2026-05-27 · hand-built in BrightSoft tokens · ICP = Ryan · Power BI has no dev tools, so we built one
What changed from v1
The first round used gpt-image-2 to render the report card. It looked AI-generated. So this round rebuilt the mockups in actual HTML/CSS using the real Nexlytics design tokens — Inter for headings, JetBrains Mono for code fragments, semantic severity colors, BrightSoft cobalt + PBI yellow. Pixel-crisp on retina.
Each mockup is a real 1:1 square (what Meta will see). PNG downloads below each are 1080×1080 — Meta-ready. Tap on mobile to save to camera roll.
Files are Onwards-coded: tier-one metals + mining + the kind of file Ryan would actually open at 6:47am on a Wednesday.
A · my pick
Verdict Stamp
Single-pane stop-the-scroll. Giant amber grade + score on the left, blood-red issue count + verdict pill on the right, three real DAX findings below. Reads in 0.5 seconds. Maximum punch.
Nexlytics
QLD_Iron_Margin_FY26.pbix · 8.4 MB
D
38/ 100
47 issues6 critical · 14 high · 27 medium
Do not ship
Critical findings
Net_Revenue divides by Gross, not Net. Margin shows 4-7% low on 11 downstream visuals.
RLS role SalesRep filters by USERPRINCIPALNAME. Non-AU viewers see zero data on 6 pages.
DIVIDE() called without third argument on 14 measures. Wrong-number risk on refresh.
The inheritor frame, dialled all the way up. Shows score CRASHING (67 → 38) since the offshore team's last sprint. Three findings flagged as NEW since prior scan. Sells the Diff View feature + the offshore-handoff anxiety in one image.
scan/Tailings_FY26.pbix
Scan #4 · vs. scan #3 · 8 days ago
38Health
Validate the numbers and budget significant cleanup.
67 → 38 since last scan
New since scan #3
NEW
Net_Revenue measure now divides by Gross, not Net. Margin will appear 4-7% low on 11 downstream visuals.
NEW
RLS role SalesRep filter changed. Non-AU viewers see zero data on 6 pages.
NEW
Calendar table dropped. 9 measures depend on it. Refresh will fail on next attempt.
Full product-screen feel. Dark top bar, big amber score card, all five dimensions across one row, triage table beneath. Densest take, looks most like a real app screenshot. Closest to "this is what you get for $9".
Nexlytics
QLD_Iron_Margin_FY26.pbix · Scan #4 · 27 May
VerdictValidate the numbers and budget significant cleanup. 6 critical findings block sign-off.Do not ship
D38 / 100
DAX
22/100
Gov
41/100
Perf
58/100
Lineage
34/100
Comp
71/100
Top 3 to fix first6 critical · 14 high
Critical
Net_Revenue divides by Gross, not Net. Margin shows 4-7% low on 11 downstream visuals.
Critical
RLS role SalesRep filters by USERPRINCIPALNAME. Non-AU viewers see zero data on 6 pages.
High
DIVIDE() called without third argument on 14 measures. Wrong-number risk on refresh.
Inherited a .pbix file? Get the verdict in 90 seconds.
Primary text
If you've signed off on offshore .pbix files this week, this is for you.
You open the file. You have no idea what it is or what it does. The last person built it under nonsense deadlines. Now it's yours to sign off on.
You can do the read by hand. Most of the time you do. It just takes forever.
I built Nexlytics because I was losing a day a week reading offshore output before it could ship to clients. Five AI agents check DAX, governance, performance, lineage, and compliance, the same lenses a senior consultant would run.
90 seconds per file. $9 for 7 days. Then $59/mo. Cancel anytime.
It's what Onwards uses on every PBI engagement. Every Monday.
A · Verdict Stamp is my pick for cold-feed smoke test. Giant amber D + blood-red issue count + "Do not ship" pill carries the stop-the-scroll. Three findings do the demo work below. Most thumbnail-friendly.
B · Diff Lens if we want to lead with the inheritor / offshore-handoff frame. The 67→38 delta + three NEW pills make the offshore-team-broke-things story land in one image. This is the one that pairs hardest with the ad copy's "last person built it under nonsense deadlines" line.
C · Triage Stack if we want maximum "this is what $9 gets you". Densest, most product-y. Best for retargeting — too information-dense for cold feed but lethal for warm.
Recommendation: run A cold ($30/day × 3 days, Lane 1). If CTR ≥ 3% and LPV ≤ $0.50 → graduate to the locked $1k B-vs-E offer test with A + B + C as the 3-image rotation per variant.