Week 16, in review

Store Performance Analysis: Traffic Growth and Conversion Priority

Date: July 22, 2026

Scope: Shopify store performance since engagement start, funnel diagnosis, and next-phase priorities

1. Summary

Traffic acquisition work has succeeded. Session volume is up more than 4x year over year and gross sales are up 84% over the trailing year. Absolute order volume has increased in every comparison window.

Conversion rate has declined. The decline is not evidence of a store that has always converted poorly. Rolling monthly data shows add-to-cart rate held between 5.8% and 7.3% from September 2025 through May 2026, inside normal ecommerce benchmarks. The collapse is approximately sixty days old and has two distinct components.

The store's conversion problem is a mismatch between a much larger, less pre-qualified audience and product pages that were built for a small, high-intent one. Product imagery is the primary point of failure, and the failure is specific and documentable rather than aesthetic.

2. Revenue

2026 is tracking at roughly double the 2025 full-year pace within five months.

3. Funnel by Rolling 30-Day Window

Two separate events are visible.

Event one, May 22 - Jun 21. Add-to-cart fell from 41 to 24 in absolute terms while sessions rose 19%. Completed checkouts more than halved, 15 to 7. Dilution does not produce an absolute decline at that traffic level. Something reduced purchase behavior directly. Product description revisions shipped during this window and are a candidate cause. Inventory conditions are another (see section 5).

Event two, Jun 22 - Jul 21. Sessions rose from 704 to 3,148 while add-to-cart rose from 24 to 29. The incremental traffic, 2,444 additional sessions, produced approximately 5 additional carts, an incremental add-to-cart rate near 0.20%. The new traffic is functionally non-shopping at current page quality.

4. Attribution, Last 90 Days

Organic Google accounts for approximately 98% of non-direct sessions and drives the aggregate rate. Smaller channels continue to convert normally, which indicates the store is capable of converting appropriate traffic.

5. Inventory

Sell-through data shows Physics Flashcards at 100% sell-through with zero ending inventory on both the physical and the digital download variant.

A digital download at zero available inventory is a configuration error, not a stock condition, and represents unrecoverable lost revenue for every visitor who reached that page while it was unavailable. This should be verified and corrected before any other work.

Physics is also one of the highest-review products in the catalog, which makes an outage on that SKU disproportionately costly and a plausible contributor to the May-June decline.

6. Diagnosis: Audience Composition, Not Store Regression

The imagery on the site during the 6-7% add-to-cart months is largely the same imagery on the site today, which at first appears to rule imagery out as a cause. It does not. Whether an asset blocks conversion depends on the interaction between audience and asset, not on the asset alone.

The earlier audience was small and arrived with high intent, largely students already searching for these specific products under exam pressure. That audience tolerates weak product imagery because it has already decided to buy.

The current audience is five times larger and arrives from broad organic search. It contains a far higher proportion of visitors who are evaluating, comparing against alternatives, and who apply the visual scrutiny habits developed across other ecommerce sites. The same assets that were adequate for a decided buyer are disqualifying for an evaluating one.

Behavioral session recordings confirm the pattern on current traffic: visitors swipe past the styled opening images, seek an image showing the actual product, attempt to zoom, fail to resolve usable detail, briefly scan the description, and move to reviews. Reviews are the fallback when imagery has failed to answer the question.

Published research supports the mechanism. A May 2026 industry survey found 72% of consumers expressed concern over AI-generated shopping content, 31% said visibly AI-generated marketing reduces brand trust, and 63% flagged inconsistent AI imagery specifically as a sign of an unreliable seller. As of July 2026, 61% report trusting brands less after learning imagery was AI-generated, with sentiment researchers describing the effect as structural rather than transient.

A shopper study adds the operative nuance: 71% could not distinguish AI from real imagery at a glance, but reliably noticed inaccuracies in detail. Trust tracked realism and detail accuracy rather than provenance. The conclusion is that the problem is not that imagery is AI-generated. The problem is imagery that fails inspection.

Separately, business legitimacy is now the leading cause of cart abandonment, and 82% of shoppers report being more likely to trust a site displaying a third-party verified trust badge, with measured A/B conversion lifts of 4.7% and 19.7% in two published retailer tests.

7. Specific Defects

7.1 Typographical errors in packaging artwork

The Abdomen Flashcards packaging contains two misspellings, and one of them is in the name of the target credentialing exam.

Both originate in the source artwork and therefore appear on the printed box, the title card inside it, and every listing image showing that artwork.

