Sign in with your Landing Partners account
Pick yourself in the menu top right (BM or Expert). The agenda and tiles then show your clients and the questions addressed to you. Your choice, tab and filters are remembered in this browser. Someone who is away today (Vacation organisation) shows 🌴 and the day they are back next to their name; hover it to see who covers their clients.
Everything that needs your attention today. By client shows one row per client (click to open the details); By category lists the same items per type, one line each; click a line to open it.
⚠ Problems — data not coming in (no sales, no ad spend, source errors, stale Shopify data), a sharp drop in net sales, or a connection the cockpit can't read (Meta ad account missing or not accessible, Klaviyo key expired).
✉ Client questions — client emails without a reply from the team, from the last 30 days (BMs only).
💬 Team questions — Slack messages where you were @mentioned and haven't replied in the thread.
📣 Ad requests — open requests from Notion (experts only); red when overdue.
📉 Act proactively — red/orange clients with the numbers that are off; a client the team acted on in the last 7 days (seen in Meta, Slack or email, or ✓ Done) says '🛠 the team acted lately' instead and is no to-do.
What a client does and when: drops, pop-ups, sales, collabs, events. On the client page the tab 📅 Planning under the key numbers draws it on a timeline (four weeks ahead; 2, 10 or 13 on a click) with Today, the next meeting and the retail moments (Black Friday, Christmas, the sales) as markers. Add a line with a date picked from the calendar, drag a block or a bar to move it or its edges to change the dates, ✕ asks before it leaves a line out. The AI proposes lines from the meeting prep or from a screenshot or document you hand in (✨ Read a planning…); a proposal waits for ✓ Keep. Only confirmed lines reach the screen for the client (which shows the same timeline) and the recap mail. My agenda → By planning shows your clients on one timeline two weeks ahead; Tools → 📅 Planning every brand; the AI's proposals fold open under its top line, each with ✓ Confirm and ✕ Deny. The Today block says what goes live today, Next up what starts tomorrow.
Click a client (a tile, a name, an agenda row) to open its page. It is made for a check-in: one screen that says where the client stands and what your first click is. On top Where we stand: the status word with one sentence (the AI's reading, when there is one) and the period's key numbers as tiles (net sales, new customers, MER, CAC, ad spend), each with its change against last year or, after the slash, its target; the deviation is red or green, the biggest one outlined. Last 7 days, MTD, Last month and Year to date change the period, as on the Performance tab. Under the tiles one line of tabs, one open at a time: 📈 What happened (open from the start: one chart per day, the sales as columns in front of last year's grey ones, the new customers as a line, a light green background where for 3 days or more both were above last year, light red where both were under, and what we did as an icon on its day: hover it for what it is, click it for what exactly happened and the why (✎ correct); ✗ Not agreed asks one sentence on why not, sends it in Slack to whoever did it and keeps it under Communication until someone clicks ✓ Handled. 📝 Details under the chart holds the rest: the sentence, the period of the chart (7 days, MTD, last month, 8 weeks), + Note on a day, the totals, the legend and the list in words), 🔍 Why (the AI's reading in full with the researcher's reasoning behind it; it is asked for when you open it and there is none, 'Read it again' when it is old), 📊 Numbers per period (the last 7 days, the last 14 days and the month to date side by side, each against the same dates last year, with ✨ What stands out on top, Google in its Channels for a client with Google spend) and 📅 Planning. The block ends with the question box: ask the researcher anything about this client; it answers in your language and links to the blocks its answer leans on. Then The work: five tabs with the number that is open on each, closed until you click one. 💬 Communication: on the left what waits on us, five at rest: the meeting with 'Open the prep ›' (or '✨ Prepare a meeting'), an action someone did not agree with (Show on the timeline, ✓ Handled), an unanswered mail or message from the client (Reply ›, Mark as handled), no contact for 14 days (Mark contact), the overdue plan items, the team's open questions, a Google alarm, the alerts (Hide 14 days, Out of scope) and the notes from management ('+ Note for the team'; 'n hidden or handled' brings a line back); on the right what was said, the last five first: the mails both ways, the Slack messages, the meetings with their recap, each opening on a click; under them '✨ Prepare a WhatsApp check-in' and '✨ Send a content reminder'. 📣 Ads: the open ad requests and the budget skill's proposals, beside the changes in Meta per day and person, the budget skill's moves and the ad requests done in the last 30 days ('📣 All ad requests' under it for status, notes and edits). ✉️ Email: the open Klaviyo requests beside the campaigns that were sent. 