Think Outside the Click #78

In partnership with

If you’re still optimizing partner and performance channels on last‑click and CTR, you’re playing a 2026 game with a 2014 scoreboard.

The way people discover, research, and buy has shifted to feeds, creators, AI answers, and social—while your attribution model is still rewarding whoever slaps on a coupon at the end.

In a world of AI results, answer engines, and “no‑click” journeys, the brands that win will be the ones that measure the full scope and impact of views and incrementality, not just clicks.

How Jennifer Anniston’s LolaVie brand grew sales 40% with CTV ads

For its first CTV campaign, Jennifer Aniston’s DTC haircare brand LolaVie had a few non-negotiables. The campaign had to be simple. It had to demonstrate measurable impact. And it had to be full-funnel.

LolaVie used Roku Ads Manager to test and optimize creatives — reaching millions of potential customers at all stages of their purchase journeys. Roku Ads Manager helped the brand convey LolaVie’s playful voice while helping drive omnichannel sales across both ecommerce and retail touchpoints.

The campaign included an Action Ad overlay that let viewers shop directly from their TVs by clicking OK on their Roku remote. This guided them to the website to buy LolaVie products.

Discover how Roku Ads Manager helped LolaVie drive big sales and customer growth with self-serve TV ads.

The DTC beauty category is crowded. To break through, Jennifer Anniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.

Learn more.

The click is the wrong hero

The click used to be the atomic unit of the internet.

Then tracking broke, AI moved the goalposts, and the platforms got a lot better at optimizing off their own signals.

Even if you remove AIO (Google’s AI Overviews), LLM (Chat GPT, Claude, Perplexity, etc.) impact click, and MTA (Multi-Touch Attribution) has been suboptimal for years…

That is a lot of acronyms!

So, even before the advent of AIO and LLMs and a shift in search behavior, there were many reasons to go beyond click-based measurement, Multi-Touch Attribution logic (MTA), and last click attribution.

Key Data Points: The Shift to Zero-Click 

• 2024–2025 Trend: Roughly 58.5% to 60% of Google searches in the U.S. ended without a click in 2024–2025.

• 2026 Projections: Analysts anticipate this will surpass 70% by mid-2026 as AI-driven answers continue to expand.

• Mobile Disparity: On mobile, up to 77% of searches now end WITHOUT a click, compared to roughly 47% on desktop.

• Impact of AI: Queries triggering AI Overviews (AIO) have an 83% zero-click rate.

• The "Vanishing" Click: For every 1,000 searches in the U.S., only about 360 clicks now go to the open web (non-Google properties). 

Growth in Zero-Click Searches (2017–2025) 

The trend shows a steady, accelerated climb in zero-click behavior: 

• 2017: ~48.5%

• 2019: ~50.3%

• 2020: ~64.8% (Pandemic peak)

• 2022: ~57.5%

• 2024: ~58.5% (AIO rollout)

• 2025: ~60%+ 

Why Are Clicks Declining?

Clicks are falling because the web's answers now live in three powerful ecosystems before users ever reach a site.

First came Google’s zero‑click era—now touching roughly 60% of all searches. AI Overviews, snippets, and local packs satisfy intent directly in the SERPs, meaning more search but fewer visits.

Then social feeds sealed themselves off. TikTok, Instagram, and YouTube turned from referral engines into self‑contained discovery platforms. Users scroll, watch, and buy without ever leaving the app.

@listmanguy

5 Most Viewed Tiktoks #viral #fyp #tiktok #top5

Now, AI assistants have joined the mix. Tools like ChatGPT, Gemini, Perplexity, and Claude capture part of the search intent entirely within the chat. When they do send a click, it’s usually a deep‑funnel decision, not casual exploration.

Together, these “answer systems” are quietly rewriting the web’s traffic map—growing attention, shrinking clicks.

Your Thinking About Attribution Wrong

If you’re anchoring decisions on last‑touch, you will under‑invest in the channels that actually create demand and over‑invest in whatever your ad platforms can easily claim. Eric Seufert calls it the “measurement myth”: the idea that if you can just perfectly attribute every click, you’ll perfectly allocate every dollar.

In reality:​​

• Identity is fragmenting (privacy, ATT, cookies), so deterministic click‑to‑conversion chains are at best incomplete.

• Platforms are already doing probabilistic measurement and “signal engineering” under the hood, while many marketers are still reading raw last‑click reports like gospel.

Here is an example of how this often plays out

• Meta shows you a modeled ROAS based on probabilistic lift and calibration tests (incrementality experiments feeding into MMM).

• Your analytics platform then credits only the final click (often branded search or direct), making paid social look weak.

• The platform’s optimization uses its probabilistic view; your budget decisions use your simplistic last‑click view. That gap is where a lot of money gets misallocated.

Meanwhile, the user journey has gone post‑click:

• An AI summary names your brand but never sends a referral URL.​

• A creator explains your solution in a video; the user remembers your name and later searches for it on Google.

• A partner comparison piece drives the first visit, but conversion happens weeks later by branded search on another device.

All of these are “zero‑click” or “invisible‑click” interactions that shape demand but never show up as a nice blue link in your reports.

