Turn Weak Creative Insights into ROAS Wins with Gemini
Marketing teams drown in ad variants but still lack a clear view of which creative elements actually drive performance. This guide shows how to use Google’s Gemini, connected to your campaign data, to pinpoint high-impact assets, messages and formats and systematically improve ROAS. You’ll learn strategic steps and concrete workflows you can apply in your own stack.
Inhalt
The Challenge: Weak Creative Performance Insight
Marketing teams are producing more creatives than ever: multiple formats per campaign, dynamic components, localized versions, and endless test variants. Yet when it’s time to decide which creative elements actually drive clicks, conversions, or ROAS, most teams are blind. They see top-line metrics per ad or ad group, but not the deeper patterns: which headlines, images, hooks, or CTAs consistently win across channels and audiences.
Traditional analysis relies on manual exports from Google Ads, Meta, and other platforms, followed by spreadsheet gymnastics or dashboard tinkering. This is slow, brittle, and biased towards what the analyst happens to look at. With dozens or hundreds of creative variants live at any moment, it’s nearly impossible for a human to connect the dots between performance, audience, placement, and creative attributes in time to inform the next creative sprint.
The business impact is significant. Budgets keep flowing into ads with weak messaging because nobody sees the pattern in time. Creative teams optimise for aesthetics instead of measurable impact. Acquisition costs creep up, experimentation slows down, and competitors who use data-driven creative iteration steadily win higher ROAS. Over time, you’re not just wasting spend – you’re eroding your ability to learn what your market actually responds to.
The good news: this problem is painful, but it’s solvable. With modern AI models like Gemini connected to your ad and analytics stack, you can finally analyse creative performance at the component level and at the speed of your campaigns. At Reruption, we’ve seen how AI-driven insights can transform messy, underused data into clear creative guidance for marketers. In the rest of this page, you’ll find practical, non-theoretical steps to use Gemini to make your creatives smarter, your tests faster, and your budgets more effective.
Need a sparring partner for this challenge?
Let's have a no-obligation chat and brainstorm together.
Innovators at these companies trust us:
Our Assessment
A strategic assessment of the challenge and high-level tips how to tackle it.
From Reruption’s experience building AI-first marketing workflows, the teams that win with Gemini don’t just plug it into Google Ads and hope for magic. They deliberately design how Gemini interacts with campaign data, creative assets, and business goals. Because we build working AI products inside organisations—not slideware—we’ve seen where Gemini brings real lift in creative performance insight, and where process and data readiness matter more than yet another dashboard.
Anchor Gemini to Clear Creative Performance Objectives
Before you ask Gemini to analyse anything, align on what “good” looks like for your marketing organisation. Is your primary lens ROAS, qualified leads, first-time purchases, or repeat orders? Creative insights are only useful if they connect directly to these business metrics. Define a short set of KPIs and acceptable ranges (e.g. target CPA, minimum conversion rate per audience) and make those the frame for Gemini’s analysis.
Strategically, this means treating Gemini as a decision-support tool rather than a reporting gadget. In prompts and workflows, emphasise trade-offs: for instance, that it should prioritise creatives that keep CAC under a certain level even if CTR is lower. This prevents the model from optimising for vanity metrics and aligns insights with the economics of your channels.
Design Data Flows Before Dashboards
Gemini can only surface meaningful creative performance patterns if it has structured access to the right data: creative attributes (copy, colour, layout, offer), campaign structure, audiences, and outcome metrics. Many teams jump straight into BI tools and visualisations without first deciding how these elements will be tagged and joined. That leads to “nice looking” insight that can’t be actioned at the creative component level.
From a strategic perspective, invest in a minimal but robust data model for your ads: how you name assets, how you tag variations, and how you track experiments across platforms. Connected with Looker and Google Ads, Gemini can then reason over consistent attributes instead of trying to infer patterns from inconsistent naming conventions. This upfront discipline is a multiplier on every AI insight that follows.
