Selling things online seems relatively straightforward: Get a website, some product photos, and sign up for a payment processor. Bam, you're ready! But on the high end, it can be much more complicated than that. An online store overflowing with SKUs and traffic becomes a high-stakes affair, since small tweaks can have a huge effect on revenue.

That's why the purveyors of big e-commerce stores are turning to a new breed of enterprise technologies, deploying the machine learning and statistics techniques usually shorthanded as "A.I." to get as many people as possible from discovery all the way to checkout.

That process starts with choosing which products to stock. Eight-year-old startup Edited is a merchandizing database for the fashion industry that promises to "optimize assortments" and "eliminate blind spots." From luxury retailers like Net-a-Porter to British fast-fashion darlings Boohoo and Missguided, retailers turn to Edited "on a daily basis," according to Edited senior retail analyst Katie Smith. "It's really helping them drive their product assortment, timed correctly and priced right."

Edited scrapes the websites of "hundreds and hundreds" of apparel retailers, vacuuming up the metadata (color options, available sizes, etc.), and keeping track of things like when an item is discounted and when it goes out of stock. Edited can answer a question like, "Which midi dresses sold out at full price in the last three months?"

Beyond the metadata, Edited is also able to assess styles and trends using automatic image analysis. The system cross-references everything else it "knows" about a product to make sure its "judgment" is correct. For example, in an image with a shirt tucked into jeans with a belt, the machine is able to identify which item is the actual product via context clues. To identify patterns like polka dots and stripes, Edited uses a convolutional neural network, a machine-learning architecture that excels at extracting features from images.

Previously, this market research was all done by hand--buyers would go to rival stores in person and manually catalog everything they saw, relying on memory and intuition to identify trends. The scale and systematic nature of Edited go far beyond the analogue possibilities. "There are so many nuanced ways to describe a fashion product," Katie Smith tells Inc. "Being able to standardize which products go together is something that's never been done in the industry before." Edited also provides detailed, customizable analytics dashboards to its customers.

Edited doesn't index marketplaces like Amazon, although it has plans to do so in the future, but if e-commerce is a room then that giant platform is always the elephant.

Outlier, a young business intelligence startup, positions its A.I.-enabled service as a way for independent retailers to stay competitive with that pachyderm. "At Outlier, we are seeing the growth of Amazon put pressure on e-commerce companies to adopt new technologies like A.I. to get an edge," CEO Sean Byrnes writes in an email.

Outlier's product is a number-crunching system that cross-correlates the avalanche of data points that a modern business generates. For example, Outlier automatically detects subtle changes in the behavior of customer cohorts--women over 50 who received a certain marketing email, say--and surfaces them for human attention. Byrnes notes that 70 percent of Outlier's customers are in the e-commerce space.

Edited and Outlier aren't the only two startups addressing this market. There's Feedvisor, which helps merchants succeed on Amazon Marketplace through automated smart pricing, rather than focusing on the independent retailers who defy the behemoth. Alibaba-backed Twiggle uses natural language processing to make e-commerce product search work better.

E-commerce companies are often early adopters, since it's easy for them to track when using a new tool impacts revenue. The success of startups moving into this space indicates that A.I. translates to ROI.