The State of PPC 2026 report surveyed 1,306 PPC professionals globally. The findings are clear but uncomfortable: measurement is where everything breaks down.
53% of practitioners say PPC is harder than two years ago. The top reasons aren't about budgets or competition. They're about less transparency, less accurate measurement, and less control over what the platforms are doing.
And the client side mirrors it. When asked about satisfaction with their agency, the number one area of dissatisfaction was "helping with better measuring ROI."
This isn't a minor gripe. This is the thing that's breaking the agency-client relationship across the industry.
What's actually gone wrong
Three things have converged to make measurement genuinely difficult:
Consent Mode and signal loss. Privacy regulation has reduced the data Google and Meta receive from browsers. Consent Mode v2 models conversions where consent isn't granted, but the modelling introduces uncertainty. Smart Bidding strategies that rely on conversion data are now optimising on incomplete signals.
GA4 has become unreliable. The real issue isn't just that GA4 replaced Universal Analytics. It's that GA4 and Google Ads attribute conversions differently, leading to significant discrepancies between the two platforms. On top of that, GA4 was so difficult to use compared to Universal that many advertisers effectively stopped using it as a source of truth. It hasn't been properly configured, consent mode integration hasn't been handled correctly, and the data it reports has become something most people don't trust. The result is that one of the main tools people relied on for measurement no longer serves that purpose for most accounts.
Platform signals are hidden behind the black box. While Google has improved transparency in some areas - Performance Max now provides search term data, for example - much of the signal that drives how and where your ads are shown remains opaque. The algorithm decides who sees your ads based on signals you can't see or influence directly. That's fine when measurement is solid, but when it isn't, you're trusting a system you can't verify.
The funnel tracking problem
Here's a specific example of what we see regularly.
Most ecommerce Google Ads accounts track purchases. That's the conversion action Smart Bidding uses to optimise. But purchases are the last step in the funnel. By only tracking the end point, you're giving the algorithm one data point per customer journey instead of four.
The conversion funnel has four stages that Google Ads can track:
- View item - someone looked at a product page
- Add to cart - they added it to their basket
- Begin checkout - they started the checkout process
- Purchase - they completed the order
We recently found accounts that were missing key funnel steps - no add to cart, no begin checkout, no view item. Just purchases. It's more common than you'd expect.
That means Smart Bidding is working with a fraction of the signal it needs. It can't distinguish between a campaign that drives lots of product page views but no purchases (wrong audience) and a campaign that drives fewer views but higher intent (right audience, lower volume). Without the upstream events, it's guessing.
Why upstream funnel events matter more than ever
Before Consent Mode, Google could see most of the purchase journey through cookies and browser signals. The platform could infer funnel behaviour even if you weren't explicitly tracking it.
That's no longer the case. With consent rates varying by market, Google is seeing significantly less browser-level data. The conversion actions you explicitly set up are now the primary signal source.
But there's another reason the upstream funnel events are important that goes beyond just giving Smart Bidding more data points.
When you track add to cart, begin checkout, and view item events, you're telling Google what type of user takes action on your site - even when they don't purchase. Google uses this signal to understand the characteristics of people who engage with your products. It builds a profile of high-intent users and goes after more people like them.
Without those upstream events, Google only knows about the people who bought. That's a small subset. With them, it knows about everyone who showed purchase intent - the ones who added to cart but didn't check out, the ones who started checkout but dropped off. That's a much larger and richer signal set, and it gives the algorithm a better picture of who to target with the view that some of those users will convert later in the funnel.
In a reduced-signal environment post-Consent Mode, this matters more than it ever has.
What a proper measurement setup looks like
Based on managing ecommerce and lead generation accounts across multiple verticals, here's what we consider the baseline for any ecommerce account:
Conversion tracking:
- All four funnel events (view item, add to cart, begin checkout, purchase) set up as conversion actions
- Purchase set as the primary conversion action for bidding
- Funnel events set as secondary (providing signal without inflating conversion counts)
- Enhanced Conversions enabled (sends hashed first-party data back to Google for better match rates)
Reporting:
- Server-side conversion verification (comparing Google's reported conversions against your actual order data)
- Profit tracking where possible (revenue minus cost of goods, not just top-line revenue)
- Cross-platform deduplication (so Google and Meta aren't both claiming the same sale)
Monitoring:
- Automated alerts for conversion tracking failures (tags stop firing, consent rates drop)
- Regular audits of conversion action configuration (primary vs secondary, attribution windows, counting methods)
This isn't aspirational. This is what's needed to make Smart Bidding work properly in 2026.
The agency gap
The State of PPC report found that 20% of clients are considering replacing their agency with AI tools.
That's not surprising when the core complaint is measurement. If an agency can't tell you what's working, what's the value? You might as well let the platform auto-optimise and save the fee.
But the agencies that own measurement - that can connect ad spend to actual profit, that can tell you which campaigns are driving real business outcomes rather than just clicks - those agencies are irreplaceable. No AI tool is setting up server-side conversion verification or building custom attribution for your specific business.
The measurement gap is also an opportunity gap. If your current setup can't tell you whether your ads are profitable, that's the first thing to fix. Everything else - creative testing, audience strategy, bid optimisation - is downstream of having the data right.
Where to start
If you're running ecommerce PPC and you're not sure about your measurement setup, start here:
Check your conversion actions. In Google Ads, go to Goals > Conversions > Summary. Do you have add to cart, begin checkout, and purchase? Are they all receiving data?
Verify Enhanced Conversions. Under each conversion action, check whether Enhanced Conversions is enabled. If it's not, you're leaving match rate on the table.
Compare reported vs actual. Pull your Google Ads conversion count for last month. Compare it to your actual order count. If there's more than a 15% gap, something needs investigating.
These three checks take about 20 minutes and will tell you whether your measurement foundation is solid or needs work.