Prepare Your Creative and Performance Teams to Work Together
AI-powered creative insight with Gemini reshapes how performance marketers and creatives collaborate. If one group owns data and the other owns messaging, insights will stall in the middle. Treat this as a change management topic: decide who owns the prompts, who interprets Gemini’s recommendations, and how suggestions are turned into new assets or tests.
Strategically, build a shared ritual: for example, a weekly “AI creative review” where both teams look at Gemini’s summaries of winning elements and underperformers, then commit to specific adjustments in the next sprint. This mindset shift—from opinion-based feedback to evidence-based creative iteration—is where a large share of the value is created.
Use Gemini to Reduce Risk, Not Replace Human Judgment
Gemini is powerful at summarising patterns across thousands of data points and proposing optimisation ideas, but it does not understand your brand risk, regulatory constraints, or internal politics by default. If you let it drive creative direction without guardrails, you risk on-brand issues, compliance problems, or overfitting to short-term performance quirks.
Think of Gemini as a way to de-risk creative decisions: it can tell you which elements look promising or weak, and simulate how changes might impact key metrics. Humans still decide which ideas are acceptable and how far to stretch the brand. In your strategic setup, explicitly encode these boundaries into your prompts and workflows (e.g. forbidden claims, sensitive topics, tone of voice rules) so that Gemini narrows down options without crossing red lines.
Plan for Iterative Adoption Instead of a One-Off Project
Unlocking strong creative performance insight with Gemini is not a “launch once and forget” initiative. As your campaigns, offers, and markets change, so do the patterns that matter. A static ruleset or a single integration will decay in value. You need an operating model that treats AI-driven analysis as a living capability.
Strategically, that means starting with a focused use case—such as identifying winning headlines for one priority campaign—and then extending Gemini’s role over time to images, formats, and channels. Incorporate feedback loops: track which of Gemini’s recommendations were implemented and what happened to performance. At Reruption, we build these loops into the product from day one, so your team learns with the system instead of re-running lengthy AI projects every quarter.
Used thoughtfully, Gemini can turn scattered campaign data into clear, actionable creative insight that directly improves ROAS and reduces wasted spend. The real leverage comes when it’s wired into your naming conventions, analytics stack, and creative workflow—not just plugged into Google Ads as another report. Reruption specialises in building exactly these AI-first capabilities inside organisations, from robust data flows to tailored Gemini prompts and automations; if you want to see what this could look like for your own marketing team, we’re happy to explore it with you without the usual consulting overhead.
Need help implementing these ideas?
Feel free to reach out to us with no obligation.
Real-World Case Studies
From Automotive to Banking: Learn how companies successfully use Gemini.
Best Practices
Successful implementations follow proven patterns. Have a look at our tactical advice to get started.
Connect Gemini to a Clean Creative Performance Dataset
The foundation of useful creative performance insight is a reliable dataset. Start by consolidating data from Google Ads, your analytics platform, and—if available—Looker views that already join campaigns, ad groups, ads, and conversions. Ensure that each creative asset (headline, description, image, video) can be linked to performance metrics like CTR, conversion rate, CPA, and ROAS.
In practice, work with your analytics or data team to expose a view that includes: campaign name, ad group, asset IDs, asset text, asset type, impressions, clicks, conversions, cost, revenue, and audience segments. Once this exists, configure Gemini (via the Google Cloud console or within Looker with a Gemini integration) to query this view. The goal is that you can paste or reference sample rows and ask Gemini to analyse patterns without manual copy-paste from multiple tools.
Use Gemini to Identify Winning and Losing Creative Patterns
Once Gemini can see structured performance data, you can start asking it to surface high-impact creative elements. Rather than merely listing top-performing ads, have Gemini cluster creatives by attributes—such as value proposition, offer type, tone, or imagery—and compare performance across those clusters. This moves you from ad-level to insight-level analysis.
Here is an example prompt you might use inside a notebook, Looker integration, or Gemini chat window after pasting a sample table export:
System role / instructions:
You are a marketing performance analyst focused on creative insights.
