Fairing today made Advanced Attribution available in beta, a product that adds a tailored second question to the "How did you hear about us?" survey shown to shoppers after checkout, so that a customer who answers "Podcast", "YouTube" or "ChatGPT" can go on to name the specific show, creator or query behind the purchase. The New York and London-based company said the feature will reach general availability in January 2027.

In Short

Many online shops ask buyers where they first heard about a brand, and Fairing now follows that answer with a second question asking which podcast, creator, TV placement or AI assistant it was. This matters for brands spending money on podcasts, influencers and TV, where people rarely click an ad, so standard tracking tends to give the credit to a later Google search instead. Fairing says early users got up to 50% more usable answers from the same number of orders, although that figure comes from the company itself and has not been independently checked.

What the product does

The mechanism is simple. Fairing's core product is a post-purchase survey, typically placed on the order confirmation page of an e-commerce store, which asks customers a single question: "How did you hear about us?" The industry shorthand is HDYHAU. According to Fairing, the company helped popularise the format, and Advanced Attribution is built directly on top of it.

When a shopper selects a channel from the list, the new product triggers a follow-up question whose content depends on that selection. According to Fairing, a customer who picks YouTube can then identify the creator who influenced them, a customer who picks Podcast can identify the show, and those who cite ChatGPT or another AI assistant can describe what they asked. The effect is to convert a channel-level response into what the company calls a "named source".

A second component deals with the messiest part of any survey: the free-text box. Advanced Attribution adds managed auto-suggest lists, which the company says keep response options current without brands having to maintain them by hand. As a customer types into the "Other" field, matching options appear in real time. The purpose, according to Fairing, is to turn answers that would otherwise be ambiguous into structured data that can be counted and compared.

That second feature is less glamorous than the first, yet arguably as important. Free-text responses are notoriously hard to aggregate. One respondent types a podcast's full title, another abbreviates it, a third misspells the host's name. Without normalisation, a brand running sponsorships across a dozen shows can end up with hundreds of variant strings and no reliable count. Auto-suggest pushes respondents towards a canonical label at the point of entry rather than leaving analysts to reconcile the mess afterwards.

The number, and what it does not say

The headline figure is Fairing's own. In early rollouts, according to the company, Advanced Attribution uncovered as much as 50% additional usable attribution signal from the same order volume, with no change to survey placement or response rates.

Several details are absent. The release does not say how many brands took part in those early rollouts, over what period, or in which product categories. "Usable attribution signal" is not defined, so it is unclear whether the increase reflects more respondents answering, more answers being classifiable, or the same answers being broken into finer categories. The "up to" qualifier also means 50% is a ceiling rather than an average. No third-party audit or methodology note accompanies the claim.

That last point is not trivial. A follow-up question can only add detail to customers who already answered the first one, so the claim that response rates did not change is plausible on its face. Whether more granular answers translate into better budget decisions is a separate question, and one the announcement does not attempt to answer with data.

Why clicks miss so much

Fairing frames the product around a gap that measurement specialists have discussed for years. According to the company, brands are spending more than ever on channels that never generate a measurable click. A customer hears about a product on a podcast, watches a creator use it, or asks ChatGPT for a recommendation, then searches for the brand days later. Pixel-based and modelled attribution systems credit that final touch, usually search or direct traffic, and the channel that created the demand goes unrecorded. As discovery spreads across audio, video, influencers and AI, Fairing argues, the gap is getting wider.

The argument is well documented. Academic work covered by PPC Land in March 2025 showed that last-touch attribution can significantly underperform alternative methods when consumers see advertising across several publishers before converting. Google's own reporting options have drawn criticism on similar grounds: the "Google paid channels last click" model assigns 100% of conversion value to the last Google Ads channel a customer touched. Meta, which sells exactly the sort of upper-funnel inventory that last-click systems undervalue, published a white paper drawing on 307 studies that concluded advertisers undervalue Meta by 31% at the median when relying on rules-based attribution rather than incrementality measurement. That study came from a platform with an obvious stake in the result, but the direction of the finding is consistent with independent work.

Industry surveys point the same way. The IAB's State of Data 2026 report found that up to 75% of marketers say attribution, incrementality tests and marketing mix models underperform, with 48% of mix-model users saying creator and influencer marketing is underrepresented in their models. A survey-based approach sidesteps some of these problems because it does not depend on observing a click, a cookie or a device identifier. It brings others, which are discussed below.