On a product sold to a credential-seeking audience, a visible spelling error in the exam name works directly against the legitimacy signal identified above as the leading cause of cart abandonment.

This product line is being retired and redesigned, which resolves the issue at source. Remaining action is limited to removing or replacing the affected listing images.

7.2 Zoom does not resolve

Session recordings show repeated failed zoom attempts. Where images do not contain sufficient detail, zoom functionality cannot compensate. Both the image source resolution and the zoom implementation require attention.

8. What Product Imagery Must Contain

A reference composition was generated using only the existing product description as input. Its construction indicates the elements a product image can carry, several of which are absent from the images reviewed.

Elements that should be present:

• Readable product content. Actual card or page content at legible size.
• Format and scale. Physical cards and packaging shown together.
• Specification overlay. Deck count, card count, and format availability stated visually.
• Differentiating content. Ultrasound imagery on cards, which distinguishes these from generic anatomy flashcards.
• Environmental context. Desk staging that reads as a real study setting.

All specification data required for the overlay already exists in the product description. The eight decks, the per-deck question counts, the physical and digital availability, and the ARDMS exam alignment are all documented in copy on the same page as the images that fail to convey them.

The specification data required for effective product imagery already exists in the product copy. Bringing the imagery in line with it requires no new information, only a brief drawn from what is already written.

8.1 Constraints on generated imagery

Generated imagery is acceptable for staging, lighting, composition, and background. It is not acceptable for anything a buyer would treat as a product fact.

Product facts that must be depicted accurately include packaging, card layout, deck structure, and stated counts. Any generated composition should be checked against actual product photography before publication.

Recommended production method: photograph the real cards and the real packaging, then use current-generation tooling for compositing, lighting, background, and overlay. This achieves reference-level quality with nothing invented.

9. Opportunity Sizing

Modeling add-to-cart rate improvements at current traffic and $72.40 AOV, holding downstream ratios constant:

The 6% row is not hypothetical. It is the level this store sustained for eight months on the same product catalog, prior to the traffic composition shift. Returning to it requires no additional acquisition work.

10. Measurement

Track add-to-cart rate as the primary KPI for this phase.

Conversion rate is confounded by traffic volume. Continued acquisition success will continue to suppress it through denominator growth, producing a misleading signal and understating the effect of conversion work.

Secondary metrics: absolute completed checkouts, cart-to-checkout rate, checkout completion rate.

Baseline for measuring the July 20 homepage rebuild: add-to-cart rate over the 30 days following deployment, compared against the 0.9% July figure.

11. Data Notes and Caveats

• Reached checkout exceeds added to cart in several windows. This is consistent with express checkout paths that bypass the cart. Cart-based ratios should be read as approximate.
• Google Search attribution is incomplete prior to search console installation. Channel-level historical comparison is limited to the post-installation period.
• AOV varies by channel from $27.00 to $138.00. Projections use $72.40, the Google organic figure, as the conservative anchor given that channel's share of volume.
• The June 2025 revenue spike is an outlier against an otherwise flat 2025 baseline. Single-month year-over-year comparisons against it overstate any decline.
• Sell-through comparison windows in the source reports are mismatched in length. The stockout finding is reliable; the percentage changes on that report are not directly comparable.
• Benchmark add-to-cart ranges are general ecommerce figures and vary by category and price point. The store's own eight-month 5.8-7.3% history is the more relevant benchmark.
• The relative contribution of imagery, description revisions, and the Physics stockout to the May-June decline has not been isolated. All three are plausible; the stockout is the cheapest to verify.
• Imagery findings in section 7 are drawn from a partial sample of listing images rather than a complete catalog audit. No conclusion about coverage across the full image set is claimed.
• Traffic intent composition for the June-July session increase has not been characterized. Landing page and query analysis would establish what proportion of the new traffic has commercial intent and would refine section 9 materially.

12. Recommended Sequence

1. Verify and correct Physics Flashcards inventory, particularly the digital download variant. Lowest cost, immediate revenue recovery.
2. Remove or replace listing images carrying the misspelled artwork. The product line is already scheduled for redesign, so no reprint is required.
3. Audit the full image set across all listings before committing to imagery work. Findings to date are drawn from a partial sample and are not sufficient to scope a catalog-wide treatment.
4. Add specification overlays and physical-product context where absent, built from existing product description data.
5. Repair zoom at both source resolution and implementation level.
6. Measure homepage rebuild against add-to-cart rate at the 30-day mark.
7. Audit landing page query intent for the June-July traffic cohort to validate opportunity sizing.
8. Implement trust signal elements, including third-party verification badging.