🎨 Creatives: the creative refresh round ('Open the creatives ›') and the open creative requests beside the delivered rounds and the requests done; the rhythm and 'Run now' under them. 🛒 Webshop: the open development, data / IT and other requests beside the ones done. + New request on the tab line opens the five kinds (ad, creative, development, Klaviyo, data / IT) for this client. Then three closed rows: 📌 Our plan & actions (Plan = the goal from the strategy, what was agreed and what the AI suggests; Done = what the team, the agents and the tests did; Next = the first things to do with who and when; open, the client's 🎯 Strategy on top, the plan list with who does it, by when and its status, '+ Add activity', and ✨ The AI suggests under it: take it by choosing who, or ✕ Decline; 'Manage ›' on an ad request or a request opens its tab of The work), 📋 Instructions (the targets and the guardrails that hold, the goal from the strategy and the first instruction of each section; open, every section in full) and 📂 More details (who works on it and the links, 'Stepping in? Brief me on this client', 🧠 What the AI learned, 📄 Client file, 📊 Numbers and charts, 📅 Meetings & contact, 🌍 Sales per country, 🛠 Changes in Meta, ✉️ Klaviyo, 🤖 Skills).
Under 🧰 Tools → 🌡️ Month check you see whether it's a good or bad month for all clients together: total sales vs last year, how many clients grow or decline, and why — is it broad (market, season, calendar) or a few big clients, do weekends or moving holidays (Easter, Mother's Day, Black Friday…) shift the comparison, and are the accounts still efficient (MER). The page opens 🔒 Anonymised: no client names, no € amounts and no client counts (only percentages and MER), so you can show it to a client as a market overview; the toggle at the top switches to 👥 With client names (names, amounts, counts, biggest drivers and the per-client table).
Experts see the open ad requests from Notion in their agenda. Change the status with the dropdown next to a request (it asks for confirmation and updates Notion right away). Done also posts a message in the client's Slack channel; Done and Archived requests disappear from the agenda. In a new ad request (+ New request → 📣 Ad request, or 'Turn into ad request' on a client's message), '📁 Pick a folder' next to Creatives and Copy lists the folders of the client's Drive folder (the Drive link on the client's Notion page, two levels deep): pick one and its link lands in the field; pasting a link still works. Without a Drive folder on the Notion page the cockpit takes the folder named after the client on the Landing GDrive ('LP x …') and says it guessed; when there is none either, it says to add one.
Anthony and Charlotte can set a top priority for a client under 🧰 Tools → 🚩 Top priorities (what has to happen, for the BM, the expert or both, optional due date). It shows at the very top of the agenda of that client's BM and/or expert, and the client moves to the top of the list. Whoever handles it clicks ✓ Done.
For brand managers, a client shows up in the agenda when a bi-weekly, bi-monthly or monthly meeting with that client is planned in the next 3 working days (found in your calendar by the meeting title: "bi-weekly", "tweewekelijks", "bi-monthly", "monthly", "maandelijks" …), so you can prepare it. On the morning of the meeting the cockpit writes a meeting prep, on the numbers through the day before: an elevator pitch in the language of your mails with the client that you change, ask the AI to change and approve. Want it sooner? '✨ prepare the meeting' on your agenda writes it at once; when the numbers are read again on the day of the meeting, the AI puts the new figures in what you say and in the takeaways on the screen for the client. You decide first: '✨ Prepare it now' and '✨ Rewrite…' ask which period the prep is about (the last 7 or 14 days, this month so far, last month or your own dates; the team's day rule is chosen for you; the key numbers beside the prep follow it), propose what to talk about (click a topic to leave it out; the numbers are always in) and leave room for your own story, what you want to tell or what the cockpit can't know: the prep is built around it. At rest the prep is the one message and the steps of the meeting ('In this order' is a bullet list: type, Enter for a new point, delete what you don't want); the reasoning, what to check, the actions to agree, the follow-up, the numbers and your notes are rows you open when you need them, and the researcher answers your questions in the chat beside the prep. Communication and storytelling: one step of the meeting brings ideas for the client's communication the coming weeks or months: around a moment (Black Friday, the sales), something that plays at the brand (a drop, an anniversary, a collaboration) or something that works in the ads; all the researcher's ideas, each with why it fits this brand, are under 📎 Behind this prep → The research, and you can ask him more in the chat ('what do we tell until Christmas?'). The prep is its own page: 'Open the prep ›' in the 📅 Meeting prep row of the client page (or 'prep ready ›' on your agenda) opens it as one task with a start and an end, in four steps: 