×
Workflow

How Mavlers turned manual search term reviews into a weekly AI workflow

Project Overview

Function

AI-Powered Google Ads Search Term Analysis + Automated Decisioning

Use Case

Analyse up to 100,000 search terms per account, apply per-client goal logic, and write decisions straight into the team's Google Sheets, weekly, with no manual review

Products / Services Used

  • Google Sheets + Apps Script
  • Google Ads (search term data)
  • Two-pass AI (Gemini + GPT-4o-mini)
  • OpenRouter (AI routing layer)
  • Railway (central admin server)

Every account manager was spending 2 to 4 hours per account, per week, manually reviewing Google Ads search terms for wasted spend, missed negatives, and new opportunities. With 15 to 20 accounts per manager across a 50-plus person team, only a fraction of accounts could realistically be reviewed every week, and decisions varied depending on who was doing the reviewing.

‍

We built a two-pass AI system that runs directly inside each manager's existing Google Sheet. It analyses up to 100,000 search terms per run, applies the correct goal logic for each client, CPA, ROAS, CPC or CTR, and writes negatives, opportunities and keywords to add straight into the Sheet, automatically, every week.

Performance Insights

Key results

100k

Search terms analysed per run

5-15min

Down from 2-4 hrs per account

100%

Consistent AI-driven decisions

Weekly

Every account covered, without fail

Client Objectives

The search term review bottleneck

1.
Hours lost to manual review, every week

Each account took 2 to 4 hours to review by hand, reading through search term reports line by line, every single week.

2.
Not every account got reviewed

With 15 to 20 accounts per manager, only a handful could realistically be covered each week, leaving the rest unchecked.

3.
Decisions were inconsistent across the team

Two managers looking at similar terms could reach different conclusions, with no shared logic keeping the review fair.

4.
No account carried client-specific goals

CPA, ROAS and CPC clients were reviewed the same way, with no logic built in to reflect what "good" actually meant per account.

5.
No audit trail or history

Once a decision was made, there was no record of what was reviewed, when, or why, making it hard to track what had changed.

Mavlers Strategy

How the system works

Phase 1 - Core AI analysis

1.
Two-pass AI classification

Every search term is analysed by two AI models in sequence, the first classifies it into one of 12 decision types, the second cross-checks uncertain results before anything is finalised.

2.
Runs inside the existing Google Sheet

Each manager opens their client's Sheet, confirms the goal settings, and clicks Run. No separate login, export or new tool to learn.

3.
Results written back automatically

Negatives, keywords to add, opportunities and items for review are written into separate tabs, with a summary dashboard and history updated on every run.


Phase 2 - Built to scale past 50 users

1.
Moved off a shared server

A single shared office server crashed under the load of 50 concurrent users, so the system was rebuilt to run on each manager's own laptop, removing the single point of failure.

2.
Per-user API keys

A shared API key meant one manager hitting a rate limit blocked the whole team. Every user now runs on their own key, isolating usage.

3.
Central admin control

A lightweight auth server tracks access, daily run limits and budget per user, with a silent installer and auto-update pushing changes to every machine, no IT involvement needed.


Phase 3 - Per-client personalisation

1.
Goal-aware logic

Campaign objective, goal type (CPA, ROAS, CPC, CTR) and target value are set once per client, and the AI evaluates every term against that client's actual number, not a generic rule.

2.
Plain-English custom rules

Instructions like "always negative competitor terms" or "seasonal campaign Oct to Dec only" are written in plain English and applied automatically on every run.

3.
Different logic, same terms

A ROAS client and a CPA client seeing identical search terms now get different, and correct, recommendations, automatically.

‍

Results

What this means for account teams

Every account gets reviewed, every week

All 15 to 20 accounts per manager are covered without fail, not just the ones there was time for.

Decisions are consistent, not personal

The same logic is applied across every account and every manager, removing reviewer-to-reviewer variance.

Hours go back into strategy, not spreadsheets

1,500 to 4,000 hours are freed up across the team every week, time that now goes into account strategy instead of manual review.

The system keeps getting cheaper and more automated

Planned upgrades, direct API access, token optimisation, one-click push to Google Ads, are already in motion to cut cost and manual steps further.

Related Case Studies

Results that speak
for them selves.

Migrating 9,000+ pages across 12 languages

Successful migration of 9,000+ pages across 12+ languages, with no critical downtime.

80% CPR reduction and 3x ROI over 6 years

Multi-channel strategy driving 50%+ monthly registration growth.

35% cost reduction while scaling SEO delivery globally

Ogilvy needed to scale SEO delivery across multiple markets without expanding headcount. Mavlers stepped in as their invisible production partner.

View all case studiesView all case studies
Mavlers Agency
contact@mavlers.agency
© Copyright 2026 Mavlers Agency. All rights reserved.
Privacy PolicyResource Hub