Goesfor — logo candidates, top 5 of 20

The product's whole voice is the vernacular of an official statistical table: heavier top and bottom rules, a label stub, and figures right-aligned on a shared column edge so magnitude is legible at a glance — the 4-digit row breaking left is what right-alignment is for. The mark tells the product's one story without a verdict: navy rows are the bond-lodgement record, and one hueless ink row is the reader's own figure, simply placed within the distribution. Blocks are drawn on a strict mono pitch (14px glyph, 6px gap) so the mark is literally tabular-nums discipline made geometric, and the block texture survives a 24px favicon crop where real digits would not. The wordmark is plain IBM Plex Mono 600 in ink — the type system's own voice, nothing drawn.

Hue mapping:
#16191d -> wordmark, table top/bottom rules, and the reader's own figure row (the reader is deliberately hueless)
#454c56 -> stub-column label dashes (secondary text)
#123a63 -> bond-lodgement data figure rows (navy means the data and nothing else)

• A booktabs table in miniature — 3px top and bottom rules only (no vertical grid, the correct statistical-table idiom), #454c56 stub labels, right-set navy figure blocks on a 24-unit row rhythm, and the reader's row at y=84 set in hueless ink among the data — with the wordmark in Plex Mono like tabular figures. This is the register of an ABS release, not a startup. The one real weakness is scale: sixteen small rects at 24px height become undifferentiated texture, and the reader-row distinction goes first.
• A booktabs table — 3px top/bottom rules only, no vertical rules, right-aligned figure blocks all terminating at x=140, varied-width #454c56 stub labels, and the reader's row set in ink among navy data rows — is statistical-publication vernacular executed with real understanding. Row pitch is a consistent 24px, column pitch 20px, everything on whole units. Hue mapping is the system verbatim (navy=data rows, ink=reader, ink2=secondary text). The cost is scale: at 24px the table is texture, and the ink/navy row distinction dies in one-colour reproduction.
• This is a statistical publication compressed: booktabs-weight 3px rules with no verticals, right-aligned figure cells all terminating at x=140 flush with the rules, 24-unit row pitch, grey stub dashes, and the reader's row set hueless among navy data rows — every convention is the ABS's own, and Plex Mono for the wordmark completes the tabular conceit. The strongest provenance and credibility in the field. Its one real weakness is scale: fourteen 14x16 cells at 24px height become texture, so the favicon crop is busy rather than crisp.

The product's one move is putting a tenant's rent on the same scale as what comparable properties actually go for, so the mark is that scale: a y-axis with graduated ticks, drawn with the hairline discipline of a statistical release rather than a startup glyph. The navy band is the dataset occupying a range on the scale (band ends registered on major graduations, the way an interval is read off an axis), and the reader appears only as a longer ink tick — hueless per the token system, positioned inside the span without any colour or geometry that could deliver a verdict. Nothing decorates: every element is either the instrument, the data, or the reader, which is the same three-part vocabulary the app's colour tokens already enforce.

Hue mapping:
#16191d -> the scale itself (spine, graduations), the wordmark, and the reader's own figure (the long 3px tick) — the reader is deliberately hueless
#123a63 -> the span of bond-lodgement data, plotted against the scale — navy means the dataset and nothing else

