, including the quoted figures, which belong to the speakers. KeepVerbatim and promoted directly under the pain ladder: the same difficulties, in customers' own words. KeepVerbatim, placed with the case studies as evidence of depth.
Closing offer trio — curated portfolio · “We integrate your systems” · comprehensive service — and the “Get free advice” contact block KeepVerbatim. KeepVerbatim. After a pain-led opening this is where the reader recovers the full scope, and it already does that job. RewordTrio reworded around the process → tool → automation → AI order; the contact block kept as written, named person included.

The solutions surface (/digital-solutions/) inherits its direction from the homepage: KEEP leaves “We implement your new Digital solutions” and the portfolio paragraph untouched; PAIN-FIRST opens each product entry on the difficulty it removes; ADVISORY-DEPTH orders the same entries by where they sit in the process ladder rather than by product family. Source: src/data/baseline/pages/digital-solutions.json.