Category: Positive Sparks News
Platform: WordPress
What if the real growth lever in Google Ads wasn’t more budget, new creatives, or broader targeting, but teaching the algorithm which users were actually worth winning? That was the working theory behind this 30-day test.
The hypothesis
Why did standard Smart Bidding feel efficient on paper but underwhelming in the actual P&L? Because it was largely optimising around last-click conversion events, while a meaningful share of buyers were visiting the site 3 or more times before purchasing. In this account, those repeat visitors were often the highest-value customers, not the cheapest ones.
We believed Google Ads was underpricing early-session intent.
So we fed the system more meaningful journey signals: add-to-cart, deeper product-page engagement, and time-on-site thresholds tied to high-intent sessions. The idea was simple. If we could help the bidding model distinguish between casual traffic and users showing layered buying behaviour, we could push budget toward higher-value click paths rather than just cheaper conversions.
Actionable takeaway: If your customers rarely buy on the first session, don’t let bidding optimise as if they do.
The technical setup
How did we actually build it? We created a custom value-based conversion action in Google Ads weighted around predicted customer value, not just the first transaction.
That value model included:
- first purchase value
- 30-day repeat purchase probability
- micro-conversion signals from pre-purchase sessions
- offline subscription signup data imported back into Google Ads
We also implemented Enhanced Conversions to improve match quality and support offline conversion import for subscription signups that happened after the initial ad click. Then we changed bidding from Maximize Conversions to Maximize Conversion Value and set a tROAS floor of 400%.
In plain English, we stopped asking Google to find us more conversions at any cost and started asking it to find us more revenue-weighted customers above a profitability threshold.
Actionable takeaway: If you only pass purchase events, you’re giving the algorithm a flat picture of buyer quality. Add value layers before you scale spend.
The 30-day result
So what happened after 30 days?
| Metric | Before | After | Change |
|---|---|---|---|
| Unique converting users | 100 baseline | 127 baseline | +27% |
| Blended ROAS | 3.2x | 4.8x | +50% |
| Cost per incremental user | 100 baseline | 82 baseline | -18% |
The headline result was the 27% increase in unique converting users, using Google’s own reported metric. Just as important, blended ROAS improved from 3.2x to 4.8x, and cost per incremental user fell 18%.
That combination matters. We weren’t just buying more conversions by paying more for them. We were acquiring more converting users while improving efficiency.
Actionable takeaway: Don’t judge bidding tests on CPA alone. Look at unique converters, blended ROAS, and incremental acquisition cost together.
The nuance
Will this work for every account? Not exactly.
This setup performed best in brands with more than 500 conversions in the last 30 days. At that level, Google had enough recent signal density to learn from value weighting and offline feedback loops without becoming unstable.
For smaller accounts, we recommend starting with a 90-day lookback window before pushing hard into journey-aware value models. Bear in mind, the goal isn’t to force complexity too early. The goal is to give the system enough clean data to make better decisions.
Actionable takeaway: If volume is low, widen the learning window before changing bidding strategy.
The highest-impact change
What made the biggest difference? Honestly, it wasn’t the micro-signals on their own. The single highest-impact change was moving from CPA-style optimisation to value-based bidding and uploading offline conversion data.
That gave Google a better definition of what a good customer looked like.
If you’re deciding where to start, do that first. Then layer in journey signals once the value model is stable.
What are you seeing in your own account right now? Are your highest-value customers converting on the first visit, or are they giving you multiple signals before they buy? And if you haven’t switched to value-based bidding yet, what’s holding you back?