1 where we stand (beside it the period's sales, MER net and CAC against the target and new and returning customers against last year, each with its green, orange or red dot), 2 what you say, with the core of each sentence in bold, then side by side 3 since the last meeting (first the follow-up of what we agreed then, then the changes in Meta, what the agents did and the tests that ended, each with its why) and 4 what comes next (first what you agree with the client today and what you ask them, then what already runs in our plan: who does it by when, the tests that run; pass a playbook action on, mark it done or choose who takes it and by when in its row); ✓ Approve sits in the bar at the bottom. Once the meeting has started the page opens on 📝 Your notes: type while you talk, it saves by itself (the foot of the box says 'Saving…' and 'Saved 11:42', also when you come back later). Nothing to press after the meeting: the recap comes by itself once the Fyxer notes are in your mailbox; '✨ Make the recap now' turns your notes into the actions and the mail without waiting (the Fyxer notes are then not folded in by themselves: make it again once they are in), and the line under the box says what a recap was made from. The headline is the big picture in two or three sentences, top-down: the webshop as a whole against last year first, then where it sits (orders or basket, new or returning customers) and whether it comes from the spend or the efficiency; the story under it says why and what we propose. Under the headline one line names the biggest moves against last year (software); a click opens the tables: this period and the last 7 days against last year (the moves that matter highlighted), Meta and the other channels, the campaigns, and last year around this period (spend per week and when the scale-up started); a creative near fatigue or a MER flattered by low spend is named under 'To check'. The row says who sits at the table, from your calendar invitation. Click a step of the run of show, a question or a line to change it in place (Enter saves, an emptied line disappears, the dashed line adds one) and drag it by its number (or the ⋮⋮ grip at its left) to change the order; decisions and questions are one block, 'What we propose next'. After the meeting the cockpit looks for the Fyxer notes every quarter of an hour, from 45 minutes after the start, in your mailbox and in the mailbox of whoever approved the prep; as soon as they are in, it lists the actions: who does what by when. You check them, and the cockpit then writes the recap mail from those actions the way the BMs write their own (the attendees by name, a bold heading per topic, what we do and what you do, two or three direct questions, the next meeting; in the name of whoever makes or sends it: when you cover a client while its BM is away, it ends with your name); the update for the expert is the actions themselves, posted in the brand's Slack channel with the expert tagged when you send it out. The recap is not a to-do on your agenda: send it right after the meeting, from the meeting's page. The client page still shows it until it left. After the meeting it reads like a client email: 1 on the left what was said (the Fyxer notes, read from your mailbox, and your own notes, the same field as in the meeting, with '✨ Make the recap again'), on the right what comes out of it, in two steps. 2 The actions (what we agreed, who does it by when; the update for the expert sits folded under them): change one where it stands, then ✓ These are right: write the mail (every action needs an owner and a deadline; a change after it takes 'right' off and the mail follows when you mark them right again). 3 The recap mail (to the client's people from the invitation, your colleagues in cc): change the text in place, tell the AI under it what to change with Enter, '✨ Fold in what is prepared and done' puts the AI's analyses and the handled actions into it. A follow-up that is a colleague's work (visuals → Sybi, development or Klaviyo → Buba, data → David, by name or by the words) gets a yellow line '🧩 Looks like a request for …': 'Make it a request' opens the request form prefilled, 'Not a request' hands it back as a draft. 'Done' on an action marks it handled (the mail can say so); it is not an approval. Then send it out: Send to … sends the mail from your own Gmail after a yes (only the person in whose name it is written can send it, and a [ ] the cockpit did not know is never sent), or ↗ Draft it in Gmail puts it in your drafts to send there. Either way the update is posted and the actions go where they belong: yours on My agenda (📅 From the meeting, with ✓ Done, in Today when due), the client's in the mail, a colleague's as a 🧩 request in one click, a decision for management by Slack DM to Anthony and David. There is no archive step: the cockpit sees the recap mail in your Sent mail and closes the meeting by itself; the recap stays under 📅 Meetings & contact ('✓ Sent another way, close it' when you mailed it differently). Sent from Gmail or another way, the update still goes to the expert in the brand's Slack channel when the meeting closes, if it was not posted yet. The update for the expert is yours