convey.school
0 of 0 boxes ticked
for Hannah · go in order, do the boxes

Brand Onboarding.

This is the very first thing we do with a new brand. We are not building anything yet. Onboarding is where we gather every metric, collect every asset, and build deep context, so that when we build the site, the flows, and the ads later, we build the right thing. Two people run in parallel: Dougie owns brand and creative, Dale owns data and measurement. Use the tabs above to see each lane, or the two side by side.

What onboarding actually is

Onboarding is the intelligence phase. The whole point is to understand the brand better than the founder does before we touch a single thing they own. We read their customers, their reviews, their numbers, their inbox, and the recording of our first call, and we turn all of it into one context pack the whole team builds from.

If you remember one thing: in onboarding we gather and understand. We do not build. The website, the Klaviyo flows, the ads, the CRO fixes, those are separate courses that come after this.

In scope here

  • Pull every metric (Shopify, GA4, Meta, Klaviyo)
  • Read the customers and every review
  • Read the current positioning and find the gap
  • Find the top 3 buying cohorts
  • Collect every brand asset into one Drive
  • Decide brand voice, aesthetic, and design direction
  • Audit what is tracked and where the money is

NOT here (later courses)

  • Building or redesigning the website
  • Building the Klaviyo flows
  • Launching or changing ads
  • Implementing CRO fixes
  • Publishing anything live

The one place we do more than observe: we decide the brand direction. What is the voice? What aesthetic fits? Is there room to improve their branding? What does the founder actually say about their own brand? That decision is a deliverable of onboarding. Everything else on this page is gather and understand.

The two shared inputs, both lanes start here

Before either lane splits off, both Dougie and Dale sit inside the same two sources of truth about this brand:

Input 1 · The call

The Fathom recording of the onboarding call. This is the founder in their own words: what they think the brand is, their goals, their pain, and anything they have said about their design and branding. Read the full transcript, do not skim it.

Input 2 · The inbox

Every email and text exchange with the brand up to now. Scrape the inbox for the whole thread history. This is context they have already handed us: prior asks, constraints, what they care about, hard nos.

Runs onGmail connectorYou + Claude

Which agent does what, and when

The most important thing to learn: a person does not do this by hand. You direct agents, and each agent owns a specific job. Learn who to call at each point.

AgentLaneWhat it owns in onboarding
DelphiDougieThe 5-stream intel sweep: voice of customer, reviews, the top 3 cohorts, per-cohort search demand
agent-reachDougieThe scraping skill: Reddit, TikTok/IG comments, YouTube, reviews, who tagged them
SloaneDougiePositioning teardown, competitive read, brand voice, aesthetic room-for-improvement
DesignerDougieUI/UX fit: does the current design match the brand, where is the friction
roastDougieStress-tests the new positioning before we commit to it
ShoppyDaleShopify data pull: orders, AOV, catalog, top products, repeat rate
MargoDaleAttribution and growth-math read: true CAC, brand-search waste, channel profit
Gmail connectorBothInbox and text scrape for context

Why every prompt on here is long

The prompts in each step are written to pull the most detailed output possible. A lazy prompt gets a lazy answer. Copy them as-is, fill the [brackets], and you get depth every time. When you write your own, use the same six moves:

MoveWhat it does
Give it a role"You are my brand strategist" sets the standard it answers at
Number the deliverables(1)(2)(3) forces it to cover every part instead of one
Demand verbatim + sourcesQuotes and permalinks, not paraphrase, so it is provable
Set a minimum count"at least 20 quotes", "top 15" stops it going shallow
Ask for VERIFIED vs ESTIMATESeparates fact from guess, no fabricated numbers
Say "be exhaustive, do not summarize"Turns a summary into the full picture
→ if an answer comes back thin, add "go deeper, you missed detail" and it expands.
Rule of the course: do the "Do this now" boxes. Reading alone teaches 20 percent. Doing teaches the rest.
Every brand is different

A brand with no ad spend means no Meta pull. A brand with no email list means Klaviyo is greenfield, note the gap and move on. A brand with 4,000 reviews is a goldmine, a brand with 12 leans on competitor and category chatter instead. Read the brand in front of you and take what applies.

