06 — AI & Automation
The reporting and admin that eats your week, handled
Automated reporting, feed management, alerting and internal tooling. Practical plumbing rather than a strategy deck about artificial intelligence, and only where the time saved is worth more than the thing costs to build.
Written audit back in 48 hours. No obligation, no pitch deck.
01
What's included
The work itself.
Most engagements use some of this, not all of it. The audit establishes which parts are worth paying for in your case.
Reporting
- Looker Studio dashboards
- Scheduled exports
- Multi-source blending
- Board reporting
Feeds
- Feed rules
- Title automation
- Custom labels
- Stock and margin syncing
- Supplemental feeds
Alerting
- Spend anomalies
- Feed disapprovals
- Stock-outs still bidding
- Tracking failures
- Site downtime
Applied AI
- Product description drafting
- Bulk categorisation
- Review summarisation
- Support-macro drafting
02
Our approach
How this work is done here.
The order matters. Most of what goes wrong in this discipline is a later step being taken before an earlier one.
- 01
Time the task before automating it
How long it takes, how often, and who does it. A job costing two hours a month rarely justifies a build that takes twenty. This step regularly ends with us recommending you leave something alone.
- 02
Automate the reporting first
It is the most repeated task in most teams and the least contentious to replace. One dashboard, refreshed automatically, usually removes a recurring half-day of copying figures between spreadsheets.
- 03
Alert on what fails quietly
Disapproved products, out-of-stock lines still taking spend, a tracking tag removed by a theme update. These cost money for weeks precisely because nothing announces them, and an alert is cheap next to the loss.
- 04
Use AI where a human still checks it
Drafting descriptions for a large catalogue, categorising products in bulk, summarising reviews into objections worth answering. Generated text goes out reviewed, never straight to a live product page.
03
What you get, and when
What lands, and how often.
Reporting you can read without a translator, on a schedule you can hold us to.
A written audit in 48 hours
Before anything is agreed. What we found, what we would change first, and what we would expect it to be worth. You keep it whether or not you hire us.
A weekly note
Plain English, sent every week: what we did, what it did, what is next. Written by the specialist doing the work, not summarised by an account manager.
A monthly review
A longer session with the numbers behind it, including the things that did not work. Thirty minutes on a call, or a document if you would rather read it.
A change log you can read
Every change is dated and recorded, so when a number moves you can see what moved it. This is also what makes it possible to hand the account to someone else.
04
Indicative pricing
What it costs to start.
From
£1,000
per month
Rolling monthly. No minimum term and no exit fee.
What moves the figure
There is no fixed workflow count. The audit works out which tasks cost more in time than they would cost to automate, and the scope is whatever pays for itself on those numbers.
When this is the wrong fit
This is not a machine-learning product and we are not going to claim an algorithm runs your account. It is reporting, feed logic, alerting and scripting, plus language models used for the drafting work they are genuinely good at.
05
Tooling
What we work in.
Your accounts, your logins, your data. We bring the access we need and leave it behind when we are done.
- Looker Studio
- Google Sheets
- Google Ads scripts
- GA4
- Merchant Center
- Klaviyo
- Shopify
- WooCommerce
06
Questions
The things people ask before they enquire.
Answered here rather than held back for a sales call.
Is this actually AI, or is it scripts?
Mostly scripts, scheduled queries and dashboards — and those solve the majority of what wastes your week. Language models are used where they fit, chiefly drafting and classification at a scale a person would find tedious. We will always tell you which one you are paying for.
Who owns what you build?
You do. It runs in your accounts, your Google Sheets, your Looker Studio, and you get the documentation with it. Nothing is hosted somewhere only we can reach.
What happens when something breaks?
Automations break when the thing underneath them changes — a renamed column, a revised API. Maintenance is part of the monthly fee while we work together, and the documentation is written so someone else could take it on if we stop.
Will AI write our product copy?
It can draft it, at a scale that makes a large catalogue feasible. Someone who knows the products reviews every line before it publishes, because generated copy invents specifications with complete confidence.
What happens to our data?
It stays in your accounts. Where a language model is involved we use business tiers that exclude your data from training, and we do not put customer personal data through one. If a task would require that, we will say so and propose another route.
How many automations does the fee cover?
There is no fixed count. The audit works out which of your recurring tasks cost more in time than they would cost to automate, and the scope is whatever pays for itself on those numbers. Bring the task and how long it takes; if the sums do not work we will say so, and if an off-the-shelf tool already does it we will point you at it rather than build one.
Find out what is actually wrong.
Send us the URL and tell us what is not working. You get a written audit back within 48 hours: what we found, what we would fix first, and what we would expect it to do. If it turns out you do not need us, we will tell you that too.