• Measurement-not-judgement made literal, and it is the product's own vocabulary — RentScale is an existing release-layer component. The build is careful: 2px spine on even coordinates, majors on a 28-unit rhythm with minors at midpoints, the navy data span (y52–108) bounded by graduations, and the reader's 3px ink tick at y=85 deliberately between graduations, crossing the spine, hueless as declared. Reads as an instrument, survives one-colour ink, and the spine-plus-band crops to a favicon. Marginally penalised on craft for the reader tick's odd 85/3 placement being easy to misread as error rather than intent.
• 'Measurement, not judgement' drawn literally — and the product actually ships a RentScale component, so this is the interface's own furniture as a mark. Tick system is disciplined: majors every 28px (y=24…136) with minors exactly halfway, the navy data span (52–108) terminating precisely on major ticks, and the reader's 3px ink tick crossing the spine longer than any graduation. Hue mapping is the system's: navy only on the data span, everything the reader owns in ink. Minor ticks will mush at 24px but spine + band + reader tick survive.
• A graduated scale with the data's navy span (52-108, spanning exactly two major-tick intervals) and the reader's longer ink tick crossing the spine is measurement-not-judgement made literal — and RentScale is an actual component in this product, the rare mark that is provenance by construction. Tick hierarchy is disciplined: 28-unit majors, interleaved minors at half length. But everything is 2px stroke, which is sub-half-pixel at 24px; the graduations that make it a scale rather than a bar vanish first.

The product's one irreducible object is a distribution summarised by quartiles and a median, and a square bracket enclosing a tick is that object with everything decorative removed — it is simultaneously the typography of citation, which is the register a statistical release lives in. The median tick is the heaviest element (12u vs 8u bars, 3px vs 2px at favicon scale) because the median is what the reader came for; the bracket frames it without pointing anywhere, so nothing in the mark implies direction, growth, or verdict. Navy is used strictly in its system sense: every navy pixel in the mark depicts a statistic of the bond-lodgement data. All geometry sits on whole units in 4u multiples, so the mark stays crisp at 24px and crops to a square standalone favicon.

Hue mapping:
#123a63 -> bond-lodgement data (the bracket is the interquartile range, the tick is the median — both statistics of that data)
#16191d -> wordmark text

• The best-built mark in the set: square statistical brackets (8px spines at x=32/128, 16px arms) enclosing a 12px median tick whose centre (84) sits exactly on the bracket span's centre (84) — every rect on whole units, perfectly symmetric, one navy meaning one thing. The bracket-and-median is genuine statistical notation, not dashboard decoration, and it survives 24px, one-colour ink, and a favicon crop of the bracket alone. Only reservation: it can momentarily read as '[|]' punctuation before the IQR reading lands.
• The IQR as a square statistical bracket with a heavier median tick is a real notation, not dashboard clipart, and '[|]' is an ownable signature. Coordinates check out: bracket span 32–136 and median rect 78–90 share centre x=84; 8px stroke family consistent across verticals and serifs; all whole units. Single navy for bracket+median is defensible since both are statistics of the same data. Mark (96 tall) slightly overweights the 72px wordmark, and bold simple geometry survives favicon and ink reproduction well.
• The best-drawn mark in the set: bracket span 32-136 centres exactly on the median bar's centre at 84, serifs are a consistent 16 units, and the median's 48-112 height nests symmetrically inside the brackets' 32-128. Single navy, single meaning. Two reservations: perfect symmetry is slightly less honest than a skewed statistic, and the [|] silhouette flirts with reading as code punctuation or a pause glyph rather than an interquartile range — distinctive to a statistician, ambiguous to a landlord.

The quartile is the product's one load-bearing statistic — every cell the tenant reads is Q1/median/Q3 of real bond lodgements — so the mark is that statistic reduced to its minimum legible form: three vertical bars (Q1 6u, median 10u and tallest, Q3 6u) bridged by a navyWash band. The median sits left of the box's centre because rent distributions skew right, so the glyph is drawn the way the data actually looks, not as decoration. Everything is whole-unit rects with square corners and one non-neutral hue; at a 96×96 favicon crop the bars render at 1.5–2.5px and stay legible at 24px. The wordmark is the exact string in the serif stack at weight 600 — masthead type in the ABS/OWID register, rendered by the type system rather than drawn.

