Blog
-
Modern Search: How to Upgrade Your Google Ads For AI-Driven Success
Search advertising has changed. The old playbook of hyper-segmented account structures, dozens of tiny ad groups, manual bids, rigid keyword match types, and a complex network of negtive keywords worked when the system had to be spoon-fed intent. Today, Google’s machine learning (and the data it can use) is the engine that finds real, incremental customers at scale. Advertisers who don't work to facilitate this through consolidated, data-first account architecture will find performance drifting down while competitors who empower ML improve efficiency, scale, and insight.
Ready to upgrade your Google Ads to modern search and get real results?
Below I explain the strategy, the tracking you must get right (consent mode, enhanced conversions, server-side tracking and value-based bidding), the operational benchmarks that unlock smart bidding, the organizational trade-offs, and a real-world case study showcasing results in action.
What is Modern Search?
Modern search = simpler, consolidated account structures + broad match (backed by rich 1st-party conversion data) + robust tracking + value-based bidding + letting Google’s ML combine keyword signals with audience & behavioral signals to find incremental customers.
This is not “set-and-forget” advertising — it’s an intentional architecture that gives machine learning the signals it needs to make better decisions in real time.
Why Consolidation Helps Machine Learning
- More signals per learning unit. When you consolidate ad groups and campaigns around meaningful themes (not micro-segmentation), each bidding unit collects more impressions and conversions. More signals = faster, more confident, and more stable ML decisions.
- Better use of Google’s audience and behavioral signals. Broad match keywords + strong first-party conversions allow Google to combine intent (what users type) with audience signals (past behavior, demographics, likelihood to convert). Bids and auctions become informed by people as well as keywords.
- Value-based decisions. When you feed purchase values and revenue into Google (via enhanced conversions or server-side conversion feeds), smart bidding can optimize toward business value, not just conversion volume.
- Reduced internal noise. Fewer tiny ad groups and redundant keywords reduces conflicting signals (e.g., competing bids across many similar ad groups) that confuse ML and waste budget.
If you cling to old over-segmentation and manual fiddling, you’re essentially handicapping the ML that could be finding more efficient, incremental conversions for you.
Tracking & Data: Building Data Foundations
Machine learning is only as good as the data it receives. That means investing in tracking and privacy-safe data flows:
- Consent Mode (Google): Respect user consent while still sending partial conversion signals to Google so ML can model conversions even when not all users consent to cookies.
- Enhanced Conversions: Server-side or first-party hashed conversion data (email, phone) tied to conversions that improves match rates and attribution accuracy.
- Server-side tagging / conversion importing: Reduces signal loss from browser restrictions and ad-blocking, and lets you send richer payloads (currency, order value, SKU).
- First-party data hygiene: Clean user IDs, CRM imports where appropriate, and reliable purchase values.
- Value-based conversion configuration: Tell Google the value of each conversion (not all conversions equal) so bidding can prioritize profitability and LTV.
Turning on smart bidding without this infrastructure is like putting a powerful engine in a car with flat tires: Despite its power, it won’t get you where you want to go.
Volume Benchmarks for Modern Search
To give smart bidding what it needs to succeed, adhere to these minimums:
- At least 3,000 impressions per ad group per week. This gives ML volume to understand how different creatives and queries perform. (Think of this as the minimal statistical signal for ad group-level learning.)
- At least 10 conversions per campaign per week. Smart bidding needs conversion momentum to evaluate and scale toward similar users.
- No keywords with fewer than 10 impressions per week. Low-impression keywords stifle learning and create budget leakage; prune or combine them.
- Avoid manual bid adjustments. Manual device/location/time adjustments are usually redundant — ML analyzes real-time demand across these dimensions continuously.
- Avoid multiple landing pages in the same ad group. Structure ad groups by theme and tie them to a single high-volume landing page. Reduce the total number of landing pages used in paid search; where niche pages are important, use Dynamic Search Ads (DSA) and Performance Max to capture those lower-traffic long-tail pages without fragmenting signals.
These are minimums — hitting them doesn’t guarantee success, but missing them will almost certainly limit how much ML can improve performance.
Why Broad Match + 1st-Party Data Works
Broad match has a bad reputation among old-school search pros — but combined with robust first-party conversion data, it lets Google’s ML understand semantic intent across many query variants. The advantages:
- Captures queries you didn’t anticipate. ML finds relevant searchers that would be missed by rigid match types.
- Leverages audience & behavioral signals. Bid strategies can now factor in likelihood to convert based on Google signals (not only the literal keyword text).
- Improves auction accuracy. When audience likelihood, historical conversion behaviour and real-time signals are combined, auctions produce better matches between ad and user — which improves CTR, conversion rate and ROAS.
The key caveat: only do this if you’ve solved for high-quality conversion data (enhanced conversions, server-side feeds) and have consolidated structures so the signal is concentrated.
You can still use a combination of broad match and exact match for high-volume exact match terms in the same ad group, but be careful not to overdo it! Avoid the urge to segment ad groups by match type – this will only stifle your ML opportunities and add an inefficient set of negative KWs to manage.
For negative keywords, use only terms that are irrelevant to your brand. If you’re trying to force traffic to different ad groups, you’re limiting your opportunities with AI.
