Thank you!
We'll get back to you soon.
EN
UA
RU
ENG
UA
RU
ENG
Bots handle the routine.People make the decisions.
Most of our clients are eCommerce businesses. That means thousands, often millions, of products, feeds that update several times a day, and budgets spread across hundreds of campaigns.
Checking all of that by hand is close to impossible, so we write our own tools. They handle the routine work and the checks, and leave the team time for strategy, analysis, and new tests.
12MMerchant Center products checked every hour
16,000+Ad groups matched against the website every day
1 min 45 secFor the system to spot a problem in an account
60 hrsA month freed up for the team on one project
[ Our rule ]Our team is responsible for strategy, client communication, and performance. Automation takes the checks and the thousands of repetitive clicks off their plate.
[ 01 ]
Disapproved product monitoring in Merchant Center
Google doesn’t always tell you right away when individual products drop out of Shopping. Often you only find out once sales dip.
Our system checks Merchant Center every hour, sends you a Telegram alert when products get disapproved, and shows how many sales those products used to bring in.
Connect your store →How it works
Checks every hour
24 automatic checks a day, with no manual work needed. Each takes under a minute, even across millions of products.
Shows what’s at stake
Matches disapproved products against 90 days of Google Ads conversions.
Always current
Each check is compared with the previous hour, so the chat doesn’t fill up with the same thousands of products. People read the alerts instead of scrolling past them.
Catches a broken feed
If the product count in the account drops by 20% or more, you get a separate alert, even at night. This usually means something has gone wrong with the product feed.
Results
~12MProducts monitored across five clients
Days → hoursResponse time to disapprovals for an online store with 100,000+ products
$7,000Saved for a client by catching dropped products in time
Sample alert
[shema] BOT
🚨 New disapproved products!
📦 The Merchant Center check found new issues.
🔢 Disapproved now: 4,843
📈 New in the last hour: 1,639
💰 Sales from these products: 1,136
📄 Open the sheet
[ 02 ]
Automated search campaign launches
Big retailers usually lean on Shopping and Performance Max, because running search campaigns by hand across a huge catalog is close to impossible. And the site changes every day: brands and categories come and go, sections move around.
Our system builds campaigns, keywords, and ads from your feed and checks the ads against the site’s inventory every day, Dynamic Search Ads (DSA) included. We write the scenarios and approve the changes.
Let’s talk →How it works
Scenarios from the feed
For example, thousands of parts × thousands of car makes and models turn into campaigns with keywords and ads.
Checks the site against the ads every day
Pulls fresh feeds from the client’s site and compares what should be live with what’s actually running in Google Ads. It launches new campaigns and pauses outdated ones.
A person approves in Telegram
The system prepares a list of changes with a reason on every line, like “not in feed.” Nothing changes until someone taps Approve, and it never touches bids or budgets.
Prices and availability in ads
Updated automatically several times a day, with no manual edits.
No dead-end pages
A crawler removes pages with no products in stock from your ads. When products are back, the page goes back into the ads on its own.
A new account when the old one fills up
Once campaigns no longer fit within Google Ads limits, the system creates the next account on its own.
Results
+20%Average profit lift from feed-based search for an auto parts store
50,000Routes launched in 3 business days for an online bus ticket service, with prices and availability updated twice a day
3,722Operations in a single sync run, about a week of manual work
1,126Ad groups fixed so they could serve ads again
Sample alert
[shema] BOT
📊 DSA audit — online store
📋 New suggestions: 4
🆕 Create: 0
▶️ Resume: 4
⏸ Pause: 0
👉 Approve or reject the changes below
📄 Open the sheet 12:01
✅ Approve all❌ Reject
[ 03 ]
AI anomaly detection
Built in-house. The system forecasts your Google Ads metrics (Bing Ads can be connected too) and compares the forecast with what actually happens. If a metric differs from the forecast by more than 10%, the system sends a chart to Telegram, Google Sheets, WhatsApp, or another channel you use.
Set it up →Why it matters
~20 hrsHow long the average advertiser takes to notice an anomaly on their own
1 min 45 secHow long our system takes to spot a problem and start looking for the cause
Results
20x+Faster than the average advertiser at reacting to anomalies
$40,000+Average daily spend of the accounts we monitor, so every hour of downtime is expensive
[ 04 ]
Reports and bots that free up the team for analysis and growth
Part of the daily job is pulling numbers together, comparing them, and noticing when something’s off. Bots and self-updating reports handle that for us.
Let’s talk →How it works
Good Morning Report
Every morning a bot sends yesterday’s key metrics to Telegram. No logging into ad accounts to compare numbers.
Low-balance alerts
When an ad account has less than 7 days of budget left, the bot warns you ahead of time so your ads don’t stop.
PPC KPI Report
Updated every hour, it projects how many purchases you’ll have by month end, broken down the way your business needs.
Data in BigQuery
We stream data into BigQuery so decisions rest on complete, unsampled numbers.
Weekly client report
Plan vs. actual, what changed this week, and what we did. The report builds itself; a specialist reads it over and sends it.
/Our tools in the press
How our systems have saved budgets and sales, as told in the media.
MMR · Feb 8, 2024Protecting seven-figure budgets: AI anomaly detection in Google AdsRead the article (in Ukrainian) ↗
MC Today · Apr 24, 2023The chatbot that saved us at 1,000 orders an hour: how we fixed our processes and spared the company’s nerves and moneyRead the article (in Ukrainian) ↗
RetailersWhat to do with your online ads during power outages: advice from shema.team’s CTORead the article (in Ukrainian) ↗
/FAQ
What is shema.team’s anomaly detection system?
It’s a tool we built ourselves. It forecasts Google Ads or Bing Ads metrics and sends a chart to Telegram, Google Sheets, or WhatsApp when a metric strays more than 10% from the forecast. On average it spots a problem within 1 minute 45 seconds. Advertisers without a system like this take about 20 hours.
How does disapproved product monitoring work?
Every hour the system checks Merchant Center through Google’s official API, picks up new disapprovals, and matches them against 90 days of Google Ads conversions. If an issue is worth your attention, it messages you on Telegram. The history is kept in Google Sheets and BigQuery. Right now it watches about 12 million products.
Does the automation change anything without a person signing off?
Not the important ones. Those go live only after a specialist approves them, for example with a button in Telegram. Bids, budgets, and campaign settings stay in human hands. The system never deletes ad groups; it only pauses them.
Who is this for?
Online stores and projects with large catalogs: thousands of products, brands, and categories, with inventory that changes daily.
Which platforms do you work with?
Google Ads, Merchant Center, Bing Ads, and Meta Ads through their APIs, plus GA4, BigQuery, Power BI, and Looker Studio. Our systems run on Google Cloud (Cloud Run), work with Google Sheets and Apps Script, and send alerts to Telegram. We also use n8n, Binotel, and AI tools such as Claude. If you need an integration with something else, let us know and we’ll look into it.
Ads run by people who love what they do
Tell us which routine eats up most of your team’s time, and we’ll tell you what can be automated.
Let’s talk