Why last‑click and click‑only models fail partner programs

If you’re running affiliate or partner marketing, last‑click is not just wrong; it’s actively hostile to your best partners.​

Here’s how it quietly breaks your program economics:

• They hide true lift. Where brands run proper holdout or geo tests, they often find that partner traffic produces meaningfully higher conversion rates, AOV, and downstream value than non‑partner traffic, even when the last‑click numbers look similar. The lift is in the delta versus a no‑partner world, not in the raw click report.​

• They over‑credit the bottom funnel. Coupon, cashback, toolbar, and retargeting partners are structurally favored in last‑click models because they appear at the very end—often after demand is already created by content, media, or creators.​

• They erase the assist value. Many buyers first discover your brand through a review site, category authority, or podcast, then later convert directly or via brand search, which last‑click assigns to “Direct” or “Search” and not to the partner who did the heavy lifting.

Far too many in partner marketing fall into the easy / default trap of last‑touch: the numbers are clean and definite—and also wrong.

The cost is real: you starve top‑ and mid‑funnel partners while over‑funding whoever has the last cookie.​

Sometimes we call them the “Cookie Monsters!”

What “beyond the click” actually looks like

The right way to respond as an advertiser is to “triangulate”: use experiments (geo tests, lift tests) as ground truth, then calibrate MMM and attribution models to that, instead of worshiping whatever your last‑click report says.

Eric Seufert’s answer to broken attribution is not “give up,” it’s “upgrade the measurement stack.”

The modern playbook mixes experiments, modeled data, and platform signals instead of waiting for a perfect user‑level click trail. For partner marketing, that means:​

1. MMM is your friend

   1. Set up for impression data collection with your network and your partners

   2. We are testing a ton in this area to improve insights into what partners, partner types, and subsections of traffic are incremental and to what degree

2. Run incrementality tests, not dashboard arguments

   • If you are actually using MMM (Media Mix Modeling/Marketing Mix Modeling), it makes experimentation that much more effective. We recommend:

   • Use geo- or cohort-based tests: turn partner exposure off in some geos or cohorts, and compare revenue, new customers, and LTV to those in exposed audiences.

   • This can require an already mature higher volume program and partner volume and involvement, but it is possible

3. Track touches leading up to the last

   • For partners, mirror this: report on first‑touch, assisted‑touch, and multi‑touch paths where partners appear early, not just last‑touch conversions.

4. Optimize to higher‑order KPIs

   • Borrow from Brian Balfour’s growth models: focus your partner program on incremental revenue, new‑customer contribution, and LTV/CAC—not just short‑term ROAS.

   • Build a simple quant model tying partner touches to retention and monetization, so a “low ROAS” content partner that drives high‑LTV cohorts doesn’t get cut just because they rarely get the last click.

5. Align incentives with incrementality

   • Move beyond pure CPA to hybrid deals (base + performance kicker) for partners who reliably drive incremental, high‑quality customers, even if they lose in last‑click logs.​

   • Use test results to create tiers: “incremental growth partners” get higher commission or co‑marketing, “neutral” partners get standard terms, and “cannibal” partners get cut or de‑prioritized.​

In other words: elevate partners from “traffic vendors” to demand-creation partners and pay them accordingly.

Zero‑click, AI answers, and the “Answer Economy.”

A growing group of SEOs and growth leaders argues that the “click economy” is being replaced by an “answer economy.” AI‑driven search and assistants are less about sending traffic to ten blue links and more about returning one trusted answer.​

That has two big consequences for partner and performance marketers:

• Discovery is multi‑surface and often no‑click. Users discover you through AI answers, snippets, aggregators, chats, and devices, where the referring click is never attributed back. Jordan Koene points out that in this world, a user can ask a thermostat for a recommendation, then Google your brand name later, and you’ll never know the true first‑touch.

• Platforms optimize on their own signals, not your pixel. Seufert’s work on signal engineering shows how platforms can generate and optimize against richer intent signals than simple clicks, especially when user‑level tracking is constrained.​​

For partners, this means their influence is increasingly upstream and qualitative: being the expert cited in AI answers, the brand mentioned in a creator’s story, or the product recommended in a trusted community. The click is just one possible manifestation of that influence—and often not the most important one.

Your job shifts from “which click gets credit?” to “where is demand really being formed, and how do we test and model that at the portfolio level?”

Why this shift is strategically unavoidable

The move beyond click and last‑touch is not a nice‑to‑have; it’s where growth, partners, and platforms are already going.​

When you do this well in partner marketing:

• You unlock real scale. Once you can prove incrementality, it suddenly makes sense to invest aggressively in creators, content partners, and category authorities your CFO previously saw as “unprofitable” on last‑click.

• You become the brand serious partners choose. Partners talk. Programs that pay fairly for their true contribution—first‑touch, assist, and incremental lift—keep the best publishers, creators, and B2B partners and attract more of them.

• You finally align with how people actually buy. As Balfour stresses, growth is a system of loops and compounding interactions, not a linear funnel. Modern purchase journeys are multi‑touch, multi‑device, and increasingly mediated by AI. Measuring only the last click is like judging a football team solely on who carried the ball into the end zone.

If your attribution is stuck in 2014, your partner program strategy will be too. The next wave of performance and partner leaders won’t define success by who caught the last click. They’ll be the ones who can prove, with data, who actually moved the needle.

And they’ll think—very intentionally—outside the click.