The Ficklr Fable
On the seductive and historically predictable habit of trading what works for what’s next
Most people think the hype works like this: something new appears, everyone gets excited, it disappoints, the headlines move on, and some of what was built persists.
Imagine a roller coaster. When a cycle starts to form, people tend to ask, “How do I get on this ride?”
The smarter question is, “What’s the structure underneath?”
The Gartner hype cycle was first published in 1995 as the Hype Cycle for Emerging Technologies. Those who follow Gartner’s dozens of annual domain-specific reports may recognize the curve.
Pull back from the shape, and a longer pattern emerges.
Carlota Perez studied every major technology wave since the 1770s in her 2002 book Technological Revolutions and Financial Capital: steam, railways, steel, oil, computers. Her research describes the same engine in every wave: people build something, then investors get excited and start funding the excitement instead of the building.
The Gartner curve shows what it looks like up close: attention rising and falling over a few years. Perez found a similar shape that stretches across decades.
But neither one accounts for everything. Both miss what people experience inside the cycle. Both ignore the regulations that could slow things down, or the absence of regulations that lets a frenzy run unchecked.
Money, attention, lived experience, and laws (or absence of laws). Put all four together, and you see a full system.
When these four are aligned, you get a golden age. Post-war America: highways, suburbs, a rising middle class, all built on oil and automobiles, all regulated just enough to hold together.
When the four pull apart, you get a frenzy. Dot-com speculation. Crypto mania. Investors fund the expectation of more money instead of the building itself.
That’s when a crash comes. Every time. Every wave. For at least two hundred years.
The ride is designed to be exciting. It’s not designed to be safe, and not everyone has to be on it.
This is a story about what that ride does to a burgeoning company that couldn’t get off. The names are made up, but the pattern is not.
The vertical search gold rush
The year is 2005, and Noodle is the front door to the internet. Everyone uses it to find everything, but the real money is in specialized verticals: shopping, travel, jobs, and real estate. Venture capital agrees. The gold is in the niches.
DealDive crawls retail sites, indexes what it finds, and compares prices across merchants. Users love it. Advertisers pay for it, and the revenue holds up.
The office has good snacks and even better energy. Nothing in the numbers says stop. They are not wrong. They might just be early.
The thing about building on someone else’s land
It takes two years for the landscape to shift, and when it does, it doesn’t come from a competitor. Noodle starts doing comparison shopping itself.
Noodle does what noodles do: it absorbs everything. Travel, jobs, real estate, shopping. The traffic that used to flow out to specialists stays on Noodle.
Advertisers shift their spend to Noodle’s search ads, and the referral traffic that kept DealDive alive starts to dry up. Two numbers tell the story: cost per crawl is climbing, revenue per merchant is falling, and the gap between them is widening.
But underneath DealDive’s thinning consumer product, something else was built. Something that still works.
The useful byproduct
To compare prices, you first need to have the prices.
DealDive had built something substantial: a crawling system that visited thousands of retail sites and an indexer that organized what it found into clean, structured records. Prices, availability, product attributes, and merchant reliability. It wasn’t the fastest, but it was thorough, and years of domain knowledge were baked into every normalization rule.
“Maybe the layer underneath can stand on its own. Maybe the thing built for one purpose turns out to be useful for many.”
New name, new direction
DealDive pivots. The consumer comparison engine is out. In its place: B2B data services for retailers and platforms, built on the same crawling infrastructure that powered the comparison engine.
“We already have the records, we already have the system, we just need new customers.”
The company renames itself. DealDive.com becomes Ficklr. The old name told you what the company did, but the new name doesn’t tell you much of anything. It doesn’t even have an “e.”
Get Sean Madrid’s stories in your inbox
Join Medium for free to get updates from this writer.
But the pivot is easier to name than to execute. The crawling system was designed to serve one thing, the comparison engine, and making it serve a wider variety of customers means rearchitecting it. Progress is slower than expected, and the team estimates 18 months to a full index.
The next wave doesn’t wait
While Ficklr is still working on the rebuild, the social web is already here.
Peephole has hundreds of millions of users, and they aren’t just searching anymore. They’re connecting, sharing, reviewing products, and recommending them to each other. The web has gone social, and it’s rewriting the rules of commerce in real time.
