AI automation for marketing agencies and a client report that builds itself
An agency finds out what it earned on a project once the project is over. The monthly client report is assembled by hand from five panels, and everything agreed with the client sits in email and Slack. We pull campaign data and tracked hours into one place, add an alert when the hours budget is burning, and an AI agent that answers your team from your own procedures.
- 14 daysto the first working automation, counted from the end of the workshop
- EUR 3,500 netper week of workshop, a published price we do not negotiate
- 5.55 pointshow much margin rose year over year in our founders' own agency after the AI agent went in. The caveats on that number are further down this page
The problem
Why does project profitability only show up after the fact?
Team hours sit in a spreadsheet or in Asana, rates sit in the contract, media cost sits in Google Ads and Meta Ads, invoices sit in the billing system. Each of those places tells the truth about its own slice and none of them tells the whole one. Project profitability only exists once somebody puts them side by side, which happens after the month closes. Burnt hours cannot be taken back at that point.
The monthly client report is the same illness, only visible from the outside. A specialist exports the campaigns, adds GA4 figures, pastes it into Looker Studio or into slides, formats it to that particular client's template and writes the conclusions. The work the client is actually paying for starts at the last sentence.
You already track hours somewhere, because payroll and client invoices depend on it. So the problem is not that nobody counts hours. The problem is that counted hours and the project budget never meet in the same place in the same week, and by the time they do, the month is closed.
- A project drops below zero and you find out when you raise the invoice
- The client report is glued together by a person every month, by hand, from five panels
- Briefs and decisions get lost in email and Slack, so somebody has to reconstruct them
- A new specialist spends weeks asking the team about procedures and account history
- The owner is the bottleneck on every proposal and every quote
Why now
What changed, so this takes weeks rather than a quarter?
Ad panels, analytics and task systems now have read interfaces, so the numbers come out without an export to a spreadsheet. Google Ads, Meta Ads, GA4, Search Console, Asana and ClickUp hand over their data on request. Three years ago wiring that together was a quarter-long project. Today it is counted in working days.
The second change is the warehouse. Campaign data and the time log dropped into BigQuery stop being files somebody keeps up to date and become one source that Looker Studio draws the client report from and your profitability view from. The same measurement serves both sides. Collection runs through n8n, Make or a direct API call. We describe the individual pieces separately: the campaign report, client reports and the reporting layer under them.
The third thing is the AI agent. A rule can line the numbers up, but it will not answer an account manager asking how many hours are left on a project and whether it will hold to the end of the month. BOND, our company AI agent, answers that in a plain sentence in 5 to 10 seconds, from your data and your procedures.
What changes
What does an agency have after the first month?
The client report builds itself
Campaign data, GA4 results and team hours land in one document, in the format agreed for that specific client. The specialist checks it and writes the conclusions instead of building from scratch.
A finished draft on the first day of the month, the same measurement for every account
Profitability visible during the project, not after
Hours times rate, media cost and invoices lined up per client and per project. You can see which client actually earns, rather than which one looks good in a revenue list.
Counted daily instead of once a month, after the books close
An alert before the project goes below zero
Hours budget usage compared against plan, with a notification to the account manager at the thresholds agreed in the workshop. A reaction is possible, because the message arrives while the project is running.
Thresholds at 70 and 90 percent of the hours budget, notification in Slack
The team asks the agent, not the owner
Procedures, quotes and the history of work with a client in one knowledge base the AI agent answers from. A new hire stops blocking two other people just to learn what the report standard looks like.
An answer from company data in 5 to 10 seconds, in English or in Polish
A form enquiry reaches the CRM already qualified
The enquiry funnel goes into HubSpot or Pipedrive with the segment, the service scope, the budget and the source. Nobody retypes it and nothing dies in an inbox after hours.
A CRM record with every field filled, seconds after the form is sent
A draft quote instead of a blank page
The AI agent assembles a draft quote from previous proposals of similar scope and hands it to a person for approval. The owner still signs the proposal off, they just do not start it at ten in the evening.
A draft built from previous quotes, always approved by a person
Proof
Do you work this way in your own agency?
Yes. One of our founders runs a performance marketing agency, and the AI agent inside it was built with the method described above. It is the only proof we can show without asking a client for permission to publish their numbers. Below is half a year of measurement, together with what those numbers do 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 Google Ads, Meta, the CRM, the task system, the SEO tool, Analytics, Search Console and the data warehouse. First commit 1 March 2026, 23 people use it today.
8 of 38 systems it works in
Google Ads
Meta Ads- HubSpot
- ActiveCollab
- Ahrefs
GA4
Search Console
BigQuery
- Before
- In the twelve months before launch the agency ran on a margin of minus 4.42 percent. Across a twenty month window there was not a single run of three positive months.
- After
- May, June and July 2026 were three positive months in a row, the first such run in the whole window. Revenue barely moved, so the change sits on the cost of delivery, not on sales.
Margin up 5.55 percentage points year over year, on revenue up 1.5 percent
Operations run by the agency agent in consecutive months of 2026, from the first commit in March. From 83 operations in March to almost 63,000 in August, with users growing from 4 to 23. This curve is here because a rollout nobody uses looks exactly like a successful one in a report.
