test
CRO Testing Concepts
June
2026
Three concepts for June. PDP tests DC03 and DC04 are queued and waiting on Intelligems billing, so June moves off the product page and onto three surfaces with clear room to work. Each concept below is grounded in the Fairing post-purchase survey, the May manual survey, Shopify orders, GA4 traffic, and Doich's own testing history. Approve, push back, or pick a subset.
1
Why we're testing
Today the cart drawer is a line item and a Check Out button. Everything in between is white space. We use that space to do three things: show how close the buyer is to free shipping, point them at the next pack tier with the live Buy 3 Get 1 Free promo, and add small trust signals at the decision point. Three additive moves on a surface that currently sells nothing.
Concept illustration only. Final design will differ. Threshold matches the current $45 announcement bar; review count and badge wording confirmed before launch.
Evidence
| Signal | Source |
|---|---|
| Shipping is the most common reason buyers say they almost didn't buy, cited equally by first-time and repeat buyers. The majority of orders also fall below the $45 free-shipping line, so the nudge-able cohort is the larger half of the distribution. | Fairing post-purchase survey · 90 days · "What almost stopped you?" |
| Doich already runs Buy 3 Get 1 Free, Buy 4 Get 2, and Buy 5 Get 3 in the announcement bar. The cart drawer makes no reference to that promo. A second cluster of buyers describes friction finding or applying advertised discounts. | doichfoods.com announcement bar · observed 2026-05-26 · Fairing 90d |
| A separate cluster of buyers describes trust hesitation at checkout (scam/fraud worry, not enough reviews surfaced, missing nutrition info). Small trust signals at the cart resolve this without changing the buy box. | Fairing post-purchase survey · 90 days |
Priority Reasoning
The cart drawer is currently empty real estate at the highest-intent moment on the site. The three additions are independent levers, so the wireframe stage can decide which combination becomes V1 versus V2.
2
Why we're testing
The homepage hero subhead today reads "Designed to crave, built to snack" — a generic indulgence line. The dietary claim that actually does the persuasion work ("Low sugar. 3x more protein than the leading cookie dough.") sits below the fold under the flavor block. Search-acquired buyers reorder roughly 1.6 times more often than social-acquired buyers, and they over-index on dietary vocabulary. We move the macro claim up into the hero and add a small allergen / star-rating strip so first-time visitors can verify above the fold.
Concept illustration only. Final design will differ. The macro / dietary line and the review count match copy that already appears elsewhere on the site; review count to be verified against Judge.me before launch.
Evidence
| Signal | Source |
|---|---|
| Search-acquired buyers reorder multiple times at a noticeably higher rate than social-acquired buyers, and they over-index on dietary vocabulary. The dietary cohort is Doich's highest-LTV cohort and the homepage hero is the first thing they see. | Manual customer survey · May 2026 · Q1 / Q2 cross-tab |
| A cluster of buyers describes trust hesitation at the buy decision (scam fear, fraud worry, "not seeing enough reviews", missing nutrition info). The current homepage above the fold has no review count, no allergen badges, no press signal. | Fairing post-purchase survey · 90 days |
| Search demand for "high protein cookie dough" is up sharply year-over-year with low competition. The macro claim is the term that's growing, so leading with it on the homepage hero matches the rising search intent. | Keyword research · Slack 2026-05-19 |
Priority Reasoning
Copy-and-badge level change on the highest-impression surface. No new feature build. The macro line and the allergen badges already exist on the PDP and the product tiles; we are surfacing what's already true.
3
Why we're testing
The ambassador apply page today is a multi-step form. One question per screen, no statement of what an ambassador gets, no estimated time, no preview of the reward. Meanwhile the AMBASS20 code carries roughly a third of all orders, so the program already works once people are in it. We replace the form with one screen that leads with the reward, shows the full form, and issues the code on submit. The primary metric is form-fill (apply) rate, not purchases — this page exists to recruit ambassadors, not to sell directly.
Concept illustration only. Final design will differ. Reward copy and ambassador-approval policy confirmed with the team before launch.
Evidence
| Signal | Source |
|---|---|
| 297 sessions landed on the ambassador apply page in 7 days (255 mobile, 42 tablet, zero desktop) — roughly 1,200 sessions per 30 days at run rate. Form-fill rate is not currently tracked, so the test will instrument a submit event and measure single-screen apply against the existing multi-step form as the primary metric. | GA4 landings by device · 7 days ending 2026-05-21 |
| The AMBASS20 code appears on roughly a third of all orders. The program already drives meaningful volume; the bottleneck is the page where new applicants land. | Shopify orders · 60 days · May 2026 |
| Buyers who self-identify as ambassadors repeat at a noticeably higher rate than the general population. Each apply-page conversion compounds. | Fairing post-purchase survey · 90 days · ambassador-language segment |
Priority Reasoning
The current page is a Tally / TypeForm embed with no value proposition shown up front. Replacing it with a single-screen apply that leads with the reward is a low-effort, high-leverage move. Primary metric is apply-form completion rate; downstream revenue is a secondary, lagged read given the ambassador cohort's higher repeat rate. Instrumentation prerequisite: the test cannot read without a custom event firing on form submit (and ideally on form start). Wire this into the same dev ticket as the page rebuild — without it, both variants are unmeasurable.
Engineering Fix Flag · Not a Test
→
Why we're flagging this
A meaningful slice of buyers describe friction finding, seeing, or applying a discount code. Three failure modes cluster: advertised codes don't render at checkout, SMS / text codes glitch and force a cart restart, and buyers leave the site to hunt for codes. Independently, the orders data confirms the discount layer is load-bearing — well over half of orders carry a code. This is a bug-fix flag, not a CRO test. Fix the underlying issue first; any framing experiment afterward will read cleaner.
"No discount code as advertised."Fairing post-purchase survey · 90 days
"The text code glitched had to back out to get new code."Fairing post-purchase survey · 90 days
"Needed to find promo codes and search for membership program etc."Fairing post-purchase survey · 90 days
"Confusion over discounts."Fairing post-purchase survey · 90 days
Recurring theme
Evidence
| Signal | Source |
|---|---|
| A meaningful slice of post-purchase responses describe code-related friction. Three failure modes recur: advertised codes don't render, SMS / text codes glitch, buyers leave the site to find codes. | Fairing post-purchase survey · 90 days |
| Well over half of orders in the trailing 60 days carry a discount code. The discount layer is load-bearing; failures in it leak revenue directly. | Shopify orders · 60 days · May 2026 |
| The fix is independent of any framing test. Repairing the layer first means any later test on discount visibility or code placement reads cleanly, not confounded by underlying failures. | Engineering scope recommendation |
Recommended Action
Hand to Doich's dev team. Audit the three failure modes against Shopify's discount-application logic and the SMS / popup code-issuance flow. No A/B test needed — the codes are advertised; they should work. We can revisit a discount-visibility test once the underlying layer is stable.