Real-World Review: How Much Did Our Price Comparison Gateway Save Readers in Month One?

Our price-comparison gateway has been live for a month. This empirical post-mortem answers one question: how much money did it actually save readers? All figures come straight from production—no storytelling, just numbers.

Let's start with scale

• Cached referral links: 120,000 JD Union links
• Valid links: 85,000 (70%)
• Auto-refreshed on expiry: 28,000 (re-converted in real time at click)

What does 120,000 links actually mean? It's 120,000 historical price anchors for products. Each one represents a chance to help a reader confirm the lowest possible price.

Core mechanism: expired links auto-regenerate

JD Union links expire by default after 40 days. Our approach:

1. User clicks an internal short link
2. The system checks whether the cached link has expired
3. If expired → call the JD API in real time to re-convert → 302-redirect to the new link
4. The user never notices, and the commission is preserved

This means: even if a user clicks into a year-old article, the link still works. That's the confidence behind our "evergreen article" SEO play.

Proof of savings: 3 real cases

Case 1: seasonal AC clearance

Late September: a certain 1.5-ton, first-grade efficiency air conditioner listed at ¥3,299. During National Day with the government subsidy, the final price dropped to ¥2,499. The reader saw ¥2,499 as the historical low on the comparison bar, confirmed it was the floor price, and checked out—saving ¥800.

Case 2: iPhone trade-in

The reader's old device was valued at ¥1,200, plus a ¥500 government subsidy and a ¥300 education discount. Stacked together, the deal saved ¥2,000 versus buying direct from Apple. Our purchase guide spells out the exact stacking order, steering readers away from the common trap of using coupons first and losing part of the subsidy.

Case 3: spotting the fake discount

A certain noise-canceling headphone jumped from ¥899 to ¥1,099 in mid-September, then advertised a "¥200 drop" during National Day—landing back at ¥899. A historical-price check showed ¥899 was already the price in early September, so the reader held out for Double 11. The "fake discount" trap was avoided.

Technical post-mortem: three things we got right

1. Server-rendered comparison bar: SEO text is output server-side, no JS dependency, so search engines can actually index it
2. Real-time link conversion on click: expired links auto-regenerate, making old articles viable for the long haul
3. Matomo tracking: full funnel from impression to click is traceable, so we know the true conversion rate per article

Two things we didn't nail

1. Missing coupon data: parse_coupons() is written, but without real coupon samples we can only fall back to a degraded order-completion strategy
2. Mobile experience: the comparison bar takes up too much screen space on phones and needs a better collapse behavior

Next month's target

Double 11 is the real test. Goal: cover every Double 11 guide article with the comparison bar, use historical price data to back up "true floor price" calls, and help readers avoid the usual Double 11 tricks.

Data sourced from production environment logs; organized with AI assistance.