Written by the people who did the work.

Every volume starts with work we actually did. We publish the method, name the author, separate measurement from interpretation, and cite the source behind every market figure.

Original volumes, each with a named author

393%

Growth in AI-referred traffic year over year, per Adobe Digital Insights, 2026

Findings we did not source or run ourselves

Flagship research · four volumes

Four volumes.None of them is a content calendar.

One question runs through the series: when an answer engine is the first thing your buyer opens, what decides whether you are in the answer?

eCommerce

Volume 01Pyxl AI Visibility Reports

The AI Visibility Report

The State of eCommerceAI Visibility, 2026

42%

Better conversion from AI-referred traffic

Bonnie Winter, Co-CEO · May 2026

For the first time, traffic that arrives at an online store from an AI assistant converts better than traffic from any other channel. The volume is still small. The trajectory is not.

393%

Year-over-year growth in AI-referred traffic to U.S. retail sites in Q1 2026

Adobe Digital Insights, 2026

2–7

Domains an answer engine typically cites in one response, against ten blue links

Profound, 2025

66%

Machine-readability score Adobe assigned to retail product pages, the lowest of any page type

Adobe Digital Insights, 2026

Read the Report14 minutes · no form

eCommerce

Volume 03Pyxl AI Visibility Reports

The Cited Catalog

The product shelf movedinside the answer engines

Own it

The layer is the asset, not the subscription

Bonnie Winter, Co-CEO · June 2026

A point of view rather than a data report. Why the catalog has become the asset, what it takes for an answer engine to cite it, and why the brands that win will own that layer instead of renting it.

2028

When AI-driven search is projected to overtake traditional search

Gartner and others, search-decline projections

Two

Readers on every product page now: the shopper, and the model recommending on their behalf

Pyxl · AIRO Commerce

Rent

What a subscription buys: someone else's model, undifferentiated output, your data on their servers

Pyxl · AIRO Commerce

Read the Field Brief10 minutes · no form

B2B

Volume 02Pyxl AI Visibility Reports

The Day One Shortlist Report

The State of B2BAI Visibility, 2026

94%

Of B2B buyers used AI in their last purchase

Brian Winter, Co-CEO · August 2026

The shortlist is now assembled before a vendor knows it is in play, inside a system with no form fill, no session and no attribution. Ninety-five percent of the time, that list is the deal.

95%

Of the time, the winning vendor was already on the buyer's Day One shortlist

6sense Buyer Experience Report, 2025

47%

Build the internal business case inside an AI tool before contacting any vendor

Forrester Buyers' Journey Survey, 2026

10–40%

Website traffic decline as vendor research migrates into answer engines

Forrester, February 2026

Read the Report11 minutes · no form

Field brief · Volume 04Outside the visibility series. Applies to both motions.

Most companies find outby demand letter.

Two claim types are being filed against websites at volume: accessibility suits under the ADA, and privacy suits over the analytics and advertising tags in your tag manager. Both are templated, both are decided by code you already shipped, and they reach commerce and B2B differently. The brief covers what is being filed, what one case costs end to end, why overlays and automated scans do not close it, and the ninety-day sequence that does.

Read the Field Brief Or start with what is happening to you ⟶13 minutes · no form

Bonnie Winter, Co-CEO · August 2026. Pyxl is not a law firm and this is not legal advice.

3,117

Federal website accessibility suits filed in 2025, up 27% over 2024

UsableNet, 2026

44%

Of Q1 2026 filings named a Shopify-based site

EcomBack, Q1 2026

$5,000

Statutory damages per violation under California's wiretapping statute, with no proof of harm required

California Invasion of Privacy Act

$55K–$270K+

All-in exposure for a single case, settlement through monitoring

Published settlement data and legal industry reporting

From the report

Four camps,four gaps.

Most firms working in this space sit in one of four camps, and every camp leaves something out. Each camp’s gap is another camp’s day job, which is what makes the handoffs between them expensive.

The campUnderstands citationCan engineer itCan write and design itCan run the channels

Visibility boutiques

Understand citation. Cannot execute the commerce work that makes it pay.

Engineering shops

Can build. Cannot write or design.

Large consultancies

Move too slowly and too generically for a mid-market brand.

Performance agencies

Run the channels well. Have not built the AI-visibility layer.

Pyxl

All four, held by the same team, so the depths compound.

Our own read of the category, not third-party research. The four camps are described in Volume 01; the marks are our assessment and you should discount them accordingly.

Keep going

Research is one thing.Delivery is another.

B2BHow B2B deals actually get approvedWhat each stakeholder needs to say yes, why deals stall, and what we build to move them forward.See it ⟶

The client indexEvery client we can nameAll 711, each row carrying the industry and what we actually did. Filter it by your industry and by what you need built.See it ⟶

Before a contractSeven days, one working versionA senior strategist builds a working version of your business in seven days, yours whatever happens next.See it ⟶

Common questions

Questions we getbefore the first call.

Do I have to fill in a form to read the research?

No. All four volumes are published in full on this site with no gate, no email capture and no sequence afterwards. If an argument only holds behind a form, it is marketing rather than research.

How is the research actually produced?

Each volume comes out of client engagements rather than a content calendar. It carries a named author, publishes its method so you can repeat it, and separates measurement from interpretation — every market figure is cited to the third party that produced it, and our reading of that figure is labeled as our reading.

Which volume should I read first?

If you sell to a committee, start with Volume 02, The Day One Shortlist. If you sell a catalog, start with Volume 03, The Cited Catalog. If your concern is legal exposure in the code you already shipped, Volume 04, The Litigation Surface. Volume 01 is the market overview underneath all three.

Whose numbers are these?

Third-party research from Adobe, Forrester, Gartner, 6sense, McKinsey and Profound, each named and dated at the point of use, with the full source list at the end of every volume. Where a figure is ours, it says so and explains how it was measured.

What research does Pyxl publish?

Four volumes. Volume 01, The State of eCommerce AI Visibility 2026, by Bonnie Winter. Volume 02, The Day One Shortlist, on how B2B buying committees build a vendor list inside an assistant. Volume 03, The Cited Catalog, on the catalog layer that decides whether SKUs get named. Volume 04, The Litigation Surface, on accessibility and tracking-privacy exposure. Each names its author, publishes its method, and cites every market figure to the third party that produced it.

Reading is one thing.Seeing it built is another.

Eighteen years as an award-winning agency across brand, design and technology. We work as an extension of your team to turn any of the arguments above into a working version of your own business.

Start a Conversation