Reebok Teamsports: account usage & quality review
User: Josh Chapman (reebok-teamsports.com) · customer since 7 Nov 2025 · data as of 20 Oct 2026, 08:10 UTC · source: Modelia production DB + Stripe
Success rate, last 14 days
99.47%
1,133 requests · 6 failed (0.53%)
Requests this month (1–20 Oct)
1,440
2.3× all of September (613)
Images delivered, lifetime
3,844
1,988 in the last 30 days
Paid to date (Stripe)
$1,419
$900 of it since 10 Sep (6 × $150 packs)
1 · Generation failures, last 14 days (daily)
| Date | Requests | Succeeded | Failed | Failure % |
|---|---|---|---|---|
| Tue 6 Oct | 101 | 101 | 0 | 0.00% |
| Wed 7 Oct | 111 | 111 | 0 | 0.00% |
| Thu 8 Oct | 100 | 100 | 0 | 0.00% |
| Fri 9 Oct | 20 | 19 | 1 | 5.00% |
| Sat 10 Oct | 0 | – | – | – |
| Sun 11 Oct | 0 | – | – | – |
| Mon 12 Oct | 115 | 115 | 0 | 0.00% |
| Tue 13 Oct | 148 | 148 | 0 | 0.00% |
| Wed 14 Oct | 128 | 128 | 0 | 0.00% |
| Thu 15 Oct | 9 | 9 | 0 | 0.00% |
| Fri 16 Oct | 120 | 117 | 3 | 2.50% |
| Sat 17 Oct | 0 | – | – | – |
| Sun 18 Oct | 0 | – | – | – |
| Mon 19 Oct | 237 | 235 | 2 | 0.84% |
| Tue 20 Oct partial | 44 | 44 | 0 | 0.00% |
| Total | 1,133 | 1,127 | 6 | 0.53% |
What failed
- All 6 failures were Flatlay → Model (6 of 766 = 0.78%). Outfit Generator, Background Remover and Model → Model had zero failures.
- Root cause in every case: the GPU job hit our 90-second dispatch timeout (5 stuck in queue, 1 mid-run; 2 of them failed again after an automatic resubmit). These were infrastructure issues, not problems with his inputs.
- Every failed request was refunded automatically: 6 × 3 credits = 18 credits in October.
Longer view
- 10 failures in total since status tracking began (Mar 2026), across 3,375 tracked requests: 0.30% lifetime.
- Typical turnaround over the last 14 days (median / p90): Flatlay → Model 40 s / 80 s · Outfit Generator 33 s / 41 s · Background Remover 8 s / 31 s.
2 · Monthly volume: growing fast
164Nov
187Dec
5Jan
108Feb
34Mar
37Apr
80May
327Jun
381Jul
486Aug
613Sep
1,440Oct*
*October covers 1–20 Oct only. At this pace the full month lands around 2,100 requests. Volume has grown every month since May.
3 · Which tools he uses
Last 14 days
| Tool | Requests | Share |
|---|---|---|
| Flatlay → Model | 766 | |
| Outfit Generator | 230 | |
| Background Remover | 136 | |
| Model → Model | 1 | <1% |
How his tool mix has changed
- Nov 2025 – Feb 2026: Flatlay → Model (on-model shots from flat product photos), plus trials of Upscaler, Mannequin, VTON and Sketch.
- Feb – Aug 2026: moved to AI Fashion Lab / Smart Styler (1,000+ runs Jun–Aug), with Outfit Generator, Hat-on-Model, Pose Changer and Background Changer.
- Sep 2026 →: back to Flatlay → Model at volume (1,253 runs), with Outfit Generator and Background Remover as standard steps. Smart Styler hasn't been used since August.
- Has tried 15+ tools overall. He hasn't used Studio 4.0 (projects/looks) yet.
4 · How he works (Flatlay → Model settings, last 60 days, 1,253 runs)
Category
Catalog
100% of runs
Aspect ratio
3:2
99.8% (21:9 twice)
Background
White
98% #FFFFFF · 2% transparent
Saved model (consistent character)
87%
reuses the same saved models across shoots
Model gender
81% F
19% male
Images per request
1
always one at a time; back views 3%
In short, this is a standardised e-commerce catalog pipeline: the same aspect ratio, a white background and the same saved models every time, with one product image after another. That's a good fit for batch/bulk upload and API automation.
5 · When he works
By weekday (last 90 days)
| Monday | 735 | |
| Tuesday | 543 | |
| Wednesday | 491 | |
| Thursday | 481 | |
| Friday | 327 | |
| Sat / Sun | 0 |
By hour, UTC (last 90 days)
060708091011121314151617
UK office hours (09:00–18:00 BST), with a lunch dip around 12:00 UTC. Weekday-only professional use.
6 · Commercial history
- Nov 2025: signed up; small monthly plan (~$30/mo).
- Jun 2026: upgraded to a 750-credit monthly plan ($85/mo). We added 100 goodwill credits in April and again in June.
- 27 Jul 2026: cancelled the subscription (reason given: "Something else", no note). It ended on 25 Aug.
- Since then: buys 1,000-credit packs at $150, with purchases speeding up: 10 Sep, 25 Sep, 5 Oct, 9 Oct, 15 Oct, 19 Oct.
- He often runs out completely before topping up (balance before purchase is typically 0–85 credits), so his work is blocked while he waits to buy.
- Credits used 1–20 Oct: 3,735 (≈ $560 at pack price), a run-rate of roughly $850/month. Current balance: 810 credits, about 4 working days at his October pace.
- He is the only user from reebok-teamsports.com, on a personal workspace with no team seats.
7 · Other Reebok accounts
None found. Josh Chapman is the only Reebok user on Modelia.
| Where we searched | Matches |
|---|---|
| User accounts (email/username containing “reebok”, “teamsport”, “rbk”) | 1 (Josh) |
| Workspaces / organisations (name, slug, owner) | 1 (Josh’s personal workspace) |
| Team invitations sent to a Reebok address | 0 |
| Credit transactions (all users) | Josh only |
| Stripe one-time purchases, last 12 months | Josh only |
| Shopify app installs (shop domain, owner email/name, store profile) | 0 |
So all Reebok usage runs through one person on a personal workspace, with no teammates invited and no Shopify store connected. The meeting is a chance to expand to more users, set up a company workspace and integrate with their store.
8 · Talking points for the meeting
Strengths to lead with
- 99.5% success over the last 14 days and 99.7% lifetime; every failure is refunded automatically.
- Flatlay → Model median ~40 s per catalog image, with consistent models across 1,000+ shots.
- Usage has grown every month since May; October is on track for about 3.5× September.
Opportunities
- Volume / annual contract: he is paying pack prices at ~$850/mo with repeated stock-outs. A committed plan with auto-top-up removes the friction.
- Team seats: one person is producing the whole catalog. A team workspace would let more Reebok Teamsports staff contribute.
- Bulk / API pipeline: his settings never change (catalog, 3:2, white, saved model), so the workflow can be automated.
- Re-engage on the July cancellation, whose reason was not recorded, to understand what he needs.
Risks to have answers for
- The 6 GPU timeouts this month were all on the main tool he uses. Engineering is tracking the 90-second dispatch timeout.
- p90 for Flatlay → Model is ~80 s, so at higher volume, throughput guarantees will matter.