--- title: "Automatic Weekly Report From 4 Systems | JustAutomate" description: "One Monday email with KPIs from your CRM, GA4, accounting and project tool, plus the variance against plan. No manual collecting. Live in 2 to 6 weeks." url: "https://justautomate.ai/en/automations/automatic-weekly-report/" lang: en --- Reports and data # Automatic weekly report waiting on Monday before you open the laptop On Friday evening the workflow queries your CRM, Google Analytics 4, the accounting system and the project tool, drops the numbers into one template, works out the variance against plan and on Monday morning sends the report by email or to Slack. The management meeting starts with conclusions instead of a data hunt. Book a free consultation [See how the workshop works](https://justautomate.ai/en/services/process-workshop/) - 1 emailon Monday morning instead of five browser tabs and four people, each pulling numbers out of a different system - 4 systemsCRM, Google Analytics 4, accounting and the project tool, queried one after another over the weekend while nobody is working - 2 to 6 weeksfrom the workshop to the first automatic report in the inbox, checked against your own data from the last few weeks The problem ## Why does the Monday review start with an investigation instead of numbers? The board asks how the week went and the hunt begins. Somebody logs into HubSpot or Pipedrive and counts new leads and the deals that moved along the pipeline. Somebody else opens GA4 and tries to split paid traffic from organic, because GA4 does not hand you that split in a readable form on its own. A third person messages finance for the list of unpaid invoices in the accounting system. A fourth walks through Asana, ClickUp or Jira task by task to say what is late and why. Putting those four fragments into one slide or one email takes an hour, two, sometimes three. It cannot be done in advance, because Friday evening numbers are not Monday numbers. The job lands on whoever does it on the side: an office manager, someone from the board back office, the most junior analyst. In a week when something urgent happens, the report is late or it does not appear at all. That is exactly the week the board needs numbers rather than impressions. Do the arithmetic at your own company. Four systems, four people, an hour or two on each of the 52 Mondays in a year. On top of that, numbers that do not agree, because the same deal is worth one amount in the CRM and another in the invoicing system and nobody knows which one is true. The report that was meant to give the board certainty turns into the argument at the meeting. - An hour to three every week from a person whose actual job is something else - The report late or skipped in the week it was needed most - A deal worth one amount in the CRM and another in the invoicing system, with nobody sure which is true - Decisions taken on we seem to be doing fine, instead of a number with a variance against plan Why now ## What changed, so that the report can assemble itself? Every system a person queries by hand today has an API. HubSpot, Pipedrive and LiveSpace return leads and pipeline, GA4 returns sessions and conversions split by channel, Fakturownia, iFirma, Comarch Optima and sevDesk return the invoice list and the money that came in, Asana, ClickUp and Jira return tasks with their due dates. A workflow in Make or [n8n](https://docs.n8n.io/) asks all of them on Friday evening or Saturday morning and does not wait for anyone to log in. The [CRM vendors document these endpoints themselves](https://developers.hubspot.com/), so this is a read, not a workaround. The second change is a language model inside the workflow. The numbers from four sources go into the template and the model writes a short commentary: what is off plan, what needs a decision, what is noise. The reader gets conclusions rather than a table to interpret alone. The template stays the same every week, so the board knows where to look for what. The third change is an AI agent between Mondays. The report gives you a rhythm: same format, same hour, same recipients. But the question how many invoices are more than 30 days overdue turns up on a Wednesday. BOND, our company AI agent, answers it from the same sources in 5 to 10 seconds, with no waiting for the next report. Process first. Tools second. What changes ## What does the board get, and what does the person who assembles the report today get? - ### The report is in the inbox on Monday morning The workflow starts on Friday evening or Saturday morning, queries every source in turn and drops the result into the template. The email, or the Slack message, is waiting before anyone opens a laptop. Same template, same hour, same recipients every week - ### Four systems, one document The CRM gives new leads, closed deals and pipeline value at the end of the week. GA4 gives sessions and conversions split into paid and organic. The accounting system gives revenue, unpaid invoices and cash position. The project tool gives tasks closed, tasks late and tasks newly created. No manual export to a spreadsheet and no retyping of numbers between systems - ### The variance against plan is calculated, not guessed Every metric has a target agreed in the workshop. The workflow works out the difference and lists what needs a board decision, instead of leaving that analysis to whoever opens the email. A decisions list at the top of the report, the detail underneath - ### Numbers that match to the last cent Before the switch we run the workflow on your data from the last few weeks and compare it with the reports somebody assembled by hand. Only the version where the figures agree to the last cent and the last unit goes into production. A test on history instead of faith that a connector counts the way a person does - ### A silent failure does not turn into an empty report An expired CRM token or a renamed