PPC attribution has never been perfect, but AI is widening the gap between what influences a conversion and what gets credit for it.

Someone might discover a product on social media, watch a YouTube review, read opinions on Reddit, ask an AI-powered search tool to compare options, and then return days later through a branded Google search ad.

The PPC report will show one conversion from branded search. Technically, that may be accurate. Strategically, it’s incomplete.

AI is changing where people discover brands, how they research purchases, and how advertising platforms decide who sees an ad. At the same time, you have less visibility into the decisions those platforms make on your behalf.

As a result, you can’t afford to treat platform attribution as business truth.

AI is changing where the journey begins 

The traditional search journey is already being disrupted before an advertiser receives a measurable click.

Responsive’s 2025 “Inside the Buyer’s Mind” research found that generative AI had overtaken traditional search for one-quarter of B2B buyers, with nearly two-thirds using AI as much as or more than search when researching vendors.

The shift is even more pronounced in technology. The same research found that 80% of technology buyers use generative AI at least as much as traditional search for vendor research. More than half use LLM assistants as a leading source for discovering new vendors.

If a buyer is using AI to create a shortlist before visiting a search engine or company website, visibility during that research phase matters.

If your brand isn’t included in the initial AI-generated answer, you may not even make it into the consideration set.

Google’s own search experience is accelerating this change. At Google I/O in May, the company announced that AI Overviews had reached more than 2.5 billion monthly active users. AI Mode had surpassed a billion monthly active users.

Pew Research Center found that users clicked a traditional search result in just 8% of visits when an AI summary appeared, compared with 15% of visits without an AI summary.

The important point isn’t simply that clicks are declining.

Visibility and influence can exist without a website session. A person may read an AI Overview, remember a brand name, and search for it later. They may use ChatGPT, Gemini, or Perplexity to compare suppliers before returning directly to a website or clicking a branded ad.

The discovery touchpoint influences the conversion, but the lower-funnel interaction receives the credit.

Dig deeper: What 4 AI search experiments reveal about attribution and buying decisions

Branded search often receives credit for demand generated elsewhere

Branded search is a clear example of attribution confused with impact.

A branded search campaign will often produce efficient cost-per-acquisition figures, high conversion rates, and an impressive return on ad spend. It reaches people who already know what they want and are actively looking for the business.

That makes branded search valuable. But it doesn’t mean the campaign created the demand.

The user may be searching for the brand because they encountered it in an AI-generated comparison, watched a creator review on YouTube, saw several paid social ads, or read a discussion on Reddit.

The branded ad captures existing intent.

This is why PPC reporting needs to distinguish between demand capture and demand creation.

Branded search, remarketing, and some lower-funnel Performance Max activity may appear to be the most effective campaigns in an account because they reach people who are already close to converting.

The more useful question isn’t whether those campaigns delivered conversions, but how many of those conversions would have happened if the campaigns hadn’t existed.

That’s the difference between attribution and incrementality.

AI-driven discovery creates a measurement blind spot

We’re already seeing the implications of this shift in client data. One client whose traffic was referred directly from AI platforms converted at 8.31% across the year, compared with 2.93% for organic search traffic.

The AI audience was far smaller: 565 measurable visits compared with approximately 17,000 organic visits. But 47 of those AI-referred visitors converted.

This comes with an important caveat. It’s one client example, not an industry benchmark. The measurable visits also represent only the people who clicked through from an AI platform. We don’t know how many users encountered the brand within an AI-generated answer and later returned through another channel.

That’s exactly the attribution problem.

Visible AI traffic may represent only a small portion of the overall journey. Some of the resulting demand may later appear as direct traffic, organic brand traffic, or branded paid search conversions.

In another example, we saw AI-referred visits increase by approximately 150% month over month for an industrial machinery client. This made sense when we considered the purchase journey. Buyers were asking highly technical questions, comparing features, and researching which type of equipment was most appropriate for a specific application.

AI search is particularly useful when buyers want a detailed comparison distilled into a clear answer.

This creates both an opportunity and a reporting challenge. The final paid search click may be only the last visible step in a much longer research process.

Ads are becoming part of AI-generated search journeys

With ads now integrated into AI-powered results, the relationship between AI search and paid media has become even more complicated.

Google states that ads can appear above, below, or within AI Overviews. Existing Search, Shopping, and Performance Max campaigns may appear within them when the query and the AI Overview indicate relevant commercial intent.