Now pick a lane above. Start with Dougie · Brand & Creative to learn the brand side, then Dale · Data & Measurement for the numbers side, then Side by Side to see how they run together.

Step 0 · Load the context

AgentsGmail connectorYou + Claude

Before any research, get inside the brand's own words. Read the Fathom recording of the onboarding call and scrape the inbox for every email and text with the brand. You are looking for how the founder describes themselves, their goals, their pain, and anything they said about their branding and design.

Prompt · context synthesisYou are my brand strategist. Read this Fathom transcript [paste transcript or link] and every email and text thread with [brand], oldest to newest. Be exhaustive and quote directly, do not summarize loosely. Deliver as labeled sections: (1) how the founder describes the brand in their exact words, 5+ verbatim quotes, (2) every stated goal, ranked by how often or how emphatically they raised it, (3) every pain point and frustration, each with the quote and the timestamp or email date, (4) everything they said about their branding, design, or what they wish looked better, verbatim, (5) every hard no, constraint, budget, or deadline, (6) open questions they asked that we still owe answers to. Flag anything ambiguous instead of guessing.
Do this now

Open the Fathom recording for tonight's brand. Paste the transcript into the prompt above and run it. Keep the quotes, you will reuse them in Step 5.

Step 1 · Voice of customer

AgentsDelphiagent-reach

Go where customers talk, not where the brand talks. Reddit, TikTok and IG comments, Facebook groups, YouTube comments, Amazon Q&A. Pull the real language: what they love, what they complain about, what problem they were solving when they bought.

Prompt · Delphi, stream 1ADelphi, run stream 1A for [brand] in the [category] space, and go deep. Search Reddit, TikTok and IG comments, YouTube comments, Amazon Q&A, and relevant forums for people talking about [product type]. Return at least 20 real customer quotes with permalinks, sorted by score. Then synthesize into the 3 governing feelings driving the purchase. For each feeling: 5+ verbatim quotes, the trigger moment that makes someone buy, the exact words they use, and how often it appears. Tag each finding VERIFIED with a link or ESTIMATE. Never paraphrase where a direct quote exists.
Prompt · agent-reach scrapeResearch everything customers say about [brand] and [top competitor] across Reddit, TikTok comments, YouTube, and reviews. Be exhaustive and verbatim, do not summarize. Group findings into: what they love, what they complain about, what made them switch brands, and the exact words and phrases they repeat. Give me 15+ direct quotes with a source link for each, and count how often each theme appears so I can see what matters most.
Every brand is different

A niche or pre-launch brand may have thin chatter about itself. Lean on the competitor and the category instead, the buying language is the same.

Step 2 · Read every review

AgentsDelphiagent-reach

Shopify reviews, Google, Trustpilot, Amazon, Judge.me, and influencer comment sections. This is the goldmine, the copy writes itself from here. Sort into three buckets: what wins them, what almost lost them, and the exact words they use.

Prompt · review miningPull every review you can find for [brand] from [Shopify / Judge.me / Amazon / Google / Trustpilot], plus the comment sections of their top posts. Report the total count and star distribution per source. Then cluster into: (1) the top 8 repeated praises, each with a representative verbatim quote and a rough frequency, (2) the top 8 objections, complaints, or 1-to-3-star themes, each with a quote, (3) the exact phrases customers use to describe the product, listed as a copy bank, (4) any recurring use case, occasion, or gifting mention. Rank everything by frequency and keep it verbatim.
This output is the Voice of Customer report. It is the source doc every deck pulls from. Build it first, everything else references it.