field in GA4 are the two most common reasons an automatic report quietly stops working. Monitoring raises an alert about the failed pull before the report reaches the board. An alert to the workflow owner, rather than a half empty report on Monday morning - ### Questions between Mondays Which deals have been sitting in the pipeline for more than three weeks, how many invoices are over 30 days late, how does organic traffic look month to date. The AI agent answers from the same sources the report is built on. An answer in 5 to 10 seconds, with no login to five systems Proof ## Do you collect your own review numbers this way? Yes. Since March 2026 the data behind the reviews inside our founders' agency has been fed by an AI agent, the same one behind the figures on our AI for business page. Below is the scale of that work, together with what the number does NOT prove. Built for ourselves. These are not client deployments but implementations inside the founders' own companies. We say so plainly, because proof you cannot verify is not proof. - ![JustIdea Agency](https://justautomate.ai/klienci/justidea.png)![JustBot](https://justautomate.ai/agenci/justbot-80.webp)JustBottheir agent JustIdea Agency, a performance marketing agency of around 30 people 23 people from the company work with the agent 38 systems wired in permanently 112 processes run without a human 475 thousand lines of the agent’s code The agent is wired into Google Ads, Meta, the CRM, the task system, the SEO tool, Analytics, Search Console and the data warehouse. First commit on 1 March 2026, and 23 people use it today. 8 of 38 systems it works in - Google Ads - Meta Ads - HubSpot - ActiveCollab - Ahrefs - GA4 - Search Console - BigQuery Before Before the agent started, numbers from the ad accounts, Analytics and Search Console reached the summaries whenever somebody pulled them. Nobody measured how long that took, and that matters for the figure below. After The agent loads the warehouse on its own, with no request from a person, and the team asks about numbers in a plain sentence instead of opening five tools. A person reads the result rather than collecting it. 3,506 data warehouse load operations in 30 days, with no person involved And in your company ## What if our data is not ready for a report? Four sentences we hear most often on the consultation, and what we answer to each of them. If you are after a narrower report, sales on its own has a separate page for the [automatic sales report](https://justautomate.ai/en/automations/automatic-sales-report/), finance has one for the [automatic financial report](https://justautomate.ai/en/automations/automatic-financial-report/), and paid channels have one for the [automatic campaign report](https://justautomate.ai/en/automations/automatic-campaign-report/). This page is about the one weekly document that pulls all four areas together. - Our CRM data is incomplete, the report will show nonsense. It will show exactly what is in the CRM, and that is the first effect of the rollout: the gaps are visible on Monday instead of at the end of the quarter. In the workshop we agree which fields have to be filled in for a metric to mean anything, and the report starts with the metrics that are already reliable. The rest join once the data matures. - We have a Looker Studio dashboard, why would we need an email? A dashboard has to be opened, an email arrives by itself. The difference is who does the work: a dashboard waits for somebody to look at it and read the charts, the report brings the variance against plan and a decisions list. If your dashboard really is read every Monday by everyone who should read it, we will say so and we will not duplicate it. - The board wants different numbers every week. Usually it wants the same six to ten, they have simply never been written down. The workshop ends with a list of metrics, each with a definition and a source. One off questions outside that list go to the AI agent, and the report stays fixed, because only a fixed report can be compared week to week. - We have no IT team, who is going to maintain this? The person who assembles the report by hand today. After the rollout they can add a metric, change the recipient list and edit the template, because a workflow in n8n or Make is visible step by step rather than buried in code. Access to your systems stays on your side and we do not hold your passwords. CRM HubSpot, Pipedrive, Salesforce, LiveSpace Analytics and data Google Analytics 4, BigQuery, Google Sheets, SQL databases Accounting and finance Fakturownia, iFirma, Comarch Optima, Subiekt, sevDesk, Xero Projects and tasks Asana, ClickUp, Jira Report delivery Gmail and Google Docs, Outlook and Microsoft Teams, Slack, PDF Automation layer n8n, Make, custom integrations over the API ## What goes into the report and where each number comes from The blocks a weekly report is most often built from. Which ones make the first version is settled in the workshop, because a block fed by unreliable data costs you trust in the whole report. | Report block | Data source | Example metrics | | ----------------------- | ----------------------------------------------------- | -------------------------------------------------------------------------- | | Sales | CRM: HubSpot, Pipedrive, LiveSpace, Salesforce | new leads, closed deals, pipeline value at the end of the week | | Traffic and conversions | Google Analytics 4 | sessions, conversions, the split between paid and organic | | Finance | Fakturownia, iFirma, Comarch Optima, Subiekt, sevDesk | revenue for the week, overdue invoices, cash position | | Projects | Asana, ClickUp, Jira | tasks closed, tasks past their due date, tasks newly created | | Variance against plan | targets written down in the workshop, Google Sheets | difference against the weekly and monthly target, decisions list | | Commentary | a language model over the blocks