Automatic Slack notifications only what somebody has to act on
A new lead from a form, a deal above your value threshold, a task two days past its due date, a payment landing in the bank. A workflow in n8n or Make listens for the event in your systems and sends a message where the team already sits: with context, a link straight to the record and an action button. The alert goes to the person who has something to do, not to the whole channel.
- 28 percentof a knowledge worker's week goes on reading and answering email, and another 19 percent on hunting for information. McKinsey Global Institute, 2012
- 10 secondsfrom the event in the source system to the message on the channel, when the source sends a webhook
- 2 to 4 weeksfrom workshop to working alerts, tested on your own real events from the last few weeks
The problem
Why does the team hear about an important event three days late?
A rep does not know that a key client opened the proposal. A project manager finds the overdue task at the weekly review. The board hears about a fault from the client rather than from its own system. The events are all recorded, in the CRM, in the task system and in the shop, but nobody looks there at the moment when looking would still change something.
The notifications built into those tools do not save you. HubSpot, Jira, ClickUp and the shop each send their own alerts, by email, to everyone, with no threshold and no context. Within two weeks the team writes an inbox rule and stops reading them. From then on an event that matters looks exactly like a hundred that do not.
The cost is countable. According to the McKinsey Global Institute report "The social economy" from 2012 a knowledge worker spends 28 percent of the working week reading and answering email, and another 19 percent searching for information and asking colleagues. That is close to half a week spent moving information around, and a large part of it can move without a person.
- You find out about the overdue task at the weekly review
- The client reports the problem earlier than your own system does
- An alert that matters is lost among a hundred nobody reads any more
- Somebody retypes numbers from the CRM onto the team channel every morning
Why now
What changed, so that alerts take weeks rather than a quarter?
The systems you already run can now say out loud that something happened. HubSpot, Pipedrive, Jira, Asana, ClickUp, Stripe and WooCommerce send a webhook the moment the event occurs, and BigQuery and Google Sheets can be queried on a schedule. Nobody has to wait for a person to refresh a view inside a tool.
The second change is on the Slack side. A message no longer has to be one paragraph of text: it has a header, fields with data, a link straight to the record and action buttons, for example "Assign to me" through a Slack Workflow. The alert stops being a piece of information and becomes the place where the work happens. The n8n documentation shows the same pattern from the other side: a webhook in, a condition, a formatted message out.
The third change is the AI agent. An alert tells you something happened just now, but it will not answer the question of how many events like that you had this month. BOND, our company AI agent, answers that from your company data in 5 to 10 seconds, in a plain sentence, with no logging into five systems.
What changes
What does the team get?
The alert reaches a person, not everyone
For every event we agree the recipient: a team channel, a private channel or a direct message to the owner of the record. The rule reads the owner from the CRM, so you do not have to correct it every time the team changes.
Three levels of addressing: public channel, private channel, direct message
A message you can act on straight away
The alert carries the context of the event and a link straight to the record in the CRM or in the task system. There is an action button with it, so reacting does not mean opening three tabs.
An "Assign to me" or "Snooze until tomorrow" button through a Slack Workflow
A threshold instead of a stream
Every alert has a condition you set yourself in the workshop: amount, delay, funnel stage, client type. Events below the threshold send nothing, so the channel stays readable.
A deal above EUR 20,000 to the director, a task 2 days late to the person doing it
One place instead of five tabs
The CRM, the task system, the shop, payments and the data warehouse all speak to the team on the same channel, in one format. Nobody checks five tools to find out what is going on.
HubSpot, Jira, ClickUp, Stripe, WooCommerce and BigQuery in one place
Quiet after hours
An urgent alert goes out immediately, at any hour. The rest waits and arrives as a single digest at the start of the working day, on a calendar you agree.
A split between urgent now and one digest at 9:00
Questions about numbers in a plain sentence
How many leads came in this week. How many overdue tasks does the team hold. Which orders are stuck on payment. The AI agent answers from your data, with no report to build.
An answer from company data in 5 to 10 seconds, in English or in Polish
Proof
Do you work this way yourselves?
Yes. In our founders' agency the checks the team needs to know about are run by an AI agent, which reports the result itself. 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.
JustBottheir agentJustIdea 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 the CRM, the task system, the advertising tools, the monitoring of client websites and the data warehouse. It went live in 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
- Watching team availability and checking client websites was work someone had to do by hand every time. A person had to go in, look, and pass the result on.
- After
- Those checks now run in the background and the team gets a finished result out of them instead of raw data to read through. A person steps in where a check actually found something.
8,631 team availability checks and 4,354 reviews of client websites in 30 days, with no human involved
And in your company
What if our team works differently?
Four sentences we hear most often, and how we answer them.
We already have the Slack and HubSpot integration, and the team muted it.