Fairing's own description of its business leans heavily on that distinction. The company describes itself as an attribution platform that turns direct customer feedback into marketing attribution data, operating without cookies or pixels. The approach draws on zero-party data, information customers knowingly volunteer. Research covered by PPC Land in April 2026 found that 60% of marketers struggle to predict audience behaviour, part of the reason declared data has attracted attention as privacy rules narrow what can be tracked.

Podcasts: spend rising faster than proof

Podcasting is the channel Fairing chose to illustrate the product, and the choice fits the market. Magellan AI data showed podcast ad spend peaking at $408 million in December 2025, after year-over-year growth of 26% in the third quarter and 32% in the fourth. Measurement has not kept pace. A survey published in July found that only 20.4% of respondents measure direct sales or revenue attribution from podcast activity, and just 16.8% track formal return on investment.

The existing toolset for podcast measurement is largely pixel-based. Magellan AI's methodology, for instance, matches listener data and advertiser site data to households using the Experian graph, producing a reported conversion rate of 5.22% for the first quarter of 2026. The same company in March integrated Nielsen's DMA data into podcast attributionand later extended its measurement to broadcast radio, in a market where consumers spend 31% of media time with audio while advertisers allocate 9% of budgets to it. Podscribe, a separate vendor, in July began tracking Instagram, TikTok and YouTube posts alongside podcast campaigns in a single dashboard.

The problem those tools share is the one Fairing's release highlights. Even when targeting improves, attribution at the outcome layer remains unsolved for audio, as PPC Land noted when Triton Digital opened podcast buying by audience profile earlier this month. And with video now central to podcast consumption, creator-level attribution across audio, video and social remains technically complex.

Podscale, a podcast advertising business, is quoted in the announcement as an early user. (It is a different company from Podscribe.) Jessica C., identified only by first name and initial, is its VP of Operations. "Knowing a customer heard about a brand through a podcast doesn't provide advertisers with enough information," she said. "We need to understand which shows influence purchases, so we can make better decisions about where to invest and how to scale campaigns. Advanced Attribution gives us that additional level of insight directly from the customer, helping connect individual podcast placements more clearly to revenue."

Creators and TV

The same logic applies to creator marketing, where spending has climbed without a matching improvement in proof. An Association of National Advertisers survey found 67% of marketers call measurement the hardest step in influencer marketing, noting that different brands, agencies and platforms use different KPIs and attribution models. US creator economy ad spend reached $37 billion in 2025 with $43.9 billion projected for 2026. The boundary between creators and television is also blurring: Spotter's research put 6,600 US YouTube creator channels at a scale rivalling prime-time TV, with 52% of viewing on connected TV.

Television is the oldest example of the pattern. Viewers see a spot on the big screen and act on a phone or laptop later, which is why second screen measurement often comes down to watching branded search and direct visits rise after a flight. Fairing lists TV placements among the sources Advanced Attribution can identify, although the release gives no example of how a TV-specific follow-up question is worded or what granularity it reaches, whether network, programme or individual spot.

AI assistants: the newest blind spot

The inclusion of AI platforms is the most timely element. PPC Land reported in July 2025 that ChatGPT referrals to news publishers had grown 25 times year-over-year. Referral tracking has improved since. OpenAI added UTM parameters to more ChatGPT links in June 2025, Google Analytics first suggested custom channel groups for AI chatbots that August, and in May 2026 added a dedicated AI assistant channel covering ChatGPT, Gemini and Claude.

Those fixes only capture visits that arrive by clicking a link. Much AI influence does not. Similarweb research published in June found that 55.9% of AI-influenced traffic arrives via search, compared with 40.4% for visits without AI influence, because users read a recommendation and then type the brand name into a search engine. Semrush data covering January to July 2026 showed direct traffic accounting for 56.65% of sessions to manufacturing sites, a bucket that absorbs conversations that began inside a chatbot. Earlier this month, NIQ and Similarweb said they are building a product to measure shopping inside ChatGPT and Gemini from the fourth quarter of 2026, and yesterday PPC Land covered brand rebates delivered inside ChatGPT and Claude, another interaction that leaves no click behind.