Step 3 · Current positioning, then the upgrade

AgentsSloaneroast

What are they saying about themselves right now, on the hero, the bio, the ads? Where is the gap between how they see themselves and how customers actually describe them? The upgrade is almost always say back to them what the reviews already prove, sharper and ownable.

Prompt · Sloane teardownSloane, do a full positioning teardown of [brand]. Read their homepage hero, IG bio, top 3 ads, PDP copy, and About page. Deliver: (1) exactly what they claim to be today, quoted, (2) how customers actually describe them, pulled from the VOC report, (3) the specific gap between the two, (4) three candidate positioning angles that say back what the reviews already prove, each with a one-line hero, why it is ownable, and the proof behind it, (5) your single recommended angle and why. Be specific, no generic marketing adjectives.
Prompt · roast stress-testroast this new positioning for [brand]: [paste the new angle]. Attack it from three sides and give the strongest objection from each: a skeptical customer, the strongest competitor, and the founder who has to stand behind it. Then a verdict of green-light, reshape, or kill, with the reason, and the single cheapest 48-hour test to prove or kill it before we build on it.

Step 4 · Find the top 3 cohorts

AgentDelphi

Not demographics. Buying cohorts, built from four inputs crossed together: the brand's sales data (from Dale's lane), who is actually reviewing and tagging, the product's real use case, and where competitors are weak. Rank by revenue potential times ease of reach.

Prompt · Delphi cohortsDelphi, using the VOC report, the review clusters, [brand]'s sales data from Dale's pull, and the competitor teardown, give me the top 3 buying cohorts, richly profiled. For each cohort deliver: a memorable name, the governing feeling, who they are, their job-to-be-done, the buying trigger, the exact objections to overcome, the messaging angle that lands, the channels to reach them, an estimate of size and search or trend demand with a source, and one sample creative hook. Then rank the three by revenue potential times ease of reach and explain the ranking. Tag each number VERIFIED or ESTIMATE.

Step 5 · Brand voice and aesthetic

AgentSloane

This is one of the few things we actually decide in onboarding. Define the brand's voice and aesthetic, and call out where there is room to improve their branding. Weigh what the founder said about their own branding, from the Fathom call and emails, against the market reality.

Prompt · Sloane voice + aestheticSloane, build [brand]'s voice and aesthetic guide. Inputs: the founder's own words from the call and emails, the VOC language, and their current visual identity. Deliver: (1) the brand voice as 3 rules, each with a do, a don't, and a one-line example rewrite, (2) 10 on-brand words and 10 words to avoid, (3) an honest read of the current aesthetic, what works and what looks off, (4) specific, itemized room-for-improvement on their branding, (5) where the founder's stated vision aligns with or clashes against the market reality, called out plainly. Be candid, this is the decision doc.

Step 6 · UI/UX and design fit

AgentDesigner

What design fits their branding? Is the current site design any good? Where is the room for improvement? This is an assessment, not a rebuild. We are deciding the direction, the actual redesign is a later course.

Prompt · Designer auditDesigner, audit [brand]'s current site design against their brand voice and aesthetic from Sloane. Assessment, not a rebuild. Deliver: (1) a screen-by-screen read of home, PDP, and cart, (2) does the UI match the positioning, scored with reasons, (3) the top UX friction points ranked by likely impact on conversion, (4) the design direction that would fit, specified as type, color, spacing, and layout feel, referencing the design tokens, (5) a keep list and a fix list. Be concrete enough that a builder could act on it later.

Step 7 · Asset sweep into one Drive

Agentsagent-reachYou + Claude

Collect everything and file it in one consolidated master brand Drive. Nothing scattered. This is the raw material every future build pulls from.

Lane output

Dougie's lane produces the brand context pack: VOC report, positioning and the new angle, top 3 cohorts, voice and aesthetic decision, design assessment, and the full asset library. It all lands in the Brand OS as the source for the build courses.