That usually means it sent everything, to everyone, with no threshold. We start from the other end: we list the events somebody genuinely reacts to, and send only those. The rest stays in the system, to be reviewed on demand.
Slack is already noise for us, we do not want more messages.
The first result of the workshop is subtraction rather than addition: we switch off more existing notifications than we add new ones. An alert with no threshold and no named recipient is noise, while an alert with a threshold and a link replaces asking around on the channel.
We do not use Slack, we are on Microsoft Teams.
The mechanism is the same: a webhook from the source, a condition, a formatted message to a channel or to a person. We build it on Teams, and for events somebody has to know about outside working hours we add email or SMS. The choice of tool is made in the workshop, not up front.
We do not want CRM data circulating on a channel the whole company can see.
Only the fields needed for the decision go into the message, and you approve that list in the workshop. Events carrying client data go to a private channel or straight to the account owner, and the alert keeps a link to the record, which needs CRM permissions anyway. The personal data in the message is agreed against <a href="https://eur-lex.europa.eu/eli/reg/2016/679/oj" rel="noopener">GDPR 2016/679</a> with you.
- CRM
- HubSpot, Pipedrive, Salesforce, Zoho
- Team work
- Slack, Microsoft Teams, Jira, Asana, ClickUp, Notion, Google Workspace, Microsoft 365
- Sales and shop
- Shopify, WooCommerce, Magento, BigCommerce, Stripe, PayPal
- Finance and accounting
- Xero, QuickBooks, Sage, sevDesk, Chargebee
- Data and analytics
- BigQuery, Google Sheets, GA4, SQL databases
- Automation layer
- Make, n8n, Zapier, custom API integrations
Events, alerts and the thresholds we usually start from
Each row shows what a single rule looks like: what happened, who has to know about it and under what condition the message goes out at all. The thresholds in the third column are examples, because you set them yourself in the workshop.
| Event in the system | Who gets the alert | The threshold you set |
|---|---|---|
| A new lead from a form or an ad | the sales channel, and the owner by direct message | every lead, or only from chosen campaigns |
| A deal moves to the negotiation stage | the sales director by direct message | value above EUR 20,000 |
| A task past its due date | the person doing it and the project channel | more than 2 days late |
| A payment lands in the bank account | the finance channel | above a set amount, with no invoice in the system |
| An order with an exception | the customer service channel | out of stock, declined payment, an address outside the countries you ship to |
| A metric drops in the data warehouse | the board channel, once a day | a change of more than 20 percent week on week |
| A workflow run fails | the technical channel and the person on duty | every failure, no threshold |
A first rollout usually covers three to five rules. The rest goes in once it turns out somebody reacts to them, rather than up front.
How we work
Five steps from inbox noise to an alert somebody reacts to
Every step ends with something you can judge before you decide on the next one.
Workshop
We list the events somebody in your company has to know about, and check who notices them today and after how long. We drop the ones nobody reacts to straight away: for those an alert would only be more noise.
Rule design
For each event we agree the condition, the recipient, the channel, the wording of the message and the action button. We also agree which data may appear on the channel and what stays behind a link to the record. Then we pick the tool: n8n, Make, Zapier or an API integration.
Rollout in parallel
The alerts start alongside the current notifications, on a test channel with a few people. We run them against real events from the last few weeks, so you see the actual number of messages per day rather than the one from the plan.
Switch-over and training
Once it checks out we move the alerts to the real channels and switch off the old notifications that duplicate them. Somebody on your side learns to change a threshold and a recipient alone, with no ticket raised with us.
Measurement at 30 and 90 days
We look at two numbers: how many alerts went out and how many of them somebody acted on. A rule with zero reactions is removed or gets a higher threshold, because an alert nobody reads spoils all the others.
Before and after
How the team's day changes
Before
- Somebody goes through five tools in the morning to see what happened overnight
- Everyone finds out about the overdue task at the weekly review
- Notifications from the tools land in the inbox under an archive rule
- The question "has anyone seen this" goes round the channel several times a day
After
- Overnight events sit in a single digest at the start of the day
- A task 2 days late writes to the person doing it and to the project channel by itself
- The channel only carries events with a threshold, so the team reads them
- The alert holds a link and a button, so reacting takes one click
The first alerts are visible on switch-over day. We judge the thresholds and the list of rules after 30 days against what actually went out, and again after 90 days.
For the person signing this off
What an operations director gets out of it
You stop being the channel through which news of problems travels. An event that needs a reaction reaches the responsible person on its own, and you see the same thing they do, at the same moment. The weekly review stops being the place where the backlog is discovered.
The second thing matters more in front of the board. You know how many events of a given type happened in the month and how many were handled, because the alerts and the reactions are recorded. That is a number you do not have today, and an argument in the conversation about headcount or budget.