You analyse large ad performance tables and extract patterns that can be used
by copywriters and designers.
User prompt:
Here is a table of ad performance data. Columns include:
- campaign, ad_group
- asset_id, asset_type (headline, description, image), asset_text
- audience_segment, device, placement
- impressions, clicks, conversions, cost, revenue
Tasks:
1. Group creatives by similar messaging themes and value propositions.
2. For each group, calculate average CTR, conversion rate, and ROAS.
3. Identify 3-5 creative patterns that consistently outperform the baseline.
4. Identify 3-5 patterns that underperform and should be phased out.
5. Provide concrete guidance for new creatives we should test next week.
Return the result as:
- A short executive summary
- A table of winning patterns with metrics
- A table of losing patterns with metrics
- Actionable recommendations for copy/design.Expected outcome: You get an immediately usable summary of which messages and visuals work, plus specific directions for the next creative sprint, without manually sifting through hundreds of rows.
Turn Gemini Insights into Structured Creative Briefs
Insights are only valuable if they change what creatives produce. Use Gemini to convert raw performance analysis into clear, structured briefs for your content and design teams. This reduces interpretation overhead and speeds up iteration cycles.
After you’ve generated pattern insights, follow up with a prompt like:
Using the winning and losing patterns you identified above, create
3 structured creative briefs for our next ad batch.
Constraints:
- Brand tone: factual, confident, not hypey
- Target: performance marketers in B2B SaaS
- Channels: Google Search + YouTube
For each brief, include:
- Objective and primary KPI
- Target audience description
- Key message and supporting points
- Examples of 3 headlines and 2 descriptions
- Suggested visuals or video concepts
- A/B test ideas: which element to vary first (headline, CTA, offer)This way, Gemini becomes a bridge between analytics and creative production, delivering ready-to-implement briefs instead of abstract insights.
Automate Weekly Creative Performance Reviews with Gemini
To keep insights fresh, embed Gemini into a recurring review ritual. Set up a scheduled export or view in Looker that filters the last 7–14 days of data for your key campaigns. Each week, feed this into Gemini and generate a standardised performance report that your team can review in 30 minutes.
A typical workflow:
- Step 1: Data team maintains a Looker Explore/View called ad_creative_performance_last_14_days.
- Step 2: Marketing ops exports or queries this weekly and pastes the result into a Gemini-powered notebook or chat.
- Step 3: Use a saved prompt template (like the one above) to generate pattern insights and a concise executive summary.
- Step 4: Publish the summary and recommended actions in your project management tool (e.g. Jira, Asana) as tickets.
Over time, you can further automate this by integrating Gemini via API and pushing its summaries directly into Slack or email, making AI-driven creative review a standard part of your operating cadence.
Use Gemini to Simulate Impact Before Reallocating Budget
Before shifting significant budget based on creative insights, use Gemini to run scenario analyses. While it doesn’t “predict” with perfect accuracy, it can extrapolate from existing patterns to estimate potential impact of scaling certain creatives or cutting others. This helps performance leads justify decisions and manage risk.
Example prompt:
Based on the performance data above and the patterns you found,
simulate the impact of the following changes over the next 30 days:
1. Increase spend by 30% on the top 3 winning creative patterns.
2. Pause the bottom 20% of creatives by ROAS.
3. Introduce 5 new creatives following your recommended brief.
For each scenario, estimate the change in:
- Total conversions
- Average CPA
- Overall ROAS
Clearly call out assumptions and confidence levels, and flag any
risks (e.g. saturation of audiences, small sample sizes).This does not replace rigorous experimentation, but it gives decision-makers a structured, AI-assisted view of potential outcomes before implementing major budget shifts.
Enforce Brand and Compliance Guardrails in Your Prompts
As you automate more of your creative analysis and generation with Gemini, bake brand and compliance constraints directly into your configuration and prompts. Define forbidden claims, sensitive topics, tone-of-voice rules, and any industry-specific restrictions so Gemini’s suggestions stay on-brand and compliant from the start.