Asking customers what they typed into an assistant is a different kind of evidence from panel or clickstream data. It captures intent in the customer's own words, which may be useful for understanding which prompts surface a brand. The platform mix is also shifting quickly, with Similarweb data showing ChatGPT's share of generative AI traffic dropping to 52.7% while Gemini rose to 27.3%. A survey list that treats "AI" as one bucket would miss that shift; one that names platforms and captures queries would not, provided the options stay current, which is the job the managed auto-suggest lists are meant to do.

The limits of asking

Self-reported attribution has well-known weaknesses. Customers remember the most salient touchpoint, not necessarily the first or most influential one. Recall fades over days or weeks. Some respondents pick the first option on a list. Others conflate a creator seen on Instagram with one seen on YouTube. A follow-up question adds detail, but detail built on imperfect recall is still imperfect.

There is also a question of coverage. A post-purchase survey only reaches people who bought and chose to answer, so it says nothing about exposure that failed to convert. That makes it a poor tool for measuring efficiency on its own. Survey data can suggest which podcast listeners bought; it cannot say how many listeners heard the ad and did not. It also shares a basic property with every attribution method: correlation is not causation. Survey answers describe perceived influence, and establishing whether a placement actually caused sales still requires experiments with control groups.

Word of mouth poses a further complication. Customers who heard about a product from a friend who heard it on a podcast may answer "friend", and the podcast receives no credit. Fairing's release does not describe how Advanced Attribution handles these indirect chains.

None of this is unique to Fairing. Survey-based attribution is widely used among direct-to-consumer brands precisely because it complements, rather than replaces, pixel and model data. Matt Bahr, Fairing's co-founder and chief executive, positioned the product in those terms. "Marketing measurement has traditionally been very good at telling brands where a conversion happened, but much less effective at explaining what created the demand," he said. "That's becoming a bigger problem as more discovery happens in environments that never produce a click. Advanced Attribution gives brands a layer of evidence no pixel or model can: the customer's own account of what influenced their decision, captured in enough detail to act on."

Dates, and a discrepancy

The announcement is dated September 15, 2026, and was issued from New York and London. Its text, however, states that Advanced Attribution "launches on September 10 in beta", a date five days before the release itself. The document does not explain the gap. The most plausible reading is that beta access opened on September 10 and the public announcement followed, but the release does not say so explicitly.

According to Fairing, brands can currently request early access. The company said it will continue developing the product based on customer usage and feedback and plans to expand the feature set over the coming months, ahead of general availability in January 2027. No pricing, eligibility criteria, supported e-commerce platforms or beta participant numbers were disclosed. Fairing's existing survey product is distributed through the Shopify App Store, but the release does not say whether Advanced Attribution is limited to Shopify merchants.

For advertisers, the practical significance lies less in the novelty of the technique than in its granularity. Channel-level survey data has told brands for years that "podcasts" or "YouTube" mattered. What buyers negotiate over is the individual show, the individual creator and, increasingly, the individual AI answer. Whether a survey follow-up can deliver that granularity reliably, at scale and with the accuracy the 50% figure implies, is something the beta period will have to demonstrate.

Timeline

Summary

Who: Fairing, a zero-party data measurement and attribution platform based in New York and London, led by co-founder and chief executive Matt Bahr, with Podscale cited as an early user. The product is aimed at brands running podcast, TV, creator and AI-influenced marketing.

What: Advanced Attribution, an extension of Fairing's "How did you hear about us?" post-purchase survey that asks a tailored follow-up question to identify the specific podcast, creator, TV placement or AI query behind a purchase, plus managed auto-suggest lists that structure free-text answers. Fairing says early rollouts produced up to 50% more usable attribution signal from the same order volume, a figure not independently verified.

When: Announced today, September 15, 2026. The release states the beta began on September 10, 2026, and general availability is planned for January 2027.

Where: Issued from New York and London, deployed on brands' post-purchase pages. The release does not specify supported markets or e-commerce platforms.

Why: Pixel-based and modelled attribution credit the last click, usually search or direct traffic, leaving podcasts, creators, TV and AI assistants under-measured even as spending on those channels grows. Survey data offers customer-declared evidence that does not depend on clicks or cookies, though it carries recall and coverage limitations.