Check yourself: what is the one thing this lane decides, versus just observes?
Voice, aesthetic, and design direction. Everything else in the lane is gather and understand. We do not redesign or rewrite anything yet, we decide the direction the later build will follow.
Dale's lane is about getting the data and understanding what already exists. He is not building flows or the site here. He pulls every metric and audits the current measurement and funnel, so we know the real starting line before anyone builds.

Step 0 · Access and credentials intake

AgentYou + Dale

Everything downstream depends on access. Request read access to each system. How much we can pull decides whether this is a full-data audit or a public teardown.

Every brand is different

No access to a system means note it and run the public version of that audit. A brand with no ad accounts skips Meta and Google Ads entirely.

Step 1 · Shopify data pull

AgentShoppy

The commercial baseline. Read-only. Pull the numbers that tell us how the business actually performs today.

Prompt · Shoppy baselineShoppy, pull [brand]'s full Shopify baseline, read-only, laid out clearly. For last 30 and 90 days: orders, revenue, AOV, and new-vs-returning split. Then catalog count, the top 15 products by revenue with units, repeat-purchase rate and time-to-second-order, refund and discount rate, and every active selling plan or subscription with its terms. Call out anything unusual in the data. Report numbers only, change nothing.

Step 2 · GA4 pull

AgentYou + Dale

Where traffic comes from and where it converts. Capture the measurement ID while you are in there, it feeds the measurement map in Step 5.

Prompt · GA4 readPull [brand]'s GA4 for last 30 and 90 days, read-only, and be thorough. Give me: sessions and users by channel and top sources, the top 15 landing pages by sessions and by conversion, sitewide and per-channel conversion rate, device and geo split, the top conversion paths, and a funnel from session to add-to-cart to checkout to purchase with the drop-off at each step. Note the GA4 measurement ID and any tracking gaps you spot.

Step 3 · Meta audit

AgentYou + Dale

Audit the ad account as it stands: current campaigns, spend, rough CAC, and the creative library. Check whether the pixel and CAPI are installed. Look at competitor creative in the Meta Ad Library while you are here.

Prompt · Meta auditAudit [brand]'s Meta ad account, read-only, in detail. Deliver: every active campaign with its objective, last 30/90 day spend, ROAS and rough CAC, the top and bottom creatives by spend and result, the audience and placement setup, and account health flags. Confirm the pixel and CAPI are installed and firing, list the events, and note the pixel ID. Then pull [top competitor]'s active ads from the Meta Ad Library and summarize their angles and formats.

Step 4 · Klaviyo audit, not build

AgentYou + Dale

Audit only. We are not building or editing a single flow here, that is the flow-build course. We are mapping what exists, what is missing, and how the list performs.

Prompt · Klaviyo gap mapAudit [brand]'s Klaviyo, read-only, do not build or edit anything. Deliver: list size and growth trend, every flow with its status and last-30-day revenue, a gap map versus the standard set (welcome, abandoned cart, browse abandon, post-purchase, winback, sunset), the segment structure, last-30-day metrics (received, open, click, placed-order value, unsubscribe, spam), and deliverability signals including sending-domain setup. Rank the gaps by likely revenue left on the table.
Every brand is different

No email program yet means Klaviyo is greenfield. Record that as the biggest gap and move on, do not try to reconstruct data that was never captured.

Step 5 · Pixels and tracking audit

AgentYou + Dale

What is installed, where, and what fires. Detect it at the source, do not trust the dashboard's word for it. The output is the measurement map, every ID with its install location and events.

TrackConfirm
Meta pixel + CAPIID, install location, events firing
GA4Measurement ID, events
Microsoft ClarityProject ID, recording live
TikTok pixelIf they run TikTok
Klaviyo onsitePublic key, active site tracking
Google Ads tagIf they run Google Ads

Step 6 · CRO audit, assess only

AgentYou + Dale

Tear down the current funnel: PDP, cart, checkout. Note the current conversion rate and the friction. If Clarity or a heatmap exists, watch real sessions. This is diagnosis. The fixes are the CRO implementation course, not here.