Additional Benefit: Better Strategic Insights
Modern search doesn’t just improve ad performance, it creates better data for the whole business. Because ML is optimizing across many signals, Google Ads starts to surface statistically significant patterns:
- New audience insights: unexpected in-market groups that convert at higher rates.
- Real-time demand discoveries: regions with above-average conversion or value-per-conversion that you can use for OOH or local promos.
- Timing & device insights — when and how your highest-value users convert.
- Frankfurt (in Hessen): Home to Germany's financial industry, banks, and the German
Where teams used to define audiences and hypotheses externally, ML can produce fresh, data-backed recommendations you can share across channels. For Example: you run nation-wide ads and discover a particular region is converting 40% above baseline. That insight can justify a successful OOH campaign or help a brand decide where to open their next brick-and-mortar location.
Implementation and Organizational Challenges
Adopting modern search is both technical and cultural:
- Cross-team collaboration is essential. Data engineering, web development, privacy/compliance, analytics and marketing must work together to implement consent mode, server-side tagging, and enhanced conversions. A successful modern search approach expands beyond the “four walls” of Google Ads.
- Privacy and governance discussions. Compliance teams will need to approve data flows and retention; work proactively with them and be prepared to collect and share GDPR and data-hashing documentation from Google. Avoid providing any legal advice if not qualified to do so.
- Product/website changes may be required. You’ll likely need fewer landing pages for paid search; you may need server-side endpoints for conversion events. Plan to build close relationships with web design and web development teams.
- Training and upskilling for search teams. PPC specialists used to manual control must be coached to trust data-driven consolidation and understand how to set up and discuss ML-led strategies.
Don’t assume “flipping the switch” will work. Smart bidding needs time and clean inputs. But the long-term payoff of incrementality and scalable budget increases without performance collapse is worth the investment. Be prepared to work with teams across disciplines and get buy-in from multiple stakeholders.
Incrementality and Scaling:
Why modern search beats old-school optimization
Older best practices often involved optimizing with a particular budget and a particular set of keywords. That can work on a fixed slice of traffic — until you try to scale. When budgets increase, previously high-performing manual setups usually deteriorate because they weren’t designed to adapt to higher traffic volumes.
Modern search and smart bidding solve this by optimizing toward incremental value. With proper value signals and consolidated structure, ML finds additional pockets of demand and adjusts bids dynamically so you can increase budget without destroying ROAS.
Case Study: A Theater Client in Berlin
The problem: A theater in Berlin sold a mix of advance and last-minute tickets. Historically, the PPC account was highly segmented (ad groups segmented by match type and show name, multiple generic campaigns with manually adjusted daily budgets). Budgets needed to be constantly adjusted to avoid sending traffic when upcoming ticket availability was low — otherwise users bounced and conversion rates fell. Additionally, the brand saw limited success expanding to new customers through generic search terms.
The approach: Migration to a modern-search setup:
- Consolidated campaigns and ad groups around broader themes to ensure ML benchmarks were reached
- Switched to broad match with responsive ads and allowed Google’s ML to match queries.
- Implemented server-side conversion tracking and enhanced conversions to send purchase values and ticket-type metadata.
- Configured value-based bidding (maximize conversion value / target ROAS) so the algorithm weighted higher-value ticket purchases appropriately and understood the profitability of ads.
- Used DSA and Performance Max to capture niche sub-pages (smaller sub-pages stayed discoverable without fragmenting core ad group signals).
The result: Google’s ML began to automatically modulate bids and inventory exposure:
- The system reduced wasted spend during low-availability windows because expected ROAS began to decrease as tickets became sold out. This allowed the client to reinvest effectively in future weeks with more ticket volume.
- Advanced pre-sale revenue increased because ML identified buyers who purchase earlier when upcoming shows had low availability.
- Expanding into PMAX and adding broad match generic terms saw an effective increase in conversions from nonbrand search queries – the same keywords that were previously unsuccessful.
This is a clear example of how modern search turned tracking and consolidation into actionable, incremental revenue — things that weren’t achievable with the previous over-segmented approach. This led to more time available to spend on long-term planning, forecasting, and expansion into other paid media channels.
Final Notes & Modern Search Checklist
Quick checklist for modern search optimization:
- Consolidate campaigns/ad groups by theme, not micro-intent.
- Move low-volume keywords into broader ad groups (remove keywords with less than 10 impressions/week).
- Implement consent mode + enhanced conversions + server-side tagging.
- Feed purchase values into Google (value-based conversions) and set the right primary actions.
- Use broad match + responsive search ads + DSA/Performance Max to capture long-tail inventory.
- Monitor benchmarks (3,000 impressions/ad group/week, 10 conv./campaign/week).
- Stop manual bid adjustments and let smart bidding learn (but monitor and intervene when structural issues appear).
Need help implementing modern search?
Changing a long-standing PPC approach is hard. It requires technical work, cross-team coordination, and a willingness to give ML concentrated, high-quality signals. If you work with me, I will:
- Audit your account and tracking setup.
- Produce a prioritized migration plan (campaign consolidation + tracking roadmap).
- Help implement consent mode, enhanced conversions and server-side conversion feeds.
- Set up value-based bidding and performance monitoring.
If you want to maximize the success of your Google Ads, I can help you move from “doing search” to running modern search that scales profitably. Get in touch and I’ll draft a tailored action plan that drives real results.