Last year, Orbitz raised eyebrows when it was revealed that the online travel site rearranges the order of hotel search results, displaying more expensive lodging options to users it had reason to believe, based on data analysis, were more likely to be willing to pay a premium.

In the retail space, a similar data-driven approach to product pricing is increasingly being employed. Amazon is perhaps the best and most widely-known user of dynamic pricing, where the prices of products changes regularly over short intervals, if not in real-time. But as detailed by AdAge’s Kate Kaye, Amazon is hardly alone. Other retailers are jumping on the dynamic pricing bandwagon too.

That could be a good thing. By analyzing competitor pricing data on an ongoing basis and using it to adjust their own prices, retailers can, in theory, optimize sales. And, as Kaye points out, dynamic pricing can be a tool for fighting showrooming.

But dynamic pricing isn’t without risks. Here are five things retailers should consider when evaluating whether to employ it, and how much to employ it.

1. Customer perception

Many consumers aren’t aware of the fact that retailers alter prices on a regular basis, and did so even before the advent of online retail, but as it becomes more noticeable thanks to the web, retailers must consider the perception issues it raises.

Put simply, a customer who observes that a product can become cheaper or more expensive within minutes may not be thrilled at the prospect that they could end up paying more for a product based on little more than, say, the time of day.

Particularly worrisome is the possibility that some users will notice dynamic pricing, but won’t quite understand what’s going on, resulting in a reduction of trust.

2. Data accuracy

Dynamic pricing depends on data, and when pricing is being changed on the order of hours or even minutes, ensuring that the data driving pricing decisions is accurate is critical. While there’s a growing ecosystem of data providers and the techniques by which data is collected and filtered are sure to improve, retailers shouldn’t assume that bad data won’t make it into their systems.

3. Algorithm mishaps

Wall Street and the phenomenon of flash crashes reminds us that algorithms are far from perfect and can produce costly errors. As retailers embrace dynamic pricing models which are of course based on algorithms, thought should be given to how mishaps can be minimized and what policies will govern when a mishap results in a big mistake (eg. customers being able to purchase a product at a ridiculously low price).

4. Altered customer behavior

As the existence of dynamic pricing becomes more evident to consumers, retailers will need to grapple with the possibility that it could impact customer behavior.

On one hand, dynamic pricing clearly has the potential to encourage sales, but is it possible that in some instances it could it impede sales? If customers come to believe that the price of a product might go down in the very near future, and perhaps even on the same day, it’s not unfathomable that some of them would decide to hold off on a purchase.

And as every retailer knows, a delayed purchase is much more likely to become a purchase that never happens, or happens somewhere else.

5. Overall experience

While price is an important factor in purchasing decisions and is often the most important factor, retailers should remember that their long-term success will likely depend on their ability to offer much more than that.

Customer service, selection, shipping, return policies and loyalty schemes can also help drive sales, even when a retailer can’t offer the lowest price. These things are often crucial to fostering the brand positioning and customer loyalty retailers covet, so embracing dynamic pricing without addressing overall customer experience is short-sighted.