Hue mapping:
#123a63 -> bond-lodgement data (the Q1, median, Q3 bars are statistics computed from it)
#cdd9e6 -> the interquartile band of that same data (navy's wash)
#16191d -> the wordmark, neutral ink

• An honest abbreviated box plot: #cdd9e6 IQR band (x32 w96) with 6px navy Q1/Q3 posts at x=32/122 and a 10px median at x=70 — the median deliberately off-centre, which reads as real skewed data rather than symmetric decoration. All coordinates are whole units and the wash/navy pairing matches the declared data-family meaning exactly. Loses a little on distinctiveness — a range bar with end caps is close to generic chart furniture — and the median bar's centre (75) vs band centre (80) could be mistaken for sloppiness rather than intent.
• Abbreviated box-plot is genuinely this product's statistic: Q1/Q3 posts (x=32, x=122, 6px) over a #cdd9e6 IQR wash with a heavier 10px median at x=70. All whole-unit coordinates, symmetric heights (48/64/80 tall, common centre y=80). Median sits at x-centre 75 vs band centre 80 — slight, reads as skew rather than error. Box plots are common stats furniture, so distinctiveness is middling; hue use is textbook (navy=data, wash=same data at rest).
• Q1/median/Q3 bars over a #cdd9e6 IQR wash is literally the product's statistic; whisker bars overshoot the wash symmetrically (48-112 vs 56-104) and the median at x=70-80 sits honestly off-centre of the 32-128 band. Hue mapping is exact: navy carries only computed statistics, wash the same data at lower emphasis. Docked on distinctiveness — an abbreviated box plot is respectable statistical vernacular but not far from generic data-viz shorthand.

The product's one non-negotiable honesty is that no statistic ships without its qualification — and the diagonal hatch is the product's own device for marking thin evidence (HATCH_PITCH in tokens.ts). The mark is a single square data cell whose trailing strip is hatched: navy for what the lodgements say, ochre hatch for how thin they are, joined with no gap because in this product the two are inseparable. Geometry is whole-unit and flat — a 96-unit square (croppable as the favicon), 2px boundary rule and 3px hatch strokes at 45 degrees, spacing tuned so the strip still reads as textured ochre at 24px — and the wordmark is plain IBM Plex Serif 600 in ink, kept quiet so the hatch is the only signature.

Hue mapping:
#123a63 -> bond-lodgement data (the solid segment of the cell)
#8a5300 -> evidence thinness (the diagonal hatch and its boundary rule)
#f4e7cf -> thinness wash (the ground the hatch sits on, in place of transparency)
#16191d -> the wordmark, set in ink like body text

• The only candidate whose signature is the product's own honesty device: a navy data cell whose trailing segment carries the #8a5300 diagonal thin-evidence hatch on its #f4e7cf wash, statistic never shown without qualification. Hue mapping is exact to the system. Craft is the weak point — the hatch is hand-clipped with irregular endings (line 6 starts at x=94, off the 16-unit rhythm of the others) rather than pattern-clipped, and at 24px seven 3px diagonals in a 36-wide region collapse toward mud. Still the strongest concept-to-subject fit of the ochre entries.
• The strongest concept-to-product tie in the set: a solid navy data cell whose trailing segment carries the ochre thin-evidence hatch — the statistic never shown without its qualification, which is invariant 8/9 drawn as geometry. Hatch lines are evenly spaced (16px y-intercepts), 45°, clipped inside the #8a5300 2px boundary; whole units. Deductions: three non-neutral values (#123a63, #8a5300, #f4e7cf) brushes the ≤2 hue budget even though the wash is semantically ochre-family, and the 3px hatch lines risk mushing at favicon scale.
• The thin-evidence hatch is this product's own convention, and 'never show the statistic without its qualification' is the most honest concept in the set — ochre #8a5300 on #f4e7cf is precisely the system's thinness pair. The 45-degree hatch keeps consistent 16-unit intercept spacing and the 2px boundary rect (89,33,34x94) registers with the 88-124 fill. The diagonals are texture, not a trend line. Cost: 3px hatch strokes fall to ~0.45px at 24px height and the qualification — the whole point — mushes first.