Inspire to PSL Parallel Run Testing

A rebuilt job is done when it produces the same results as the old job on real data, every cycle, and someone signs off on that. A good-looking proof is only the start. A parallel run puts Quadient Inspire and Jetletter PSL Page Builder through the same input side by side so you can find every difference before your clients or the post office do.

Set up a fair test

Freeze the input. Use the same data file for both systems, after the same CASS, NCOA and presort steps. If the two systems get different data, every comparison after that is noise.

Pick at least one full production file plus a set of edge-case records from your WFD inventory. Include foreign and military addresses, long names, missing fields, every version and the record with the most pages.

Keep test output out of the mail stream. Label folders clearly and make sure no test file can reach a press queue or a postal upload by accident.

Record counts

Start with the simplest check. Input records, output records, suppressed records and records per version or split should match exactly. If they don't, stop there. A count difference points to a filter, deduplication or condition rule that was translated wrong, and everything downstream will be off too.

Page counts

Compare total pages and pages per record. A record with an extra or missing page usually means a page selection condition or a table that breaks differently. Sort records by page count in both outputs and compare the distribution, then look closely at the records that differ.

Visual PDF comparison

Compare the PDFs page by page. A PDF comparison tool that overlays two files and highlights pixel differences is the fastest way to work through a large set. Expect small differences in font rendering or line breaks. Decide which differences are acceptable and which are defects, and write that rule down.

Pay particular attention to:

Address and IMb position relative to the envelope window
Copyfitting on long names and addresses
Table breaks and continuation pages
Variable images, crops and resolution
Bleed, crop marks and imposition layout

Check PDF settings too. We recommend PDF 1.4 for digital presses. Run a sample of PSL output through your RIP and press before sign-off.

Data spot checks

Visual matches can still hide wrong data. Pick records at random and records chosen to hit each rule. For each one, trace the input values to the printed result. Check names, amounts, dates, offers, versions and any calculated field. Confirm formatting rules such as currency, dates and capitalization came through.

Postal checks

Postal problems cost postage and can cost a mailing its rates, so give them their own pass.

IMb. Decode the barcodes from both outputs. The 20-digit tracking code (Barcode ID, Service Type ID, Mailer ID and serial number) and the routing code (ZIP, ZIP+4 and delivery point) should match the source data. For Full-Service mail, serial numbers must stay unique for 45 days, so make sure test runs don't consume or duplicate serials that production needs.
Address block. Compare line order, endorsement lines and formatting. The PSL Address Box applies USPS address formatting, so confirm it matches what your Inspire job produced.
Presort order. Output order must match the presort sequence your postal software produced, including tray and bundle breaks if your files carry them.
Tracking. If you track mail through Informed Visibility, confirm the IMbs in the new output tie back to your tracking data. See our USPS mail tracking services.

Performance timing

Time each step on the same hardware conditions you will use in production: merge time, output time and total time from data drop to print-ready file. PSL merges 20,000+ records per minute on one core and 100,000+ complex records per minute on multi-core, but your own jobs are the number that counts. Record the timing for every cycle so you have a baseline after cutover.

Sign-off criteria

Agree on the criteria before testing starts, and get the job owner or client to approve them in writing. A typical set:

Check	Pass condition
Record counts	Exact match on input, output, suppressed and per-version counts
Page counts	Exact match per record, or each difference explained and approved
Visual comparison	No defects; accepted rendering differences documented
Data spot checks	All sampled records correct
Postal	IMb, address block and presort order verified
Performance	Completes within the production window
Cycles	Passes the agreed number of consecutive real cycles

Keep a log of every difference found, its cause, the fix and who approved it. That log is your proof of due diligence and your training material for the next job.

Cutover and rollback

When a job meets the criteria, schedule cutover for a specific run. Tell the press room and the client which system produced which files.

Keep the Inspire job in place and ready to run for a set period after cutover. Write the rollback steps down before you need them, including who decides, how the data is rerouted and how IMb serial numbers stay unique if you switch back. If a problem shows up, roll back, fix and retest. Then cut over again.

Move jobs in waves, and carry lessons from each wave into the next. The full WFD conversion plan shows where testing fits.

How MhetaData helps

Parallel testing and sign-off are built into our Inspire to PSL conversion service. We set up the comparisons, track every difference to a fix and plan cutover and rollback with your team. Call (800) 625-0369.

Quadient and Inspire are trademarks of their respective owner.

Running Inspire and PSL Side by Side During Migration