above | three to five sentences: what is off, what needs a decision, what is noise | | Delivery | Gmail, Outlook, Slack, Teams, Google Docs, PDF | Monday morning, same hour, fixed recipient list | We start with the blocks whose source data is reliable today, usually two or three of them. The rest go in once the data matures, not because a connector happens to exist. How we do it ## Five steps from a manual investigation to a report that arrives on its own Each step ends with something you can judge before you decide on the next one. 1. ### Workshop We sit down with the person who assembles the report today and with someone from the board who reads it. We write down which systems have to be queried, who signs the numbers off before sending, how often somebody exports data to a spreadsheet, and who actually reads the report against who is only on CC. Metrics nobody reads are dropped. 2. ### Report design A list of metrics, each with a definition, a source and a target. The tool choice: n8n or Make, and for an older ERP with no ready connector, a custom integration over the API. The output format: Google Docs, a spreadsheet, a Slack or Teams message, a PDF for the board meeting. If a Google Sheet is enough, we do not propose an application. 3. ### Build and a test on history We build the workflow and run it on your data from the last few weeks. We compare the output against the reports assembled by hand, to the last cent and the last unit. The gaps usually show that two systems count the same thing differently, and we settle that before the report reaches the board. 4. ### Switch and training The first automatic report goes out alongside the manual one. Once they agree we switch over, turn on error monitoring and alerts for a failed pull. The person responsible learns to add a metric, change the recipients and edit the template without raising a ticket with us. 5. ### Measurement after 30 and 90 days We check whether the report is read, whether the targets are realistic and which metrics the board skips. After 90 days something usually leaves the report rather than joining it. A short report gets read, a long one gets scrolled past. Before and after ## How Monday changes for the person who assembles the report and the board that reads it ### Before - Logs into the CRM, GA4, the accounting system and Asana in the morning and retypes the numbers into one file - Calls finance for the state of unpaid invoices, having no access there - The board gets the report at noon, or not at all in a week when something more urgent came up - The first half hour of the meeting goes on deciding whether the numbers are current and whose version is true ### After - The report is in the inbox when the laptop opens, assembled over the weekend from fresh data - Overdue invoices are in the report, from the same system, every week - The board gets the report at the same hour even in a week when everyone is firefighting - The meeting starts with the decisions list rather than the question how did we do The first automatic report arrives 2 to 6 weeks after the workshop. We review the metric list and the targets after 30 days, and after 90 we cut the report down to what is actually read. For the person signing this off ## What the CEO or the operations director gets out of it The Monday question about how the week went stops requiring an investigation, and you stop being the person who has to commission it and then chase it. The numbers arrive on their own, in the same format, so comparing week to week takes a minute rather than a meeting. The second thing matters more in a conversation with the board or an investor. You have a weekly record of results with the variance against plan, the same one for months, showing a trend rather than an impression. When a metric falls for the third week running, you know in the third week, not at the end of the quarter. - Metrics written down with a definition and a source, independent of one person - A report at a fixed hour even in the week when nobody had time for it - A test on historical data, so the numbers agree with what you have been getting so far - Changes to the report made by your own team, with no ticket to us What it costs ## The workshop has a published price, the rollout is quoted after it The process workshop costs **EUR 3,500 net per week**. For a weekly report one week is usually enough, because the process is narrow: from the data sources to one document. We quote the rollout after the workshop, once it is clear how many sources are involved, whether any of them needs a custom integration and what state the data is in. You pay the cost of running the workflow directly to the tool vendor, meaning n8n or Make, and to the language model that writes the commentary, or you pay nothing at all when n8n runs on your own infrastructure. Model queries cost the vendor EUR 0.005 to 0.07 each. No commission from us on any of it. - ### Process workshop EUR 3,500 net per week, on a weekly report usually one week A self-contained piece of work. It ends with a list of metrics, each with a definition, a source and a target, plus a plan you can execute without us. - Map of the process today: who collects, from where, how long it takes, who reads it - Metric list with definition, source and target, minus the ones nobody reads - Tool recommendation: n8n, Make or an integration over the API, plus the output format - The metrics to measure after 30 days Start with the workshop - Most popular ### Workshop and report rollout Quoted after the workshop scope follows from the number of sources and the state of the data The option most companies choose. The first automatic report in the inbox 2 to 6 weeks after the workshop, checked against your own data from the last few weeks. - Everything in the workshop - A workflow in n8n or Make wired into the CRM, GA4, the accounting system and the project tool - A report template