Record at a glance
- Evidence level
- Named record: delivered scope, plus measured results the client reported
- Client
- 4over4, named because its permission to be referenced is on record
- Service
- Custom Product Engineering
- Delivered scope
- Conversion work on the print store, a product recommendation engine, and automated conversion of .ai files to SVG
- Published results
- Six, each below with the period its source records, or a note that none was recorded
- Whose results
- 4over4's, measured on 4over4's store
Context
A print store sells configurable physical goods, so two things decide whether a visit becomes an order: whether the shopper is shown something they want, and whether their own artwork can be used without a designer touching it first. 4over4's store needed work on both. Netbase was engaged to improve the conversion path and to remove the manual step between an uploaded design file and a usable, editable template.
What Netbase built
- a product recommendation engine that suggests products from browsing history and previous purchases;
- an automated conversion of uploaded Adobe Illustrator (.ai) design files into SVG, so that layers can be ungrouped and recoloured;
- import of .ai files as editable templates in the store;
- the surrounding e-commerce conversion work on the print store.
Which part is AI, and which part is not
The recommendation engine is the AI feature here. It reads what a shopper has looked at and bought before and proposes products on that basis.
The file conversion is not AI and is never described as such. Turning an .ai file into SVG is file-format workflow automation: a deterministic conversion that preserves layers so they can be ungrouped and recoloured. It is valuable because it removes a manual step, not because anything infers or predicts. Netbase builds both kinds of work, and the AI Workflow Automation service keeps the same distinction.
Results 4over4 reported
Each figure below is 4over4's own measurement on its own store after this work, not a result Netbase can promise a different buyer:
- revenue up 82% within six months;
- conversion up 48% on the store, with no period recorded in the source;
- average order value up 35%, with no period recorded in the source;
- production time down 40%, with no period recorded in the source;
- more than 200 templates published in three months;
- fulfilment time down 50%, with no period recorded in the source.
Evidence and limitations
- The revenue figure appears twice in Netbase's own materials, as 89% in the body of the case and 82% in its title. The lower value is the one published here, under the register's standing rule that the lower figure wins a conflict.
- Where the source records no period for a figure, none is attached. Restating, annualising or combining these figures is not permitted and has not been done.
- Netbase built the system; the business outcomes belong to 4over4's business, and many things other than software move revenue and order value.
- No screenshots of the store are published, and no interface is shown. The image on this page is illustrative.
- The methodology page explains how each claim on this site is sourced and reviewed.
Questions buyers ask
Does a recommendation engine produce the same lift on another store? No. The 48% conversion figure is 4over4's, measured on 4over4's catalogue and traffic. A different catalogue, a different audience and a different starting conversion rate give a different answer.
Is the Illustrator-to-SVG step an AI feature? No. It is a deterministic file conversion that keeps layers intact so they can be ungrouped and recoloured. Calling it AI would overstate it.
Can you publish more detail on the build? The record publishes the scope attested in Netbase's own documents. Anything beyond that would not be evidenced, so it is not here.
Planning similar conversion work?
Compare this with the other work records, then submit a project brief describing your catalogue and the step you want removed. OutsourcingVN is operated by Netbase JSC and is Netbase's own outsourcing-services platform, so the team that reads the brief is the team that would deliver it.