Case studies

A spec tells the factory how to build a product, right down to the label. At this apparel PortCo, more than 2,000 of them lived in various PowerPoints, Excel files, email threads and the ERP.

When an engineering or label change came through, it could touch around 50 specs. As such, a large change took about 2 weeks to work through, and roughly 15% of specs reached the factory floor with a mistake in them.

Design, technical design, product management and the floor all worked from different versions of the same product spec. Something as minor as a label change could impact 50 documents, and with several teams manually altering each, error rates were high. The company needed a single record that each team could work from.

The company expects a full-year spec error rate of <1%, from 15%. The company is now also working to cut concept-to-production from nine months to six. LightSight estimates that would be worth about $4.1M a year in added revenue and about $436k a year in cost savings, or roughly $817k a year in EBITDA and $10.5M in enterprise value at exit.

Every month on the old process sent more wrong specs to the floor, took considerable time, and delayed product launches.

Creating workarounds and settling for the old process was costing the design team hours they should have spent designing, and costing the company revenue from delayed launches.

The operator runs 259 buses from nine locations in two business units. To see which buses were free on a given day, the team had to export fleet records, the out-of-service report, the dispatch schedule and the quotes from various sources, then paste them into a workbook that worked out availability per location.

Each copy showed the day it was exported. Service entries logged after that day were missing from it, so out-of-service counts for later dates read zero.

The data needed to be a live view, as availability for buses was constantly changing.

About a fifth of the quote rows were duplicates, so open quotes were double counted, and overestimated capacity. Cancelled trips stayed in the dispatch export and counted as committed buses. Data errors like these were frequent and at times overcounted capacity by 2x.

The team uses a dashboard that reduces manual data pulls and exports, while ensuring accuracy and clear view on capacity. This allows the company to book more customers, and reduce subcontracting costs.

A board that is a week old cannot tell sales whether to take the next charter.

When the board overstates commitments, sales turns away trips the fleet could have run. When it hides a shortage, the operator finds out late and has to subcontract a bus to cover it.

Each PortCo sent a monthly reporting pack. The fund's team rebuilt the portfolio picture by hand from those packs, board decks, credit agreements and a separate spreadsheet model for each company. No two companies reported the same way, so even a simple comparison started with someone reconciling definitions.

By the time the roll-up was done, the month was half over. The documents were confidential, which ruled out putting them in a shared drive and searching them.

The month was lost to manual inputs, well before anyone opened a dashboard.

A fund in this spot usually gets pitched a dashboard. The time went earlier than that, to opening each pack, pulling the figures out, typing them into the right model and checking the totals. A nicer screen on top of that process would have displayed the same late numbers.

LightSight started with the packs. Each company's reporting was mapped to one shared structure, so revenue at one company is defined the same way as revenue at the next.

All PortCo data is centralized in one place, so the fund can track performance faster, without rebuilding anything.

A fund that rebuilds its numbers by hand sees each month halfway through the next one.

Over a hold period, that is a long run of decisions made on old numbers. The dashboard decreases the lead time between a company closing its month and the fund seeing it.