Prompt · CRO diagnosisDo a CRO diagnosis of [brand]'s funnel, assessment only, no fixes. Walk home, collection, PDP, cart, and checkout as a real buyer on both mobile and desktop. For each step deliver what works, the friction points, and the likely conversion impact. Add the current conversion rate, findings from Clarity sessions or heatmaps if available, page speed, and trust and mobile issues. Give me the full issue list ranked by expected lift so the CRO build course can act on it.

Step 7 · Attribution and growth-math read

AgentMargo

The money picture. Margo builds the true-orders base from Shopify and Klaviyo, never GA4 alone, then reads the real numbers. This is a baseline read, not a reallocation plan, the plan comes after onboarding.

Prompt · Margo baselineMargo, run the full onboarding growth-math read for [brand], baseline only, no reallocation yet. Build the true-orders base from Shopify and Klaviyo, never GA4 alone. Deliver: total orders and revenue reconciled across Shopify, platform-claimed, and GA4 last-click with the gap explained, current blended CAC and per-channel CAC using fair-split logic, the brand-search tax estimate from Auction Insights, contribution margin and payback by channel, and where the money is actually working versus wasted. Tag each figure VERIFIED or ESTIMATE with its source.
Lane output

Dale's lane produces the measurement and money baseline: all tracking IDs, current metrics across Shopify, GA4, Meta and Klaviyo, the flow gap map, the CRO diagnosis, and the CAC and attribution read. It lands in the Brand OS measurement registry alongside Dougie's brand pack.

Check yourself: why is Step 4 an audit and not a build?
Onboarding only maps and understands. Building the flows is a separate implementation course. Here we produce the gap map so the build course knows exactly what to build, we do not touch a flow yet.

The two lanes, phase by phase

Both lanes run at the same time and feed one shared context pack. Read across each row: at that phase, here is what Dougie does and what Dale does in parallel.

Phase
Dougie · Brand & Creative
Dale · Data & Measurement
Kickoff context
Reads the Fathom call and scrapes the inbox for the founder's own words and goals. Gmail
Reads the same Fathom call and inbox for stated numbers, systems, and access. Gmail
Access
Gathers brand assets, logos, social, tags into one Drive.
Requests Shopify, Klaviyo, GA4, Meta access. Full-data vs public teardown. You + Dale
Discovery
Voice of customer and every review, clustered into the VOC report. Delphi agent-reach
Pulls Shopify, GA4, and Meta baselines: orders, AOV, traffic, spend. Shoppy
Diagnosis
Reads current positioning, finds the gap, writes the sharper angle. Sloane roast
Audits pixels and tracking, then diagnoses the CRO funnel. You + Dale
Audience
Builds the top 3 buying cohorts. Delphi
Feeds the real sales and buyer data that grounds those cohorts.
The money
Decides brand voice, aesthetic, and design direction. Sloane Designer
Reads true CAC, brand-search tax, and channel profit. Margo
Handoff
Brand context pack: VOC, positioning, cohorts, voice, design.
Measurement baseline: IDs, metrics, flow gap map, CAC read.
Both lanes land in one place: the Brand OS and the brand dossier. That combined context pack is what the build courses consume. Nothing here gets built, it gets understood and written down so the build is right the first time.

Where onboarding ends

When both lanes have filed their outputs into the Brand OS, onboarding is done. What comes next lives in its own courses and its own pages:

Next courseWhat it builds
Website BuildThe actual site design and build, off the brand pack and design direction
Klaviyo Flow BuildBuilding the flows the gap map identified
Ads & Media BuyingLaunching and scaling off Margo's baseline
CRO ImplementationFixing the friction the CRO audit found
Onboarding is not the work. Onboarding is knowing exactly what work to do, and why, before we touch anything the brand owns.