to change too: click its text in the fold under the actions ('↺ The cockpit's version' undoes it). The cockpit learns from what you change before it leaves: when you send out, it compares the mail with the AI's first version (and what you asked the AI to change), the update with what it proposed and the actions with what the AI took from the notes, and writes down at most two lessons each, for this client (🧠 What the AI learned on the client page) and for how you write (your next recaps, updates and actions follow them; an update then says '✨ Written with what your earlier updates taught'). The line under the send-out says what it learned; Tools → 📊 Usage counts how often each part leaves unchanged, lightly edited or rewritten. An action's owner is who does it: Client, BM, Expert, Team (Sybi, Buba, Tanja) or Management (Anthony or David, a decision beyond the BM's mandate). A follow-up action the AI can prepare (an analysis from the cockpit's numbers, a draft, a request for a colleague) is prepared right away with the recap: the result sits under the action; type a question or a change under it and press Enter, and the AI looks again. Nothing goes out by itself, and the recap is drafted whether or not you approved the prep. What you change or ask before you approve, the AI keeps as lessons for that client (🧠 What the AI learned), so the next prep starts closer to how you want it. On the client page 'Last contact' is a link to '📅 Meetings & contact' under More info: what the last contact was (the email with Open in Gmail, or the meeting) and the earlier meetings with the recap mail, the decisions and the follow-up actions.
When a client is orange or red, Act proactively shows the one to three actions from our Client Solutions Playbook that fit this client now, chosen by AI on the client's own numbers (MER and CAC vs target, sales vs last year, trend, what the team already did), with the reasoning and a first step. ✓ Done when you did it (it asks what it gave: it helped, wasn't it, or too early to tell; the client moves to "actions taken" for 7 days); → Pass to sends the action to the client's expert or BM as a Slack message from the cockpit; Not now asks why and hides it for 14 days; Read in the playbook opens the full action. The advice is written when you ask for it: ✨ Get advice now on the client (about 20 seconds; the same button writes it again later). An advice older than a week is not shown any more, since the numbers moved on. After 7 and 14 days the cockpit checks whether the numbers improved (Tools → 📖 Playbook).
When a client is orange or red and you are already working on something that is not in the playbook advice, click Other action taken under "Act proactively" and write what you're doing. The client moves to "actions taken" (with your note) for 7 days; if it is still orange or red after that, it comes back on your agenda.
For brand managers: when there was no email and no meeting with a client for more than 14 days, the client shows up in My agenda (also when a meeting is already planned; its date is shown), and in the opened row the 'Last contact' line turns red and carries the two moves: ✨ Prepare a WhatsApp check-in and Mark contact. The cockpit reads only the sender, recipients and date of emails and the time, title and attendees of meetings in the BM's mailbox and calendar (never the content); newsletters and notifications don't count. When a meeting in the calendar might be an event rather than a meeting with the brand (a crowd, people from several companies, or the client only in the title), the cockpit asks under it: Is this a meeting with the brand? No takes it out as an event, with no prep and no contact; Yes keeps it and the question does not come back; ↩ Undo puts it back. The cockpit never decides this by itself. A planned meeting that never happened: Did not take place on its line in the client page's Communication overview (or in the after-meeting task): no recap to send, it does not count as contact, the prep is set aside and the meeting history keeps the trace; ↩ Undo puts it back. A meeting that took place but needs no recap from the cockpit (sent from Gmail, or none needed): Recap sent or not needed on the same line closes it as it is; the actions and the mail stay under 📅 Meetings & contact, the expert gets the update only when you marked the actions right. Contact by phone or WhatsApp isn't visible yet, so mark it with Mark contact. ✨ Prepare a WhatsApp check-in next to Mark contact writes a short message to the client in their language and your tone, with one concrete hook from the facts (what went live, a good number, the next drop or meeting); change it, copy it or open WhatsApp with it, and send it yourself. What you change or ask the AI to change before you copy it, it keeps as a lesson for that client (🧠 What the AI learned on the client page). Waiting for content from the client (photos, a shoot, product info)? ✨ Send a content reminder next to it on the client page asks in one line what you are waiting for and writes a friendly reminder in the same way; open it in WhatsApp, copy it, or ↗ Put it in my Gmail drafts: a reply in your latest email thread with the client, which you send from Gmail.