For example, add a persistent system instruction such as:
You are an assistant that generates marketing insights and creative ideas
for <Company>.
Hard constraints:
- Do not make promises about guaranteed results.
- Avoid superlatives like "best ever" or "number 1".
- Comply with <industry> advertising standards.
- Keep tone: professional, clear, confident, no hype.
If a requested idea would violate these rules, propose an alternative
that stays within the constraints.This reduces manual policing of Gemini’s outputs and makes your AI workflows safer to use at scale across teams.
Implemented together, these practices typically lead to faster creative iteration cycles, clearer understanding of what actually drives performance, and more disciplined budget allocation. Teams that adopt AI-powered creative performance analysis with Gemini can realistically expect leaner manual analysis (often 30–50% less time in spreadsheets and dashboards) and measurable ROAS improvements as they consistently scale winning patterns and cut losing ones.
Need implementation expertise now?
Let's talk about your ideas!
Frequently Asked Questions
Gemini improves creative insight by analysing large volumes of campaign and asset data and surfacing patterns a human would struggle to see in time. Instead of just telling you which ad has the highest CTR, it can connect specific elements—like headline themes, offers, or image types—to outcomes such as conversions, CPA, or ROAS.
When connected to Google Ads and Looker, Gemini can group creatives by messaging, audience, or format, compare their performance, and translate that into clear recommendations for new assets, A/B tests, and budget shifts. The result is a direct line from data to creative decisions, not just more dashboards.
You don’t need a large data science team, but you do need a few key capabilities. First, someone who understands your marketing data structure (Google Ads, analytics, Looker) and can help expose clean views that link creatives to performance metrics. Second, a marketing or growth lead who can define the KPIs and interpret Gemini’s insights in a business context.
On the technical side, basic skills in working with Google Cloud, Looker, or data exports are helpful. On the marketing side, your team should be comfortable with experimentation and evidence-based creative iteration. Reruption typically works directly with marketing and analytics teams to set up the initial workflows so they can run them independently afterwards.
The initial setup—connecting data, defining prompts, and running the first analyses—can often be done within a few weeks if the necessary data access is available. Many teams see useful creative insights from Gemini after the first 1–2 cycles of analysis, especially on high-traffic campaigns where patterns emerge quickly.
Meaningful performance impact, such as improved ROAS or lower CPA, typically appears after several iteration loops—usually 4–8 weeks of consistently applying Gemini’s recommendations to new creatives and budget decisions. The key is to embed Gemini into a recurring process (e.g. weekly creative reviews) rather than treating it as a one-off report.
For most teams running multi-channel campaigns with many creative variants, Gemini is cost-effective because it reduces manual analysis time and helps cut wasted ad spend. Time savings come from automating tasks like aggregating performance data, clustering creatives by theme, and drafting recommendations, which often consume hours per week per analyst.
On the ROI side, the main benefit is better allocation of budget towards proven creative patterns and faster retirement of underperformers. Even a modest uplift in ROAS or a small reduction in CPA on large budgets can easily outweigh the cost of running Gemini and the initial implementation effort. The exact economics depend on your spend level and campaign complexity, which we usually assess upfront before building anything.
Reruption supports you end-to-end, from idea to working solution. With our AI PoC offering (9.900€), we can quickly validate whether Gemini can reliably connect your creative assets to performance data and produce actionable insight in your specific setup. This includes use-case scoping, feasibility checks, a working prototype, and clear performance metrics.
Beyond the PoC, our Co-Preneur approach means we embed with your marketing and analytics teams to design the data flows, build the Gemini workflows, and integrate them into your existing tools and rituals. We don’t just hand over a concept; we help you ship a live capability, define how your team uses it week-to-week, and create an implementation roadmap for scaling it across channels and markets.
Contact Us!
Contact Directly
Philipp M. W. Hoffmann
Founder & Partner
Address
Reruption GmbH
Falkertstraße 2
70176 Stuttgart
Contact
Phone