Case Study: Winning With Banner Advertising
A small catalog spent $4,000 on banner ads in 1999 and nearly quit after a bad first week. What changed, and what it says about that year's banner economics.
A specialty kitchenware catalog we’ll call Hearth & Home (the owner asked us not to use the real name, understandably, since some of these numbers are unflattering) ran its first banner campaign in January 1999. It’s a useful case study precisely because the first two weeks were bad. The campaign only turned around because the owner refused to quit on week one, and that refusal happened to be the right call for reasons that had nothing to do with stubbornness.
The setup
Hearth & Home sold mostly through a printed mail-order catalog and a small website built the year before. The owner, a former restaurant supplier, wanted to test whether banner ads could bring in orders directly rather than just catalog requests. She booked $4,000 in banner inventory across two cooking and homemaking sites that sold ad space directly (this was well before any real ad-network self-serve tooling existed), running a single static GIF banner at 468x60 pixels advertising a cast-iron cookware set.
Week one: it didn’t work
The banner pulled a 0.3% click-through rate, in line with what most banners got that year, but almost nobody who clicked bought anything. Four days in, seven clicks had produced zero orders. The instinct at that point is to pull the ad and call banners a bust, and that’s exactly what almost happened.
Before killing it, she looked at where the clicks landed: her homepage. Visitors clicked an ad for a specific cookware set and arrived somewhere that made them hunt for it through a category menu. Most left.
The fix that mattered
She built a single page, nothing fancy, just the cookware set, three photos, the price and an order form, and pointed the same banner at that page instead of the homepage. Nothing else about the campaign changed: same banner, same two sites, same budget.
Click-through rate held at roughly 0.3%. Orders did not. By the end of week three, the campaign had produced 41 orders against roughly 380 clicks, a conversion rate near 11%, against a spend of $4,000 and average order value of $58. That’s about $2,380 in revenue against $4,000 spent, which sounds like a loss until you account for repeat purchases: mail-order kitchenware customers in her existing base reordered roughly 2.3 times a year, so the real payback window ran past the campaign itself.
What changed
Nothing about the ad. Nothing about the price. Nothing about the audience. The only variable was where the click landed, and it was the difference between a campaign that looked like a failure and one that looked like a foundation to build on.
That’s the part worth sitting with. It’s tempting to blame banner advertising itself when a campaign underperforms, when the actual problem is usually a mismatch between the promise in the ad and the experience after the click. A banner that says “cast-iron skillet, $34” and lands on a general homepage is asking visitors to redo the work the ad already did for them. Most won’t bother.
What we’d tell anyone running a similar test
Don’t judge a banner campaign on clicks. Judge the landing experience on whether it delivers exactly what the ad promised, as fast as possible. And give a campaign at least two full weeks before deciding anything, because the first few days of any new placement tend to attract the least-qualified clicks: people testing whether the banner is even clickable, curious browsers, accidental hits.
If you’re weighing whether banner ads are worth testing for a catalog or small retail business, the honest answer from this campaign is: worth testing, but budget for a landing page as part of the cost, not an afterthought. The banner was never the weak link here. The page after it was.
See more from our Paid Advertising coverage or browse everything published in 1999.