Under 🧰 Tools → 🌴 Vacation organisation you add who is away, from–to, and a replacement for every client (required). During that period the replacement sees those clients (with their problems, questions and ad requests) in their own agenda, marked 🌴. Afterwards everything switches back automatically. Overlapping absences: when you cover a colleague's clients and are away yourself in that period, those clients show on your absence as 'nobody yet' until you set who takes them over (Set replacements); 'Clients per BM on a date' shows who carries what on any day. Handover to-do's: as soon as you fill in who and when, the AI proposes per client what your replacement should do (from your open questions, meetings in the period, ad request due dates, projects, alerts and priorities); untick, change or add what you like, also to-do's not about one client. Your replacement gets them in Slack and on their agenda (🌴, with ✓ Done / Skip and a note), and can add to-do's for you; the welcome-back briefing shows what was done, skipped and still open (open ones come back to your agenda). Back from holiday: on your first login after an absence of 3 days or more, a 👋 Welcome back block on top takes you through two steps: what happened with your clients, and what changed in the cockpit while you were away. Handover briefing: when you save the absence, the AI writes per client the handover page the team used to write in Notion (workload, key focus, to-do's, watch out, plan B, contacts, links) from the cockpit's facts and your optional notes, plus the main priorities on top and how to reach you. It is under '📋 Handover briefing' on the absence, on the replacement's agenda (🌴) and inside 'Brief me on this client'; the replacement can leave notes and to-do's for you there, which you see when you are back.
When a client on Klaviyo has sent no email campaign for two weeks, the BM gets the alert 'No email campaign in N days' with ✨ See the email ›. On weekday mornings the cockpit prepares one: the AI writes it from the client's best earlier campaigns, the homepage, the newest products, the best sellers and the ads that sell in Meta, and the cockpit builds it as a real design in the client's own Klaviyo, in the brand's colours, fonts, button and logo. On the client page, under To do, the row ✉️ Campaign email shows the send details (from, to, subject, preview text, the proposed day and hour) and the mail as it will go out. Change it by asking the AI in the line under it; the email is rebuilt. 📋 Copy the link for the client gives you the subject and a link to the mail for your own mail or WhatsApp; once the client says yes, ✓ Client approved: draft in Klaviyo makes a draft campaign to the same audience and from the same sender as their last one. Nothing is sent by itself. In a meeting the screen you share with the client shows the same proposal on its own tab (Mailvoorstel), in the client's language, with ✓ Client approved for a yes on the spot. Not now asks why and removes the design from their Klaviyo; Out of scope on the alert stops it for a client whose Klaviyo is not ours.
A request for Sybi (creative), Buba (development, Klaviyo) or David (data / IT) starts from the '+ New request ▾' tab (on a client page it is for that client at once) or Tools → 🧩 Requests: the type decides the person. A data / IT request can be about Landing itself rather than a client (pick 'Landing (general)' in the client list): David gets it as a direct message and the board says Landing. The form asks what the person needs to build it: what exactly, the material and where it is, what the client expects, who signs off (the hint under Details says it per type), a deadline, a link to the material or the brief, and photos or visuals (📎 Add photos, or paste a screenshot in the details; on the client page too). The person gets the whole request in the client's Slack channel with a link that opens it. The board: Sybi's and Buba's home (My agenda) and Tools → 🧩 Requests (everyone, filter by person) read left to right, To do · In progress · In review · Done; drag a card to where it stands. A card shows the client, what, the kind, the first lines of the details, the photos, the latest note, when it is due (overdue in red, this week in orange), who asked it and whether there is a link. A click opens the card where it is, its column wider, no pop-up (on the client page's 🧩 Requests row a request is a row that opens in place): the details in full, the photos, the link, who asked it and when, the BM of the moment, the notes, the person's meetings with that client, and you work in that row: click the title or the details to change them (Ctrl+Enter or a click outside saves), the priority, the deadline and the person are fields where they stand, 📎 Photo adds a photo (or paste a screenshot in the details, × removes one), + Add a link to the material, the status in one click (To do · In progress · In review · Waiting · Done), + Note, 💬 Ask
Whoever works on a 🧩 request can leave notes: what they did, what is next, what they are waiting for. Open the request, click + Note, type and press Enter. The newest note shows in the opened request with who wrote it and when; click it to see all notes and add one. Only the writer can remove a note. 💬 Ask <requester> next to it sends what you type to whoever asked the request as a Slack DM from the cockpit and keeps it here as a note ('💬 to Nora: …'). The request itself (its text and status) does not change.
