All case studies
🛒 Case study · Retail

How E-commerce prices you

Most of the personalization on non-Amazon e-commerce sites hides inside coupons, discounts, and A/B tests rather than changes to the sticker price. The effect is the same. Two shoppers with different browsing histories can land on identical product pages and walk away paying meaningfully different totals, without either of them realizing that's what happened.

Typical markup
5-20%
Typical price spread
Top signal
Cart
Abandonment
Disclosure
Rare
Opt-out available

How it works

The mechanics underneath
your shopping price.

The most common pattern is coupon personalization. The sticker price stays the same, but some shoppers get a popup, an emailed code, or a silent auto-applied discount at checkout. Who qualifies for which discount depends on a model that tries to predict how likely you are to abandon your cart without one.

A/B testing is the second pattern. Retailers run continuous price experiments on their own product pages, sometimes showing a test price to a random slice of shoppers to measure conversion. When the test price converts better than the control, it graduates. When it doesn’t, it rolls back. Most shoppers never know they were in an experiment, and the data suggests most of the time nobody tells them.

The third pattern, documented in a 2012 Wall Street Journal investigation of Staples, is location-based pricing. Staples was quietly showing higher prices to shoppers in ZIP codes farther from competitor stores. The company’s defense at the time was that the practice was standard business. Whether it’s standard or not, it’s still happening across the sector more than a decade later, and only in rarer cases does a regulator catch it.

The signals

What they’re actually reading
about you.

Weights are approximate and based on published research, regulatory filings, and reverse-engineering studies. Sources are cited in full on the Sources page.

🛒 Cart abandonment history
+22%

If you’ve abandoned carts before and come back to complete the purchase anyway, the model flags you as low-sensitivity and suppresses retention discounts.

✉️ Email engagement
+12%

Whether you open marketing emails, click through, and which offers you acted on in the past all feed the coupon personalization layer.

💎 Customer lifetime value
+18%

High-LTV customers often see fewer discounts because the model knows they’ll buy anyway. The loyalty penalty shows up here.

📍 ZIP code
+14%

Urban dense zips with more competitor coverage tend to see sharper discounting. Rural and low-competition zips tend to see less.

📱 Device and referrer
+10%

Mobile, desktop, and tablet all see different test slices. Where you came from (Google search, direct, email, social) also matters.

🔍 Product page dwell and compare
+8%

How long you hover, whether you use compare tools, and whether you view reviews are all inputs into whether the site thinks you’re ready to convert.

📊 A/B experiment assignment
+10%

On any given visit, you’re randomly slotted into multiple concurrent experiments. Some adjust price, some adjust the nudges around it.

💳 Payment method
+6%

Some retailers offer a discount for one payment method over another. That’s a transparent choice architecture rather than hidden personalization, but the price moves either way.

Real example

Laptop bag on a major online retailer

A Consumer Reports 2022 investigation tested a $59 laptop bag across six different shopper profiles on the same major online retailer over forty-eight hours. Final checkout totals ranged from $47.32, the lowest, to $59.00 at full price, depending on cart abandonment patterns, whether the shopper had a loyalty account, and what the model predicted about price sensitivity from prior browsing.

Low end
$47.32
High end
$59.00
+24.7%

What you can do

Ways to push back
that actually work.

None of these are silver bullets, but together they can shift the signals enough to meaningfully change the number you see.

01

Abandon the cart on purpose

Add what you want, close the tab, wait 24 hours. About a third of retailers will send a discount code or drop the price. The ones that don’t are usually the ones where pricing is consistent in the first place.

02

Sign out before checkout

A logged-out session looks like a fresh visitor and often triggers different pricing experiments. Logged-out shoppers get more aggressive first-time-buyer incentives.

03

Search the site from Google

Arriving from an outside search engine can place you in a different segment than a direct visitor. Google-referred traffic sometimes unlocks promo codes that are gated from direct shoppers.

04

Check dedicated coupon sites

Honey, RetailMeNot, and Rakuten automate coupon-stacking. They’re not perfect, and they sometimes create personalization of their own, but they regularly surface codes the retailer’s own site will not.

Try it yourself

Watch a shopping price get built for you.

The simulator uses the same factor weights shown above. Change any signal and watch the price move.

Launch the Simulator

Other case studies