- Rules written down and changed with no ticket raised with us
- Alerts launched in parallel, so the old notifications hold until the tests are done
- Reactions measured at 30 and 90 days, not an impression from a conversation
- A rule with no reactions is removed, so the channel does not turn into noise
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, and runs from one to seven weeks. Slack alerts usually need one week, because the scope is narrow: the list of events, the thresholds and the recipients. We quote the rollout after the workshop, once it is clear how many source systems are in play and whether the data in them is complete.
You pay the cost of running the workflows directly to the tool provider, that is Make or Zapier, or you pay nothing at all when n8n sits on your own infrastructure. No margin from us. The same goes for model costs when an AI agent is part of the setup, where a single query costs the provider roughly EUR 0.005 to 0.07.
Process workshop
EUR 3,500 net
per week, and Slack alerts are usually one week
A self-contained piece of work. It ends with a list of events, thresholds and recipients that you could execute without us.
- The list of events somebody in your company has to know about, with thresholds and recipients
- The notifications to switch off, because they duplicate content or nobody reads them
- A tool recommendation: the native Slack integration, n8n, Make, Zapier or the API
- Rules for what data may appear on a channel, and the metrics to measure at 30 days
- Most popular
Workshop and alert rollout
Quoted after the workshop
scope follows from the number of events and source systems
The option most companies choose. Working alerts 2 to 4 weeks after the workshop, tested on your own real events.
- Everything in the workshop
- A workflow in n8n, Make or over the API, wired into the CRM, the task system and the shop
- Formatted messages with a link to the record and an action button
- A split between urgent alerts and one digest at the start of the day
- Error monitoring and training for the person who will change the thresholds
Alerts and an AI agent
Quoted after the workshop
for teams that ask about their data a dozen times a day
Everything in the rollout plus BOND, which answers questions about the state of things in a plain sentence, straight from the channel.
- Everything in the alert rollout
- An AI agent wired into the CRM, the task system and the data warehouse
- An answer from company data in 5 to 10 seconds, with no logging into systems
- Daily and weekly digests written by the agent, not by a person
If the workshop shows that the native Slack integration with your CRM plus switching off some notifications is enough, we say so and show you how to set it up. The workshop ends with a document that stands on its own, and the matter does not go on to a rollout you do not need.
Frequently asked questions
What are automatic Slack notifications?
They are a workflow that listens for events in your systems and sends a message about them to a channel or to one person. An event is, for example, a new lead, a deal above a threshold, a task past its due date or a payment landing in the bank. The message carries context, a link straight to the record and an action button, so reacting does not mean opening more tools.
How long does a Slack notification rollout take?
One week of workshop and 2 to 4 weeks of rollout, with tests on real events from the last few weeks. With more source systems, or with data that has to be tidied up first, it runs to 6 weeks. This is a single point of integration over a webhook and an API, not a data migration.
How is this different from the built-in Slack integration with HubSpot or Jira?
A native integration sends what its vendor decided it should send, usually with no threshold and no choice of recipient. If that is enough for you, we will say so in the workshop and there is no reason to build anything else. An outside workflow earns its place when the condition combines several fields, when the recipient has to be read from the CRM, or when one event joins data from two systems at once.
What if we use Microsoft Teams rather than Slack?
The mechanism is identical: a webhook from the source, a condition, a formatted message to a channel or to a person. We roll it out on Teams exactly as we do on Slack. For events somebody has to know about outside working hours we add email or SMS on top.
Will the alerts not turn the channel into noise?
That is the main risk, which is why every rule has a threshold and one named recipient, and after 30 days we look at how many alerts went out and how many somebody acted on. A rule with zero reactions gets a higher threshold or is removed. In the workshop we usually switch off more existing notifications than we add new ones.
Which CRM data ends up on the channel?
Only the fields needed for the decision, and you approve that list in the workshop. Events carrying client data go to a private channel or straight to the account owner, and the details stay behind a link to the record, which needs CRM permissions. The scope of personal data in the message is agreed with you under the GDPR.
What happens when the integration stops working?
Error monitoring is configured as standard: a failed workflow run reports itself to the technical channel and to the person on duty. Events from the period of the outage are caught up once the workflow is back, rather than lost. A silent failure of the alerts is worse than having none, because the team assumes that if nothing arrived, nothing happened.
Let us find out what your team hears about too late
The free consultation takes 30 minutes. You tell us which systems you run and which event you last heard about from a client rather than from your own tool. We show you which alerts can be set up straight away and which notifications are worth switching off first. You do not have to share any systems at this stage.
We run at most two rollouts at a time, so that someone is genuinely sitting on each of them.
Book a free consultationPS. If anyone in your team wrote "did you see this?" on a channel this week, that is exactly the problem. Information that waits for a person to look at it arrives about a day late, while an alert with a threshold and a link cuts that to a few seconds. Listing those events is where the workshop starts.