with the variance against plan and a commentary from a language model - Test on history, a parallel run, error monitoring and alerts - Training for the person responsible on changing the report alone Talk about a rollout - ### Report and an AI agent for the board Quoted after the workshop for companies that ask about numbers more often than once a week Everything in the rollout plus BOND, which answers questions about results between Mondays in a plain sentence. - Everything in the report rollout - An AI agent wired into the same sources: pipeline, invoices, traffic, tasks - An answer in 5 to 10 seconds, on Slack, in Teams or in the browser - Agent rollout takes around 2 months, after the report is live Ask about the agent The workshop ends with a list of metrics, each with a definition, a source and a target, which you can hand to your own team or to another supplier. If it turns out the report is two numbers you already have in the CRM, we will say so and show you how to set up a notification without a rollout. ## Frequently asked questions What is an automatic weekly report? It is a workflow in n8n or Make that, at an agreed moment, usually Friday evening or Saturday morning, queries the CRM, Google Analytics 4, the accounting system and the project tool, drops the numbers into one template, works out the variance against plan and sends the finished document by email or to Slack on Monday morning. A person reads conclusions instead of collecting data. How long does an automatic weekly report take to implement? One week of workshop and 2 to 6 weeks of build. The time depends on the number of sources and on whether the data inside them is in order. With two sources that have ready connectors, HubSpot and GA4 for example, you are closer to the lower end. An older ERP with no API, or data that has to be tidied up first, pushes the date towards the upper end. Which systems can be connected to the report? Any system with an API: HubSpot, Pipedrive, Salesforce and LiveSpace on the sales side, Google Analytics 4 on the traffic side, Fakturownia, iFirma, Comarch Optima, Subiekt, sevDesk and Xero on the finance side, Asana, ClickUp and Jira on the project side. Plus Google Sheets, BigQuery and SQL databases when some of the numbers already sit somewhere. For a system with no ready connector we write an integration over the API. What happens when one system fails to return data overnight? The workflow has monitoring and alerts. A failed pull, caused by an expired CRM access token for example, sends a notification to the workflow owner, and the report marks the block as missing instead of showing a zero. The board does not receive an empty or half filled report with no warning attached to it. Will the report replace our Looker Studio or Power BI dashboard? No, and that is not the goal. A dashboard is for digging into data when somebody has a question. The report brings a fixed set of numbers with variances and a decisions list straight to the inbox, with nothing to open. Both can read from the same sources, and the report often links to the dashboard next to a metric that is off plan. Who writes the commentary on the numbers in the report? A language model over this week data, following the instruction agreed in the workshop: what is off target by more than the agreed threshold, what needs a decision, what is noise. The commentary runs to three to five sentences and sits above the table. The model does not change the numbers, it describes them, and every figure in the commentary points to a block further down. Is our financial data safe inside a workflow like this? The workflow pulls only the fields the metrics need, and that list is agreed in the workshop. Access to your systems stays on your side, on technical accounts with limited permissions. n8n can run on your own server, in which case the data never leaves the company. What reaches the language model is aggregated numbers, not individual invoices or customer records. What if the board wants to change the metrics after a month? That is normal and it is planned: after 30 days we go through the list together. A metric from a source that is already connected can be added by the person responsible on your side, after the training. A new source means an extra connector, which we quote separately. More often metrics leave the report, because it turns out nobody reads them. ## Let us count how many Mondays a year go on collecting data at your company The free consultation takes 30 minutes. You tell us which systems you use, who assembles the report today and how long it takes them. We show you which numbers can be pulled straight away, which ones need the data tidied up first, and what your report would look like on sample data. You do not have to give access to any system at this stage. We run at most two rollouts at a time, so that someone is actually sitting on each of them. [Book a free consultation](https://justautomate.ai/en/contact/) **PS.** If somebody at your company opened five browser tabs last Monday to answer the question how did we do, that is exactly this problem. A report that arrives on its own, at the same hour, with the variance against plan, turns a meeting about collecting data into a meeting about decisions. The workshop starts by writing the metrics down. ## Related - [Automatic sales report](https://justautomate.ai/en/automations/automatic-sales-report/) - [Automatic financial report](https://justautomate.ai/en/automations/automatic-financial-report/) - [n8n automation](https://justautomate.ai/en/services/n8n-automation/) - [BOND, business AI agent](https://justautomate.ai/en/services/bond-business-ai-agent/) - [Automation ROI](https://justautomate.ai/en/guides/automation-roi/) - [Process workshop](https://justautomate.ai/en/services/process-workshop/) --- **JustAutomate** · contact: contact@justautomate.ai · form: https://justautomate.ai/en/contact/