The example Google provides is revealing: A user might ask why their pool has turned green and how to clean it. The query isn’t explicitly transactional, but Google may serve an ad for a pool vacuum based on the context of the AI-generated response.

That means ads can appear earlier in the research journey, alongside more complex queries that don’t resemble a conventional commercial keyword.

You have limited visibility into this.

Google states that advertisers can’t target AI Overview placements directly, can’t opt out of appearing within AI Overviews, and don’t receive segmented reporting for ads served within them.

This is significant. You’re encouraged to participate in a new search experience, but there’s no clear way to isolate its performance.

Google is also testing new AI-powered ad formats in Search, including Conversational Discovery ads, Highlighted Answers, AI-powered Shopping ads, and a Business Agent for Leads.

These formats may create new opportunities to reach customers while they are actively researching. They may also make it harder to answer a deceptively simple question: What exactly caused the conversion?

Platform automation can make attribution look better while making analysis harder

The attribution challenge isn’t limited to AI search. It’s also happening inside ad platforms.

In Meta, broad Advantage+ audiences often outperform tightly defined demographic audiences on surface-level metrics such as reach, impressions, and click-through rate.

That sounds positive. Sometimes it is. When targeting becomes broader and the platform dynamically adjusts creative, messaging, and delivery, it becomes harder to understand exactly why an ad performed well.

Was it the creative? The headline? The audience? The product? The offer? The placement?

In some cases, conversion volume may increase, but reporting granularity doesn’t.

This matters because the next optimization decision depends on understanding what worked. A strong message could be reflected on the landing page. A specific customer pain point could inform the next creative brief. A high-performing audience insight could shape the broader marketing strategy.

When the platform retains those insights, you’re left feeding more creative variations into the system without fully understanding which variables are driving performance.

This becomes particularly risky when the platform’s metrics aren’t aligned with business outcomes.

A campaign can deliver excellent reach, a strong click-through rate, and a low cost per click while still generating poor-quality leads or low-value purchases.

It’s like feeding an absolute monster with money and creative assets without always receiving the information needed to make better decisions.

That might sound dramatic, but it reflects a real concern for those managing limited budgets.

Automation can help with delivery. It shouldn’t remove the need for scrutiny.

---

Poor-quality traffic can affect future optimization

This is one reason you should be cautious about adopting broad targeting or broad match simply because a platform recommends it.

Broad targeting can work well in the right account, particularly when the platform has a strong volume of high-quality conversion data.

It can also bring in low-quality traffic. That affects more than the immediate campaign report.

Poor-quality traffic can influence website audiences, remarketing pools, conversion signals, and future automated bidding decisions. If an account is optimizing toward the wrong actions or feeding low-quality leads back into the platform, the system may become better at generating more of the wrong outcome.

This is where lead quality matters. A campaign generating 30 form submissions and two genuine opportunities may be less valuable than a campaign generating 12 inquiries and six qualified opportunities.

The first campaign may look stronger in Google Ads or Meta Ads. The second campaign may be more valuable to the business.

PPC teams need access to CRM outcomes, not just lead volume.

Automation also creates a new layer of reporting risk

Increased automation means PPC teams must be vigilant about conversion actions and assets used by each platform.

During our podcast, I shared an example where an account suddenly showed approximately 50,000 additional conversions within a matter of days. Google had introduced Google-hosted local engagement actions into the account reporting.

Without closely checking the conversion setup, the report would have painted an extremely misleading picture of performance.

There have also been situations where Meta surfaced an outdated promotion from a website and used it within an ad, even though it wasn’t the offer the client wanted to promote.

These aren’t arguments against automation.

They’re arguments against assuming platform defaults are automatically aligned with the advertiser’s objectives.

Conversion settings, automated assets, product selections, and campaign recommendations need regular human review.

Upper-funnel campaigns influence lower-funnel conversions

The limitations of attribution become particularly important when businesses assess upper-funnel activity.

Google and WARC analyzed long-term media investment studies from European brands and found that returns generated within the first four months were equal to the returns generated across the following 20 months.

The same analysis found that average short-term profit ROI increased from $2.50 for each $1.34 invested to $5.50 when sustained effects were included.

You shouldn’t apply these figures directly to every account. A local lead gen campaign will behave differently from an established ecommerce brand. The principle still matters.