When you add a 🧩 request (on the client page or Tools → 🧩 Requests), attach photos, visuals or screenshots with 📎 Add photos or paste one in the details (max 6); on an existing request, 📎 Photo in the opened row or a paste while changing the details. They show in the opened request for whoever builds it; click one to open it full size, × removes it.
The client page shows two MERs. MER net = net sales ÷ ad spend (net sales = gross sales minus discounts and returns, without VAT and shipping): the target and the colour use this one. MER total sales = total sales ÷ ad spend (total sales = net sales + VAT + shipping, as in Shopify): always higher, about 21% with Belgian VAT; the reporting API and the reports built on it show this MER. Compare like with like, steer on MER net. Both drop in months with many returns (e.g. after the sales), because Shopify books a return on the day it comes back, not on the day of the order.
Sales vs last year leads: growth is never red, decline is never green. A red MER or CAC (vs the target in Notion) moves the colour one step down. Grey = no data or no target.
The Performance tab shows every client as a tile. The buttons under it choose the period: Last 7 days (the last 7 complete days against the same weekdays last year), MTD (this month to date), Last month and Year to date. It opens on MTD, also in the first week of the month, when month to date covers only a few days (the agenda then uses the last 7 days). The tab remembers the period you chose last, and a tile opens the client page on the same period.
Net sales (Shopify, online store), spend and customers: BigQuery (Shopify + ad platforms). Targets, BM and expert: Notion. Questions: Slack and Gmail. Ad requests: Notion. The data refreshes every few minutes ("Updated … ago" top right).
Write it in the box above: the AI answers a question right away, and sends a bug, an idea or a project to the cockpit backlog (for an idea or a project it first asks up to three short questions, so the card says what it is for, for whom and where) (you see how it was sorted and can correct it). A bug or a small improvement can be fixed by the AI on its own test cockpit: you then get a Slack DM with the link, and the fix goes live with the next AI run (within 6 hours) unless you send it back with ✗ not fixed. New features and anything sensitive go through a person first.
Tools → 🚀 Onboarding holds a new colleague's onboarding: your first weeks, step by step, on your own clients. On day 1 you set up and learn (your tools: Slack, Drive, Notion, 1Password; meeting the team: the Monday team meeting, the team's intro videos and your own; planning your monthly feedback meeting with Anthony; the training levels, a briefing per client, for experts the house rules for Meta), then you practise without changing anything for the client (a get-to-know talk with everyone, where you tick each colleague you met; a handover talk per client; the instructions for the AI, Changes in Meta, a dry run, your agenda, ad requests, client and team questions), and then you do it yourself (an expert their first budget round and a test, a BM the recap after a meeting, a request for the team and playbook advice), and it ends with the final quiz and your first feedback meeting with Anthony. Everyone asks access to their clients in Meta Business Manager and to Klaviyo (via access@landing.partners), works a few weeks in Klaviyo and follows the clients' ad accounts in Meta while practising, and only then does the Klaviyo Practitioner Certificate, Facebook 100-101 and Claude 101 (they wait, faded with 'after …', until that work is ticked; each is done by uploading your certificate, a PDF or a picture, which management can open under 👥 New colleagues), and has a get-to-know talk with everyone in the team. A new colleague shadows the client meetings of colleagues for two weeks: Anthony, Charlotte or David pick whose in your to-do's under 👥 New colleagues (🗓 Shadow client meetings) and the cockpit invites you, without a mail to the client. Most steps tick themselves the first time you do them in the cockpit; a talk you tick yourself (✓ Done, ↩ takes it back). It starts by itself the first time someone new opens the cockpit, with ▶ Start my onboarding on that page, or when Anthony, Charlotte or David start it for you under 👥 New colleagues on that page (🚀 A new colleague: your e-mail and role; it begins at your next login), where they also follow how far each new colleague is and tick what management sets up for them (☑ Set up: accounts, Drive, Salto, hardware; what the cockpit can see ticks itself); ✕ takes someone off that list and ↩ puts them back; while it runs a banner names your next step, and the Wednesday session and the monthly training wait until it is done. If 'The cockpit knows who you are' stays open, you are not yet on your clients in Notion: ask Anthony or David.
