Traffic came back in August — 83.4k sessions and $58.6k revenue, up from July's floor of 72.5k and $48.7k — but conversion did not come back with it (0.30%, still under April's 0.37%). The reason is in the mix: the returning traffic is mobile, which is now the majority of sessions and converts at 0.22% against desktop's 0.39%. The upstream leak is unchanged — email billed $10.8k in February and $0 in August.
Revenue, Sessions & Conversion Rate by Month
Bars are monthly revenue; lines track sessions and conversion rate on their own axes. August is complete bar the last few hours of the 31st. January is excluded — ecommerce tracking was still being set up (0.07% CR, 34% engagement), so it is not a fair baseline.
Why revenue softened
Traffic recovered; revenue only partly followed
Sessions fell from ~88k (Mar–Apr) to a 72.5k floor in July, then rebounded to 83.4k in August. Revenue followed it up ($48.7k → $58.6k) but not all the way back, because conversion rate slipped to 0.30% — the recovered traffic converts worse than the traffic that left.
The email channel went quiet
Email drove 3,830 sessions and ~$10.8k in February and has since fallen to a couple hundred sessions and near-zero revenue. This is the single largest recoverable line and it sits entirely upstream of the website.
Direct & returning demand is eroding
Direct sessions are down ~11% since February (40.6k → 35.9k) after clawing back some of July's low. Returning visitors still convert ~5× better than new ones — 1.03% vs 0.21% in August — so as repeat traffic thins (partly the email effect), revenue feels it out of proportion to the session count.
Mobile is now the majority — and the soft spot
Mobile overtook desktop in July and widened the gap in August (43.4k vs 37.0k sessions). Desktop conversion climbed over the window to 0.39%; mobile peaked at 0.42% in April and has fallen to 0.22%. The store's biggest audience is its worst-converting one — which is exactly what Exp-10 tested.
Channel Sessions by Month
Where the traffic went. Watch the Email row collapse.
Conversion Rate by Device
Desktop rising to 0.39%; mobile down from its 0.42% April peak to 0.22% — and mobile is now the larger audience.
New vs Returning — Conversion Rate
Returning visitors convert far better; protecting repeat traffic protects revenue.
Revenue by Product (program to date)
BeatBuddy 2 is the revenue engine; the test program is concentrated where the money is. GA4 lists several products under two names (e.g. "BeatBuddy 2" and "BeatBuddy 2 - US") after a Shopify rename — they share one item ID, so they are counted once here.
| Product | Revenue | Units |
|---|
Product Sales, Month by Month
Each column is one month's item revenue split by product, so the monthly total and the mix that produced it read off the same bar. Hover for units. The top six products are named; everything else is folded into Other. August's shape is different: BeatBuddy 2 is still soft at $24.5k against its $38–43k spring, and the month was carried by Aeros Gold Edition's best month of the window ($17.0k, 27 units).
Show the numbers
Device Split by Product
Bar length is the product's item revenue to date; the split is the device that earned it, and the figure on the end is mobile's share. This is where the mobile conversion gap turns into a test list: BeatBuddy MINI 2 (66% mobile), Aeros Gold Edition (56%) and BeatBuddy 2 (53%) take most of their money on the weakest-converting device. MIDI Maestro Gold Edition, at 37% mobile, is the one desktop-led product in the top six.
Show the numbers
What the testing program delivered
Nine experiments in the program's first months: five wins, one mixed, two learning losses, and one with no usable data. The job of a testing program is not to win every test — it is to compound the wins, cut the losses fast, and turn each loss into the next test. Both losses here did exactly that: Exp-7 became Exp-8, and Exp-8's mobile hold became Exp-10, which won.
Single focused CTA vs. three. Hero CTA clicks +158%, add-to-cart +43%, purchases +84%, revenue +66% — ~$3,142 incremental over 19 days. Rolled out.
Jump-links nav on the flagship PDP. BeatBuddy 2 units and revenue both +11.5%, and add-to-cart converted to a completed purchase 33% more often. Fewer, better-qualified carts. Rolled out to all BB2 traffic and now shipped in the theme.
The Exp-3 winner ported to a second product. MINI 2 units 4 → 6 and revenue +50% on 19% fewer product views, with landing-page bounce -4.1pp and engagement +4.1pp at 98% confidence. Rolled out 4 Aug and now shipped in the theme — and it won with a different audience than Exp-3 did (new mobile visitors here, returning visitors on BB2).
Direct fix for Exp-7. Ran 17 Jun – 15 Jul and won clearly on desktop: purchase rate +59%, revenue +125%. Shipped to desktop; mobile deliberately held back because new mobile visitors regressed (0.10% vs 0.40%). That hold became Exp-10 rather than a guess.
The answer to Exp-8's mobile hold, measured 1 Sep to 1 Oct on its final build. Hero taps fell 31%, yet product-page reach held and every step after it improved: add-to-cart 30 → 42, checkout 11 → 23, purchasers 9 → 16, BeatBuddy 2 units 2 → 4. New mobile visitors no longer convert worse, the regression that kept Exp-8 off mobile. A consistent directional win rather than a proven number (checkout reaches 92%). Recommended for rollout; next it isolates which proof element sold, the price line or the star rating.
Add-to-cart +11.6% overall, but only BeatBuddy 2 improved downstream — premium PDPs got worse. Kept where it helps; flagged a follow-up to isolate the two changes.
Engagement -49%, BB2 PDP sessions -37%. Too many competing CTAs diluted intent. Called in 11 days and fed directly into Exp-8 — a loss that cost little and taught a lot.
The Exp-3/5 pattern on a $379 limited edition: 1 unit sold in each group over 84 days — a page that cannot answer the question at any run length. Session revenue looked +63% until we read it at product level and found the gain was other products entirely. Reverted, and it produced the most reusable lesson of the round: read PDP tests at the item level, and only test where the product actually sells.
Where the next gains are
Reopen the email channel (highest $ / lowest effort)
Restarting lifecycle & campaign email is the fastest path back to the March–April revenue line. This is upstream of testing, but it's the clearest recoverable number on this page — worth raising with the team that owns email.
Mobile-first conversion tests
A compressed mobile hero and a sticky mobile add-to-cart bar target the one place on-site conversion is slipping. Highest-priority items in the test backlog.
Earn back returning visitors
Trust / social-proof for new visitors (92% of traffic, converting at ~1/5 the returning rate) plus stronger reasons to come back. Grows the high-value returning segment over time.
Fix the cart drop-off
Add-to-cart to begin-checkout is the biggest funnel leak site-wide. A clearer next-step and reassurance in the cart is a high-priority, high-reach test.
Full detail lives in the other tabs: Insights (live 90-day findings), Experiments (per-test readouts) and Test Backlog (the prioritised queue these recommendations come from).