Measurement: Agent invocation log in the data warehouse, March to August 2026, counted at the level of a single tool call.
The split of 29,854 operations run with no person involved over thirty days, by type of task. Most of it goes on tracking team availability and monitoring client sites, the things that have to run without a break and that nobody wants to do by hand.
Measurement: Agent audit log, the 30 days before 2 September 2026, counted by the kind of tool invoked.
And in your agency
What if things work differently here?
Four sentences we hear in agencies more often than any others, and all four are fair.
We have Looker Studio, our reporting is already automated.
Looker Studio shows campaign data and it does that well. It cannot see team hours, rates and the cost of delivery, so project profitability is not in there. We add a second source to the same warehouse and leave your dashboards where they are.
We are counting on AI to improve our clients' campaign results.
It will not, and we say that on the consultation call before anybody pays. Automation takes operational work off the team and shows profitability. Campaign results improve when a specialist gets the time and the data to work on them, not when a script runs.
Our time log lives in a spreadsheet and it is kept unevenly.
It nearly always is, and that is the first thing the workshop fixes. We agree what exactly has to be logged for the number to mean something for profitability, and only then wire it up. Automating an unreliable log gives you an unreliable result, just faster.
We will not hand our clients' data to an outside company.
We work on your infrastructure or on a cloud you nominate, and the scope of access is agreed in the workshop and written down. Personal data from client accounts is processed under a data processing agreement, limited to what a given automation needs.
- CRM
- HubSpot, Pipedrive, Salesforce, Zoho
- Advertising and analytics
- Google Ads, Meta Ads, LinkedIn Ads, TikTok Ads, GA4, Search Console
- Data and reporting
- BigQuery, Looker Studio, Google Sheets, SQL databases
- Teamwork
- Asana, ClickUp, Jira, Slack, Notion, Google Workspace
- Time tracking and invoicing
- Harvest, Toggl, time log spreadsheets, Xero, Fakturownia
- Automation layer
- Make, n8n, custom API integrations
What exactly we automate in an agency
A typical split of agency work before a rollout and after it. The order is agreed in the workshop, because a different line hurts most in every agency.
| Area | Today | After the rollout |
|---|---|---|
| Monthly client report | glued together by hand from five panels | assembled from campaign data and tracked hours, ready to comment on |
| Project profitability | known after the month closes | hours times rate against budget, counted daily |
| Hours budget | watched from the account manager's memory | an alert at 70 and 90 percent of usage |
| Knowledge about a client | email, Slack, the drive and one person's head | one knowledge base the AI agent answers from |
| Form enquiry | retyped into the CRM by hand | a CRM record with qualification and source, in seconds |
| Proposal and quote | written from scratch by the owner | a draft from previous quotes, approved by a person |
| Onboarding a specialist | weeks of questions to the team | procedures and account history in one place, on demand |
The scope of the first rollout is agreed in the workshop. We take the three lines with the highest return, not all of them at once.
How we work
Five steps from five panels to one view
Every step ends with something you can judge before you decide on the next one.
Workshop
One week. We walk a job from the form enquiry to the invoice and the hours settlement, and check where the same number gets produced a second time. We come out with a process map and a list of the places where margin leaks unnoticed.
Picking three automations
From the map we pick the three with the highest return and measure them before the start, so that after thirty days it is clear whether they worked. In agencies it usually comes out as the monthly report, the hours budget and the enquiry funnel.
Pilot on your data
The first automation is built within 14 days of the end of the workshop and runs alongside the current process, with nothing switched off. We test it on your campaigns and your time log from recent months, not on a sample.
Production and training
Once verified we switch it to production with error monitoring. Someone on your side learns to change the thresholds, the report format and the field list on their own, without raising a ticket with us. A rollout usually closes in 2 to 4 weeks.
Measurement after 30 and 90 days
We check the indicators agreed before the start: how many hours go on reporting, how many projects went over budget and how many questions the team put to the agent. Only then do we decide on the next area.
Before and after
How an account manager's month changes
Before
- The first week of the month goes on client reports
- You learn the hours budget is burnt when the invoice goes out
- A new specialist spends weeks asking about procedures and account history
- Every proposal waits for the owner, because only they know the previous quotes
After
- The report is waiting on day one and the team adds the conclusions to it
- An alert at 70 percent of the hours budget arrives while the project runs
- A new specialist asks the AI agent and gets an answer in 5 to 10 seconds
- A draft quote waits for approval, not for the owner's free afternoon
The number of clients you serve does not have to change. What changes is how many team hours go on running the process instead of on the work a client pays a higher rate for.
For the agency owner
What the person signing this off gets
The first thing shows up immediately: you stop being the bottleneck. Quoting, the report standard and account history move out of your head and into a knowledge base the team gets answers from without asking you. You approve instead of writing from scratch.
The second is less obvious, and it matters more in a conversation about raising rates or selling the agency. Profitability per client and per project stops being one person's knowledge and becomes a number in a system, one you can show and defend.