You can see it in the venture money. Social startups are raising tens of millions before they have revenue. You can see it at the conferences, where nobody’s talking about crawling anymore.
This is how cycles work. They don’t wait in line. They overlap.
Then ShelfWatch calls
ShelfWatch started in the same lane as DealDive and never changed direction. While Ficklr was rebranding itself, ShelfWatch went deeper into merchant records, ratings, and taxonomy. The boring, granular work of knowing everything about retail products. Now Nile, the largest online marketplace in the world, wants that knowledge at enterprise scale, and it’s waving the kind of money that rearranges lives.
ShelfWatch needs a partner who can help them scale: broader merchant coverage, faster indexing, and normalized feeds. They come to Ficklr because this is supposed to be exactly what Ficklr does now. Ficklr looks at the requirements, looks at what it can actually deliver, and realizes the new name is the only thing from the pivot that actually shipped.
Before the pivot, Ficklr was on the same trajectory as ShelfWatch. Same crawling infrastructure, same merchant records, same domain knowledge. Had they gone deeper instead of sideways, they might be the ones sitting across from Nile. Had the pivot actually worked, they’d be the partner ShelfWatch is looking for right now. Instead, both opportunities are in plain sight, and Ficklr can’t reach either one.
ShelfWatch is still called ShelfWatch. It still watches shelves.
Here we go again
The ShelfWatch call is still fresh when the venture money starts flowing toward social platforms. Competitors reposition overnight, and Ficklr can feel the market pulling away from them for the second time.
So Ficklr chases it. Platform strategy, social features, and connectors for integration. The decisions come fast, and they all point to the same thing, which feels like clarity but isn’t, because nobody’s asking what we're solving, for whom, why us, why now. The wave is here, and they just need to be on it.
Faster, messier, broken
The capital is running out, and what comes next moves fast: budget cuts, team reductions, and a declaration that it’s time for a new direction — a system built for the social web.
People who built the crawling system, who understood why the records were clean, and who knew the normalization rules. They leave in cardboard boxes.
So Ficklr rebuilds on a new stack optimized for speed and real-time connections. The crawling that took weeks now takes hours, and that’s a real win.
But the careful normalization, the reliability scoring, the taxonomy built over years; none of it transfers. Speed goes up, but fidelity goes down, and nobody made that tradeoff on purpose.
The new connectors are live, but the results are ugly. Fields that used to be complete come back partial, prices go stale, taxonomy flattens, and records that once reliably had ten data points now have four or five, depending on the day.
Nobody can tell if the problems are a skills gap or something deeper in the architecture, and the people who would know aren’t there to ask.
It would have been done by now
The old system was slow. That was the strongest argument for replacing it. But it’s been eighteen months, and the old system would have finished its crawl by now. The thing that was too slow turned out to be faster than the thing that replaced it.
Someone circled a date on the whiteboard once. It passed six weeks ago, and nobody erased it.
Same ride, different sign
The social web reshaped the internet between 2005 and 2010. Then it was mobile. Then crypto. Now AI.
Every time, the same pattern: the frenzy, the panic, the scramble to pivot. And more often than not, something useful gets sacrificed for something exciting, because the technology changes, but the pattern never does.
The hype cycle short-circuits the decision-making process itself.
The mania makes the fundamental questions feel too slow. What, for whom, why us, why now. They sound like first-day questions, but they’re really everyday questions.
Money feeds excitement. Excitement feeds money. The back-and-forth keeps going. But when the ride stops, only the people who knew what they were building, and why, have something to show for it.
Every frenzy leaves behind something useful that the frenzy itself couldn’t see. The dot-com era left fiber optic cables that now carry most of the world’s internet traffic. Blockchain technology has left programmable contracts that are quietly rebuilding global financial infrastructure. Each one burned through the hype and settled into productive, stable use. AI is still in the frenzy phase. The infrastructure it leaves behind is still forming, and the most durable piece of it may not have made a single headline yet.
The tracks. The mechanics. The foundation. That’s where the value is, and that’s what survives a crash. The ride drew the crowd, but the steel carries the weight.
