PPCcopperchunkppc

State of Paid Media 2026: automation needs better guardrails

Can-Do Digital's State of Paid Media report is another signal that PPC is moving away from dashboard execution and towards better commercial decision-making.

Published: 1 June 2026 | Updated: 1 June 2026

The short version

Chris Nightingale and Can-Do Digital have published The State of Paid Media 2026, produced with Opteo. The public page says the report is built on analysis of 22 million clicks and focuses on the human side of automation.

The main themes are familiar if you run modern PPC accounts: longer conversational queries, fewer clean channel boundaries, more AI-assisted buying journeys, zero-click commerce, and a bigger premium on human judgement.

The useful takeaway is not "use more automation". It is "automation is now powerful enough to need proper commercial guardrails".

What I agree with

The strongest point is that PPC is becoming less about execution and more about decision quality. Smart Bidding, broad match, Performance Max and Demand Gen already handle a lot of the mechanics. The performance gap is now in what you feed them and what you allow them to do.

That means the valuable work sits around the edges of the ad platform:

  • clean conversion tracking
  • first-party customer data
  • profit and margin feedback
  • feed quality and semantic product data
  • creative systems
  • landing pages and post-click experience
  • guardrails around placements, URLs, budgets and conversion goals

That is the right direction. The platform has the auction intelligence. The advertiser still needs to provide the business intelligence.

Broad match is not the problem

The report summary makes the point that search behaviour is becoming more conversational. That is true. People are not just typing neat two-word keywords. They are asking full questions, comparing situations and using natural language.

Broad match is built for that world, but it only works when the rest of the account is mature enough. Broad match with Smart Bidding, strong conversion data and sensible structure is very different from broad match pointed at a weak landing page with raw form fills as the only goal.

This is where advertisers need to be careful with negative keywords. It is easy to over-clean an account and block useful intent because a search term looks messy in isolation. With Smart Bidding, context matters. Use search-term analysis to find candidates, but do not automatically negate everything that feels untidy.

For lead generation

For lead gen, the report's first-party data argument is the practical one. If you cannot tell Google which leads became qualified opportunities, booked appointments or customers, the machine will optimise to whatever fills forms fastest.

The fix is not another campaign setting. It is a feedback loop:

  1. Capture click IDs and UTMs on landing.
  2. Store them with every submitted lead.
  3. Connect the lead to CRM status, sales outcome and rejection reason.
  4. Upload qualified offline conversions back to Google Ads.
  5. Retract or adjust conversions when a lead turns out to be junk.

Without that, automation can scale the wrong thing. With it, broad match, PMax and Demand Gen have a fighting chance of learning from real commercial outcomes.

For ecommerce

For ecommerce, the important themes are profit-first measurement, product feeds and zero-click commerce. If Google, Microsoft, Amazon, TikTok and AI agents are all becoming discovery surfaces, then the product data becomes part of the sales pitch.

A product feed is no longer just a technical upload. It is the structured evidence that tells machines what the product is, who it is for, what makes it different, whether it is in stock, what margin it carries, and whether it deserves spend.

The measurement point matters even more. ROAS inside a platform is useful for optimisation, but it is not the same as commercial truth. Ecommerce accounts need to separate:

  • brand versus non-brand
  • new versus returning customers
  • first-order ROAS versus lifetime value
  • revenue versus gross profit
  • orders driven by discounting versus genuinely incremental orders

If you feed net profit or high-value customer data back into the platforms, automation can optimise towards better customers. If you only feed revenue, it will happily chase low-margin volume.

The guardrail problem

The report summary also highlights wasted spend inside automated systems: poor placements, irrelevant search terms and default settings that hand too much freedom to the platform.

This is the uncomfortable bit of modern PPC. The best accounts usually work with automation, but they do not leave it unsupervised. Guardrails matter:

  • placement exclusions where brand risk is obvious
  • search-term review without panic-negating useful intent
  • careful final URL expansion rules
  • asset review before generated creative goes live
  • budget and bid limits where the business cannot absorb volatility
  • separate goals for real customer value, not just easy conversions

The aim is not to micromanage the machine. The aim is to stop it having permission to spend money in ways the business would never approve if a human proposed it directly.

Decision-making beats activity

The best line from the report summary is that PPC is no longer really about optimisation. It is about decision-making.

That is exactly right. Most accounts do not need more button pressing. They need clearer decisions about what deserves spend, what needs more data, what should be left alone, and what is quietly wasting money.

Good PPC in 2026 looks less like daily tinkering and more like commercial operating discipline:

  • know what a good customer is worth
  • know which signals are reliable
  • know when not to touch a learning system
  • know when the platform report is only an attribution claim
  • know which constraints protect the business

My operating rule

If automation is making the decision, your job is to improve the data, set the constraints and judge the commercial outcome.

That is the useful frame for the State of Paid Media 2026. The work is not to outsmart Google at auction time. The work is to feed it cleaner signals, enforce better guardrails and measure the result against business data.

That is a more interesting version of PPC than keyword tweaking. It is also harder to fake.

Sources