Tools → 🎓 Training shows the things that go on after your onboarding, each as one row with a dot (green: nothing open) and what is open. 📌 Mandatory trainings (only when one runs for your role): what management asks you to do, each with the date it is due (red when it is past); do it, then tick ✓ Done, or upload the certificate when it asks for one. A banner reminds you once a day while one is open. The monthly training: five minutes a month: what came new for your role since last month (two or three lines each), then five quick questions from the course, a different five every month (80% to pass, retry as often as you like); the finale shows your score and how many months in a row you did it. A banner reminds you until it is done. The course itself (level 1 Landing Partners, who we are; level 2 the cockpit; level 3 your role; for media experts level 4 working with AI and the Meta account) you learned in your onboarding, and any chapter can be played again from the row. The Wednesday session: on Wednesdays (when there are at least two new things for your role) the cockpit plays a short session of what is new: the changes that change how you work, each as two or three lines (the small fixes and details only sit in 📢 What's new, behind 'smaller changes'); features you already used are skipped, and the row shows the last six weeks. Everything is kept per person and only you see your own; Anthony, Charlotte and David see the whole team under 👥 Team (the mandatory trainings, the month, the sessions), open one colleague's page as that colleague sees it by clicking the name, add a mandatory training with its date under 📌 Mandatory trainings (for everyone or a role; 📣 tells the people who still have to in Slack) and can send a 🔔 reminder in Slack per open thing. The course itself is part of the onboarding (Tools → 🚀 Onboarding).
On a client page, under the row 📋 Instructions, the people on the account (the client's BM and expert, their replacement during an absence, and management) write in their own words what the AI must know. First the Strategy: where we go with this client, in a fixed small format of one line each: Goal (where to and by when), Bet (what we bet on and why, at most three), Measure (which numbers prove it; the targets and the budget stay where they are) and Not now (what we deliberately leave aside). It is internal, anyone on the account writes it, and every AI decision reads it first: the meeting prep says where we stand against it ('Against our strategy', the bi-yearly reviews every bet), the recap mail ties the actions to it without quoting it, the playbook advice and Automatic budget name the bet an action serves and flag what goes against it with '⚠ not in the strategy', the handover briefing opens with it. Next up asks the people on the account to write it when there is none and to review it after six months. Then About this client (what counts for this brand, who decides, what they always ask; for a brand that is not fashion also what replaces collections, drops and sale months in our rules) and How we talk to them (tone, language, do's and don'ts in meetings, mails and WhatsApp). The meeting prep, the recap after the meeting and the WhatsApp check-in read them, and they weigh above what the AI learned (🧠) and above the skills' general rules (🛠). A lesson under 🧠 What the AI learned can be promoted with '↑ Make it an instruction'. Under an empty section, '✨ Ask the AI for a first version' lets the AI suggest one from how we worked with this client so far (the client file, the lessons, the decisions, earlier preps, your last recap mail and the client's replies); you use or rewrite it, nothing is saved by itself. For a written section, '✨ Ask the AI for a new version' (say what should change if you like) shows the differences line by line: lines that stay in grey, new lines in green with + Add, lines to drop in red with ✕ Remove, or Take all. Every change is logged with your name in the why log. Write one instruction per line: click the list to change it, Enter starts the next bullet. The third block, Advertising, says what the ads may say and show (languages per market, banned words, formats, who signs off) and what the AI may change alone or only propose; the ad request draft, the copy review and the playbook advice read it. Under it the Guardrails are fields, not sentences: the budget (min and max, all channels): one for every month, or, for a client that agrees it per month, a month of its own under 'Different per month' (a month without its own takes every month's), max change per step, the channels and markets the AI may act on, 'hands off until' a date, and only when a client wants it an amount above which a person approves (empty: nobody, the AI acts). At rest the block shows what the AI steers on this month (the MER net and CAC targets from 🎯 Targets), the budget when one is set and one line with what holds; the fields are details, under ✎ Change the guardrails. Software enforces them whatever the AI concludes and spreads what is left of the month's budget over the days left; an empty field means the agency default (shown greyed), and 'From the account' shows what the account did so far (the biggest budget step, the spend last month and this month, the countries with sales) to set them with evidence. Automatic budget (the last line, for everyone: the dry run itself changes nothing) opens on the diary of the last two weeks: one line per round the cockpit did in the account (by the AI by itself on the days set under Guardrails → 'Runs by itself on', or by a colleague's click), on a click the changes, what was left out and why, and ↩ Undo until the next round. 