If a business reduces video, paid social, or awareness activity because last-click ROAS looks weak, it may improve dashboard efficiency in the short term while quietly reducing future demand.

Branded search volumes may then decline several weeks or months later. Remarketing pools may shrink. Lower-funnel campaigns may deteriorate, even though nothing material changed within those campaigns.

Dig deeper: How to measure paid social’s impact on PPC

What PPC teams should report in 2026 

A single ROAS figure is no longer enough. PPC reporting needs to combine platform attribution with broader business indicators and structured experimentation.

1. Separate demand creation from demand capture

Don’t assess every campaign against the same expectations.

Branded search and remarketing are designed to capture existing intent.

Paid social, YouTube, demand gen, and upper-funnel campaigns are more likely to influence awareness, consideration, and future searches.

For demand capture campaigns, focus on efficiency, coverage, conversion rate, and marginal value.

For demand creation campaigns, include new customer growth, assisted conversions, branded search trends, direct traffic, audience growth, and lift testing where available.

2. Review attribution paths, not just final clicks

GA4’s key event attribution paths report helps you understand which channels initiate, assist, and close conversions. It also shows metrics, including days to conversion and touchpoints to conversion.

This is a useful starting point for understanding the broader journey.

Review:

• Touchpoints to conversion.
• Days to conversion.
• Revenue from multi-touch journeys.
• Early-stage and late-stage interactions.
• New and returning customer performance.
• Differences between high-value and low-value conversions.

A campaign with a lower last-click ROAS may still play an important role in bringing new customers into the funnel.

3. Import deeper CRM outcomes

Lead volume alone isn’t enough. Where possible, feed qualified leads, opportunities, and completed sales back into the ad platforms.

Google recommends enhanced conversions for leads as an upgraded form of offline conversion import. It uses hashed first-party customer data and supports more accurate reporting, engaged-view conversions, and cross-device conversions.

The platform can only optimize toward the signals it receives.

If every form submission is treated as equally valuable, the bidding system has no reason to prioritize the leads most likely to become customers.

4. Monitor the metrics sitting outside the PPC dashboard

AI-driven discovery means you need to look beyond directly attributed conversions.

Track:

• Branded search volume.
• Direct traffic.
• Organic brand traffic.
• AI-referred sessions.
• AI-referred conversion rates.
• New customer acquisition.
• Returning visitor conversion rates.
• Assisted conversions.
• CRM lead quality.
• Revenue by customer type.

These figures won’t prove causation individually. Together, they provide context that a last-click report can’t.

5. Test incrementality rather than assuming

Where account eligibility and budgets allow, use controlled testing.

Google describes its Conversion Lift as an incrementality tool that measures purchases, visits, and other conversions directly driven by exposure to ads. It compares users who saw the ads against a control group who did not.

Run practical tests independently:

• Pause branded search activity in selected regions while retaining coverage elsewhere.
• Compare similar geographic areas with different upper-funnel investment levels.
• Test remarketing holdouts.
• Monitor branded search demand before, during, and after video campaigns.
• Compare new customer acquisition rates rather than relying on blended ROAS.

No test is perfect. Controlled experiments answer a more useful question than a platform dashboard: What changed because the advertising existed?

6. Add regular human checks to automated accounts

Automation should reduce repetitive work. It shouldn’t remove accountability.

Review:

• Conversion actions included in bidding and reporting.
• Automated assets and extensions.
• AI-generated headlines and descriptions.
• Product selections.
• Promotions.
• Search terms.
• Lead quality.
• Audience quality.
• Website content that platforms scrape.

The platforms are changing quickly. A setup that was appropriate last month may not behave exactly the same way today.

Dig deeper: Why your B2B PPC metrics may be lying to you

Stop searching for one perfect attribution model

There’s no single PPC attribution model that can fully explain a fragmented, AI-influenced customer journey.

Last-click reporting is too narrow. Platform attribution is useful but partial.

Data-driven attribution is more sophisticated, but it’s still limited by the interactions each platform can observe.

The answer isn’t to abandon attribution. It’s to stop treating attribution as proof of causation.

In 2026, PPC teams need to report performance using multiple layers of evidence: platform data, analytics paths, CRM outcomes, new customer trends, branded search demand, AI-referred traffic, assisted conversions, and controlled testing.

The most useful question is no longer, “Which channel received credit for this conversion?”

Instead, ask: “What would have happened if this activity hadn’t existed?”