On this page
How to Increase Conversion Rates: Lessons from Bing and Expedia
Why do visitors leave without buying? Learn where shoppers drop off, what causes hesitation, and how to continuously improve your Shopify store’s conversion rate.
To increase conversion rates, find a purchase decision that the store is making harder than it needs to be. That may be an unclear product difference, a confusing filter or a checkout field that shoppers cannot complete. More traffic will send more people through the same obstacle.
Two widely discussed examples, Bing’s ad-headline experiment and Expedia’s booking form, show how a specific interface change can matter. Their results are not forecasts for a Shopify store. Use them to frame a testable question about your own shoppers.
First, make the conversion math explicit
Imagine two stores with 20,000 monthly sessions and a $100 order value. At 2.2% conversion, a simplified model with one order per converting session produces $44,000. At 0.8%, it produces $16,000. The difference is 1.4 percentage points and $28,000 in monthly revenue under those assumptions.
These are hypothetical figures, not a Shopify benchmark. They hold traffic and order value constant. In a real comparison, also check refunds, discounts, margin and whether the visitor mix changed.
Locate the decision before choosing the tactic
Possible points of friction across the shopping journey.
| Stage | Shopper question | What to inspect |
|---|---|---|
| Landing | Is this what I came for? | The match between the search or ad and the page |
| Discovery | Can I find a suitable product? | Collections, search terms, filters and availability |
| Product | Why this item at this price? | Fit, product differences, proof and delivery |
| Cart | What will I pay and when will it arrive? | Total cost, selected items and shipping information |
| Checkout | Can I finish and recover from an error? | Forms, payment methods and error handling |
Use reports to narrow the investigation, then follow the journey yourself. A large drop between two stages is a place to look. It is not proof that the most visible design element caused the loss.
Bing: evaluate the headline change with an experiment
Bing headline comparison. Experiment source: Kohavi and Thomke, Harvard Business Review.
Bing’s headline experiment increased revenue 12%, equivalent to more than $100 million annually in the US. Source: Kohavi and Thomke.
For a merchant, the useful question is whether shoppers can understand an important difference in the space where they make the decision. A clearer collection-card label or product heading is worth investigating when it answers a question the existing page leaves open.
That is an application to Shopify, rather than a proven explanation of the Bing result. Write the hypothesis before testing: “Shoppers cannot distinguish the two sizes from the card; including dimensions may help them choose the suitable item.” Measure purchases and guardrails as well as clicks.
Expedia: investigate a confusing form field
Expedia form example. Case discussed in Luke Wroblewski’s presentation, slide 81.
Luke Wroblewski’s presentation describes an optional Company field in Expedia’s booking form and reports $12 million in annual profit after its removal. The figure is profit, rather than revenue. See slide 81 of the presentation.
For a Shopify checkout, first identify the fields and errors you can actually change on the store’s plan. Reproduce the failure on the affected device. Check the label, required input and recovery path before assuming that every optional field should disappear.
A checkout may also depend on payment, shipping and app settings. The right fix may belong to another provider. The Shopify checkout optimization guide separates those responsibilities.
Use customer context to choose the right improvement
A merchant selling technical spare parts needs a different discovery experience from a beauty retailer. At Last Mile SF & Marin, the priority was finding relevant parts within a complex catalog. At Palace Beauty Galleria, Gentoo cleaned up duplicate and misspelled brand-filter values.
Both are specific implementations tied to how people browse. Neither means every store needs another filter. A small catalog may need clearer product differences first. A premium item may need a stronger explanation of its materials, fit or service.
At Vom Fass, Gentoo added product purchase links in recipe articles and related products on selected pages. That connected an existing reading task with a relevant next action. To assess a similar change, follow the path from article to product and check whether readers complete a purchase.
Write a short brief before making the change
| Part of the brief | Illustrative example |
|---|---|
| Problem | Mobile shoppers cannot tell which accessory fits their chosen product |
| Evidence | The compatibility information is missing from the card and difficult to find on the page |
| Change | Show verified fit information beside the product choice |
| Primary measure | Purchases among the affected product-viewing sessions |
| Guardrails | Returns for wrong fit, support contacts and margin |
| Review | Record the release date and compare suitable periods or run a controlled test |
The example is a proposed investigation, not a customer result. Agree on who will implement it and what evidence would make you revise the explanation. That prevents a plausible idea from becoming an indefinite series of edits.
What these experiments can tell a Shopify merchant
The useful transfer is the method: identify a specific decision, make a defined change and evaluate an outcome. A headline result from a search engine does not predict the lift from rewriting a Shopify title. A travel-booking form does not establish that the same field should disappear from every checkout.
Turn the lesson into a task in your store. If the product difference is unclear, inspect product-page value and option information. If a checkout field blocks an otherwise valid order, check the supported checkout settings and test path. Keep the hypothesis specific enough that an unfavorable result would change your decision.
Keep the result attached to the work
After release, verify that the task works and review the agreed outcome. Record stock changes, campaigns and promotions that could affect the comparison. Where a controlled experiment is practical, use it to distinguish the change from other influences.
Gentoo helps choose an opportunity against the merchant’s goal, prepares it for approval and carries the approved work through execution and outcome review. The aim is to make the next store decision easier, with the earlier result available to learn from.
Find the purchase friction worth investigating in your store. See your top three store opportunities with Gentoo.