- Profitability per client and per project counted daily, not once a quarter
- An alert on the hours budget before a project drops below zero
- Procedures and quotes written down, independent of one person in the company
- A staged rollout: the workshop ends with a document, not a commitment
What it costs
The workshop price is published, the rollout price comes after it
The workshop costs EUR 3,500 net per week and we publish that before any call, because an agency has a right to know what it is signing up for. One or two weeks is usually enough in an agency. We quote the rollout after the workshop, once we know how many systems we connect to and what state the data in them is in.
Tool and language model cost goes to the vendor directly, which is where you already pay for Make, n8n or the warehouse, or nowhere at all when everything runs on your own infrastructure. The model cost sits in the range of EUR 0.005 to 0.07 per query depending on the model. No commission from us and no margin on somebody else's invoice.
Process workshop
EUR 3,500 net
per week, in an agency usually one to two
A self-contained piece of work. It ends with a process map and a 30 day plan you can execute without us.
- A week of work inside the agency, in the systems and in the time log
- A map of a job from enquiry to invoice
- Three automations with the highest return, measured before the start
- The indicators to measure after 30 and 90 days
- Most popular
Workshop and first rollout
Quoted after the workshop
scope follows from the number of systems and the state of the data
The option most agencies choose. The first automation runs 14 days after the workshop ends, the whole thing usually closes in 2 to 4 weeks.
- Everything in the workshop
- The client report or the hours budget alert, wired into your systems
- Connections to the ad panels, the CRM and the time log
- A pilot alongside the current process, then the switch with monitoring
- Training for the person who runs it on your side
AI agent for an agency
Quoted after the workshop
BOND, rollout around 2 months
For an agency where questions about account status, procedure and a previous quote come up a dozen times a day.
- Everything in the rollout
- An AI agent answering from your knowledge base in 5 to 10 seconds
- Questions about profitability, hours budget and account results in a plain sentence
- A draft quote built from previous proposals, approved by a person
- Access for account managers, specialists and the board
The workshop ends with a document you can defend without us: a process map, a list of automations and the indicators to measure. If after the workshop we conclude there is nothing worth automating, or that tidying up the time log is enough, we say so plainly and do not propose a rollout.
Frequently asked questions
What does automation mean in a marketing agency?
It means a system taking over the things somebody does by hand today: assembling a report from campaign data, lining hours up against rate and budget, retyping enquiries into the CRM and hunting for procedures in old email. The data for it usually already exists in the agency, it just sits in five tools at once. Automation is about bringing it into one place, not about buying another platform.
Will this improve our clients' campaign results?
No, and we do not promise it. Automation takes operational work off the team and shows project profitability, so a specialist gets the time and the data they do not have today. What they do with that depends on them, on the market and on the client's budget. Anyone promising you a lift in return on ad spend after a script goes live is selling something other than what they say.
How long does a rollout in an agency take?
One week of workshop and 14 days to the first working automation, counted from the end of it. A full rollout of the first area usually closes in 2 to 4 weeks. The company AI agent is a different scale, around 2 months, because the knowledge base it answers from has to be put in order first.
Do we have to swap Asana for ClickUp, or HubSpot for something else?
No. We read the data through the interface and leave the tools where they are. Migrating a task system or a CRM is a quarter-long project, and the first automation is meant to be running in 14 days. If the workshop shows that a tool genuinely blocks measurement we will say so, but replacing your stack is not a condition of starting.
How does the system know how many hours are left on a project?
From your time log, whether you keep it in a spreadsheet, in Asana or in ClickUp. We line the logged hours up against the rate and the project budget agreed in the contract, and show the difference every day. The quality of that number depends on how honestly the log is kept, which is why we start by agreeing what exactly has to be logged.
Can an AI agent answer our team from our own knowledge base?
Yes, that is BOND, our company AI agent. It answers from company data in 5 to 10 seconds: procedures, the report standard, previous quotes, the history of work with a specific client. The rollout takes around 2 months, because most of the effort goes into organising what today sits in email and Slack.
What happens to our clients' ad account data?
We work on your infrastructure or on a cloud you nominate, and the scope of access is agreed in the workshop and written down. Only the fields an automation actually needs are passed to it. Personal data from client accounts is processed under a data processing agreement, and we agree how long it is kept.
How much does the rollout cost after the workshop?
It depends on how many systems we connect to and whether the data in them is in order. We quote after the workshop, because that is the first point at which we know the answer, and a number pulled out of the air beforehand would not hold anyway. The workshop costs EUR 3,500 net per week and you know that price before the call.
Let us start with one process
The free consultation takes 30 minutes. We take one of your processes, usually the monthly report or the hours settlement on a project, and tell you whether we see anything there worth automating. If we do not see it, we say so straight away. 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 consultationPS. If somebody at your agency said this month that a project probably came out flat, that is exactly the problem. Profitability nobody can see while the work runs is historical knowledge. Automation will not improve your clients' campaign results, but it will mean you hear about a burnt hours budget in week two rather than at the invoice.