Qualify the line. Then repeat it.

The first physical proof

Prepare one packet with 48 labels on two Avery 22805 sheets, a starter card and a flat sleeve. This moving-kit calibration packet is distinct from the invitation and its mixed sample assortment. This is the calibration sample for the production line. After it passes, time a representative batch and establish a repeatable work instruction.

The prepared digital layout has passed all-48 decoding and template-geometry checks. Physical printing, adhesion, live claiming and household transfer have not been performed. The saved test files and manifest are retained for the internal pilot and are not available from this page. Candidate allocations and printable production-code files are kept out of this public report.

1. Confirm the printer, paper and settings

• Confirm the physical printer and its ink, condition and ability to feed the selected stock. The saved test packet was prepared for an existing HP ENVY queue. The production candidate needs its own permitted tray/media procedure; the old queue is not qualification for the new machine.
• Confirm the package is Avery 22805, US Letter, 24 square 1.5-inch labels per sheet. The final PDF: explanation uses the rectangles from Avery's downloaded official blank template; a similar-size label product may have different spacing.
• Select standard US Letter, single-sided / Duplex None, Actual Size / 100%. Disable Fit to Page, Scale to Fit and borderless enlargement. The earlier development queue defaulted to duplex; explicitly verify single-sided output on the selected queue. These instructions do not change the driver automatically.
• Select the media/feed settings appropriate to the actual stock and printer. Record the queue, driver, tray, media type and quality settings.

2. Calibrate on plain paper before using label stock

• Print both saved label pages on plain paper at the settings above. Overlay each on a blank physical label sheet and check the printed square boundaries against the die cuts.
• Measure horizontal and vertical displacement at the four corners and a central label on each sheet: ten positions total. Proposed acceptance is no more than 0.03 inch (0.76 mm) displacement in either direction at every position, with no clipped code, quiet zone or text. Record the values before declaring pass.
• Stop on skew, scaling error, clipping or feed failure. Correct the relevant setting and repeat the plain-paper check; record wasted sheets. Do not compensate by stretching the QR: explanation image.

3. Print and inspect one complete packet

• Print the two label pages single-sided on two stock sheets. Record all sheets used and rejected.
• Confirm 24 labels per sheet, 48 total, with the human-readable codes matching the saved manifest. Confirm no code or text lies outside its label, and no QR: explanation code is stretched or its surrounding blank margin clipped.
• Keep the 48 printed candidates together. They were generated offline, pass the source code format/checksum contract and are unique within this packet. This does not establish global allocation uniqueness, a registered database entry or a functioning live route.

4. Perform physical optical checks without opening live targets

• Sample positions 1, 10 and 24 on each sheet: six physical labels covering a corner, the central region and the opposite corner.
• On two phones, decode every sampled label in two recorded lighting conditions: well-lit indoor light and the ordinary light where the labels will be used. This is 6 labels × 2 phones × 2 conditions = 24 attempts. Record device, operating system, decoder/app and actual conditions; do not claim a measured light level without a meter.
• Use a decoder that displays the encoded text without fetching it. Do not open, tap through, navigate to or automatically preview the encoded production URL during optical calibration. Compare every result to that label's exact manifest target.
• Proposed acceptance: all 24 attempts return the exact target on the first attempt. Record retries and failures rather than omitting them. Any mismatch stops the run for investigation. This is an optical test, not a sign-in, claim, privacy or placement test.

5. Check adhesion on spare blank labels

• Preserve the 48-code packet. Use three sacrificial blank labels from remaining stock on a clean, dry corrugated-cardboard test panel. Record the surface and conditions.
• Apply each with ten firm fingertip presses. Inspect at 15 minutes and 24 hours. After the 24-hour inspection, make ten dry-finger rubs across each label and inspect all edges and corners.
• Proposed acceptance: no edge/corner lift greater than 1 mm after these checks. Record observed lift, damage and conditions. These proposed thresholds are an initial screen, not an established durability standard or a claim about moisture, heat or long-term use.

6. Pack, measure and decide

• Print and trim the starter card. Place the two sheets in sheet order with the card in a sleeve or envelope that holds Letter sheets flat. Do not fold the sheets or let sleeve adhesive touch labels. Keep the manifest separate; do not print household details on the sleeve.
• Record preparation, printing, inspection, assembly and packing time separately; record used/spoiled stock, ink observations, packing cost, freight and any unmeasured cost.
• Stop on feed, registration, scan, adhesion, code/manifest or material cost failure. Preserve the failed sample and record the correction before making more packets.
• Accept physical preparation only when its tests pass. Release a representative repeat batch only with a recorded result and the TAGD team’s qualified household workflow. First authenticated claim and verification of the exact placed code remain required; the current owner runbook does not establish working authenticated persistence.

Product workflow before distribution

The square-stock exporter still needs correction: selecting 24-up square stock routes through 30-up address-label geometry. The independent test layout bypasses that defect. Physical print acceptance does not establish working authenticated persistence, household claiming, transfer, privacy or recovery. Those flows must be qualified before distribution.

Commission all four stations

• Labels: correct the square-stock production exporter, verify identifier allocation and manifests, qualify the exact stock/tray, then test registration and scans on the actual printer. The saved independent 24-up file is a calibration aid, not an exporter repair.
• Invitations: lock the assortment and paper dimensions; prove front/back alignment, color, dry time, crease quality and trim. Time 25 complete invitations, including assembly and packing, before extrapolating a 250-piece run.
• Packaging: approve a fitted dieline or blank-mailer size. Check insert retention, adhesive cure, closure and the actual shipping package. For custom cutting, demonstrate the exact board, tooling, registration and finished boxes per hour on the quoted machine.
• 3D: explanation parts: qualify the same part/profile on both machines. Record dimensions, mass, time, rejects and bottle/insert fit. Retain the first accepted article; recheck after nozzle, material or profile changes.
• Batch release: reconcile identifiers, inspect every finished unit for identity and visible defects, log sampled phone scans and product-specific fit/durability results, and count accepted/rework/void units before packing. No live identifier is silently duplicated to replace a reject.
• Repeatability: require two consecutive representative batches to meet the documented criteria and measured schedule before advertising dependable throughput. Set an initial proposed 95% first-pass yield target, with every identifier discrepancy treated as a stop condition. Record actual performance; this is not a claimed result.

Work needed before fulfillment

TAGD owns the production-exporter and identifier lifecycle fixes. HomeHub owns its household/agent transfer and privacy workflow. Applied Sciences owns stock qualification, part design, assembly instructions and physical test evidence. Price only after materials, labor, machine time, waste, shipping, support and returns are accounted for. Unpassed product or software checks hold that product’s release; other qualified work can continue.