01
No invented icon mark
The wordmark stands on its own. I would not try to improvise a symbol unless it is genuinely strong and already brand-validated.
Separate page for comparison
This one should feel less synthetic.
The last concept was too polished in the wrong way. This version is intentionally less app-like and less “design system demo.” The aim is a public-record page with some bite: heavier rules, less generic color, no fake startup logo, and a stronger feeling that you are reading a file rather than browsing a product site.
If LookCloser says “check me,” the design should feel more like margin notes, ruled paper, and evidence sorting than glassmorphism, floating cards, or default AI minimalism.Design position
What I changed from the first pass
I leaned away from “tasteful neutral SaaS” and toward something more specific: editorial, slightly forensic, and a little rougher around the edges in a good way.
02
Color with more character
Instead of generic beige-plus-teal, this palette adds oxide, brass, and moss accents so the page feels more like a printed dossier than a generic “clean UI.”
03
Stronger ruled structure
More lines, more section breaks, fewer floating boxes. The page should feel organized by evidence, not decorated by components.
What we check
These read more like evidence categories than feature cards
The content is similar, but the framing is more matter-of-fact. I’m trying to make the site feel like it documents a method rather than advertises capabilities.
Label math
Does the claimed dose fit the serving?
Arithmetic is the first gate because sometimes it is enough on its own.
Dose vs. science
Is the amount on the bottle inside the range actually studied?
The product and the literature need to be speaking about the same amount.
Delivery route
Was the ingredient studied in the same form this listing sells?
Support should not drift casually across powders, gummies, or protected delivery.
Certifications
Do the badges appear in a real registry?
Registry checks should feel ordinary, because they should be ordinary.
Origin & sourcing
What exactly is being claimed about where it was made?
Final bottling and ingredient origin are not the same thing.
Review forensics
Do the reviews move like demand, or like intervention?
Burst timing and cluster overlap belong in the same record as label claims.
Sample result page
This is where the visual language matters most
Homepages can be debated forever. The real test is whether a shopper can land on a record and understand the answer quickly, without losing the nuance that makes the product trustworthy.
Public record · 850037549016 · methodology v0.1
Lemme Sleep Gummies with 5mg Melatonin, Elderberry
Moderate public verifiability
The short read
The core melatonin amount appears internally consistent and publicly checkable. The main limit is not contradiction but opacity: a proprietary blend hides the amounts of some named ingredients, so the public cannot tell whether each one reaches a studied level. The page should state that plainly and early.
Trust signals found
Label mathFound
Barcode confirmationFound
Supplement Facts visibleFound
NIH label filingNot found
Brand storeNot found
Named assayNot found
What checked out
Give the shopper stable ground first
These are the parts of the record that held up under the checks that ran. They are framed positively without drifting into endorsement.
✓
The dose fits the serving mass
The claimed 5 mg of melatonin fits easily inside the estimated serving size, so the label is physically possible.
Confirmed
✓
The visible label agrees with the listing on the main melatonin amount
The listing and the photo-read panel both show 5 mg of melatonin per serving.
Confirmed
✓
The barcode resolves outside the listing itself
Open Food Facts independently resolves the UPC to Lemme Sleep, which supports identity rather than quality.
Identity
What needs caution
Separate concern from contradiction
This part matters because it lets the product be serious without becoming melodramatic.
!
Ingredient amounts are hidden inside a proprietary blend
The blend total prevents a shopper from checking whether each named ingredient reaches a studied amount.
Caution
!
Magnesium appears below the range shown in the literature section
This does not prove failure. It means the label and the study range are not aligned on amount.
Context gap
What is still missing
Missing evidence should look different from a negative finding
This is one of LookCloser’s clearest doctrines, so it should also be one of its clearest interface patterns.
…
No NIH DSLD label filing located
This is a public-record gap, not a strike against the product.
No data
…
Review authenticity was not yet analyzed
The rating count exists, but the temporal evidence needed for burst analysis is not present in the current record.
Pending
What to ask and check
Turn uncertainty into next steps
This is how the product becomes useful to a human without drifting into advice.
Ask the brand or seller
What is the amount of each ingredient inside the blend?
A blend total cannot tell a shopper whether each ingredient is present at a meaningful amount.
Is this exact product filed with NIH’s Dietary Supplement Label Database?
A public filing gives everyone the same neutral reference point.
Does NSF certify this exact listing?
Brand-level presence is useful, but product-level confirmation is the real question.
Check for yourself
Compare the real bottle to the label shown in the record
The bottle in hand is the source of truth if the listing images and physical label differ.
Scan the barcode and confirm it matches 850037549016
A mismatch can point to a different variant, repack, or listing confusion.
Look the product up directly in the NSF registry
The page can guide the search, but the user should always be able to reproduce it independently.
Claim-by-claim
Keep the table, but after the synthesis
Rows are important. They just should not be the first thing a stressed shopper has to decode.
| Claim | Status | What the public record says |
|---|---|---|
| 5 mg melatonin per serving | Confirmed | The listing amount matches the visible panel and fits the serving mass. |
| Herbal ingredients support sleep | Not fully checkable | The blend hides the amount of each named ingredient, so the record cannot compare each one to a studied dose. |
| Brand holds NSF certification | Partial support | The brand appears in the NSF registry, but the current record does not confirm this exact listing. |
| Made in the USA in a cGMP facility | Claimed, not settled | The public record here is incomplete, and final bottling is not the same thing as ingredient origin. |