💬 Tell the AI under the latest round takes your own words: what the AI should remember becomes a lasting lesson for this client's rounds (under 🧠 What the AI learned), and what you want changed now ('NL −10%, BE +20%, BE has room') comes back as the changes in Meta, with what a guardrail blocks and why; '▶ Do it in Meta' makes them (the max step, the month's budget and 'hands off' hold; the 24-hour wait and a hold of the numbers do not, the judgment is yours), as a diary line under your name with ↩ Undo for a day. Under it, the dry run of the Meta skill: the cockpit reads the account and the numbers of the last 7 days (14 for a small account), says whether the rules say scale up, scale down or hold, and shows what the AI would do per campaign that spent money in that period (active ones without spend are left out), grouped by the stage the AI gave them (TOFU, MOFU, BOFU, PC, catalog, reach …) with why each campaign sits there, and a table of the daily budget per stage now and after, in € and %; in two sections. 💶 Budget scaling: the decision and one plain sentence on what the AI would change and when it can happen; a row per stage (budget now and after, the change, its share) that opens on that stage's campaigns and their moves; 'How the AI classified the account' on a click; and a box to ask why ('why a scale down?': answered in seconds) or to tell the AI what to change ('Winter retargeting is PC, not BOFU'), which runs it again without reading the account again and can be kept for the client as an Advertising instruction. Every suggestion and test says how sure the AI is that we should do it (e.g. '82 % sure'). Agree with it? Under the stage rows, '▶ Do it in Meta' (for the client's BM and expert, whoever covers them today, and management, on an account the cockpit's Meta user may change) names every change first and, once you confirm, reads the account again and makes the moves that can go now and the ones a person decides: the daily budgets (a campaign's own, or its ad sets' by the same %) and the pauses. A move whose campaign changed in the meantime (its budget, or someone touched it in the last 24 hours) or that a guardrail blocks now is left out with the reason; a move that waits is not done (run it again tomorrow). The block then shows what changed line by line, the expert gets it in the client's Slack channel, the change history names you, and '↩ Undo' puts it back for 24 hours, except what someone changed since. Rather leave it to the expert? '✓ Approve: post it in Slack for
Every text the AI writes for you to read (the researcher's answer on the client page, Ask anything, the playbook advice, the Meta round, the briefings) follows one rule: the first sentence is the answer, then at most three short paragraphs or bullets, at most two numbers in a sentence and each with its meaning (MER net 9.5 against a target of 20), what to do as a list with who and by when, a playbook action by its title, one phrase in bold. The rule is on the skill cards (Tools → 🛠 Skills, 'read it').
Under 🧰 Tools → 🛠 Skills everyone sees every skill the AI has, in three groups: the ones that talk to a client for you (meeting prep and recap, WhatsApp check-in, Meta account management), the ones that work for the team (draft reply in Slack and email, ad request draft and copy review, playbook advice, handover and briefings, feedback sorting, instructions for the AI, Ask anything, the client page's plan with its whys and tests) and the ones that make the others better (the AI shadow, learnings and rules). Per skill: what it reads, the rules it follows (readable on the card, behind a click: a prompt file for the client skills, the rules in the code for the others), the guardrails software enforces, how it learns (what the team's edits, verdicts and corrections do with it, or 'does not learn yet'), what success is (the measure, with its number of the last 30 days when software counts it, e.g. replies sent as drafted, done advice measured better, whys left as read) and its track record of the last 30 days. On the card, 'read it' opens the rules the AI follows, word for word; changing them is development work (a card in the backlog), decided by management. A client's 📋 Instructions override the rules. Under 🧠 Rules the practice suggests the cockpit proposes skill rules from the lessons that come back with three or more clients (software counts the clients and writes the evidence line, the AI words the rule, the examples are anonymised); management confirms, which makes a backlog card of the type '🧠 Skill change · small' (one rule added to a rule file: the factory may build it once the card is approved, and puts it live itself), or removes it. A message in the 💬 box about how a skill should work is sorted as a skill change too: small (one rule or wording in a rule file, track B) or big (what the skill does: new steps, other data, several files, track C, a person builds it).