Halden, Part 4: The Win
A new flavour launched in one click, a Prime Day plan built from one sentence, and the month Halden beat its own January. The last part of a four-part story.
A Fluvi story in four parts. This is the last one; start with part one.
Salted caramel
Maren had wanted to launch a fourth flavour for two years. Every time it came up, the same thing stopped it: the memory of the last launch.
When Halden launched the chocolate 1kg, the agency took three weeks to build its campaigns. Thirteen of them, each with its own budget, bids, and a naming scheme only the agency understood. Then six weeks of daily babysitting while the new listing had no sales history and every bid was a guess. The chocolate tub lost money for its first four months.
Salted caramel was ready in June: Whey Isolate, Salted Caramel, 1kg, $41.99. Prime Day was six weeks later. There was no time for a three-week build, and Halden wasn't going to repeat the four months.
Thirteen campaigns in one go
The campaigns you build for every launch are the same ones each time. Only the product changes. So Theo opened Fluvi's campaign launcher, picked the new product, typed a slug, and got the whole set laid out in front of him:
An auto campaign. Broad and exact, two tiers each. A campaign for Halden's own name. Three Sponsored Brands campaigns. Four Sponsored Display, from rivals' product pages to shoppers who'd viewed or bought before. Every one named the way the rest of the account is named, so the agents' scopes would find them.
Theo moved the main exact campaign's budget to $100 and pressed Launch 13 campaigns. They were live on Amazon in the time it took his coffee to brew. The thing that took the agency three weeks took one click, because it was the same work every time.
Bidding with no history
A brand-new product has the opposite problem from Halden's January. There's no waste to cut, because there's no data at all. The salted caramel tub had zero clicks, zero orders, zero history. In the old world, Theo would have guessed a bid for every keyword, watched, and corrected for six weeks.
Fluvi's bid suggestions don't start from zero. A keyword with three clicks and no orders hasn't converted at 0%. It just hasn't been measured yet. Its family has been: the same phrase on Halden's other products, its tier, its product, the account. So a new keyword's suggested bid leans on its family, and moves toward its own numbers as its own clicks arrive.
So "salted caramel protein powder" got a real suggested bid on day one, from how tier 1 searches convert across Halden's account and from the new tub's own price and margin. Within three weeks, as its own clicks arrived, the suggestion had moved to its own numbers.
Theo didn't need a new agent either. Rank and relevance already covered every US keyword target, so the salted caramel's keywords came under the same six forks the morning they launched. Harvesting, placements, dayparting: all of it covered the new campaigns from their first night.
The salted caramel tub was profitable on ads in its fifth week. The chocolate tub had taken four months.
The end-of-month sweep
At the end of June, Theo did what he now called the sweep. It was the last bit of hands-on PPC he still did himself, and he did it the way Maren had done the stock pause in Part 3: by typing to his assistant.
The 41 searches were each too small for the harvesting agent's ten-click rule. A few clicks on "whey pills" here, a few on "protein bars" there. Together they'd cost $212 in a month. One sentence, and they were gone.
The chocolate searches were the opposite: 19 searches that had sold through broad and auto but sat just under the harvesting agent's three-order line. Theo had drawn that line on purpose, to be careful. Once a month, he brought in the ones that were close. Two sentences, and the sweep was done. It used to take a full day.
Prime Day, in one sentence
Everything Halden had built since February had been about spending well: cutting waste, pricing each click at what it's worth, aiming at an ACoS the business could live with. For Prime Day, Maren wanted the other thing.
Over Prime Day, shoppers who've never bought a brand try one, and the ones who like it come back for months. A new customer won over Prime Day is worth far more than the order they place that week. It's the one time a year when paying more for a click than the ad order alone is worth can be the right call.
That's exactly what the growth suggestion measures. Since Part 1, Fluvi had worked out two bids for every target. Standard counts only the ad order. Growth adds the organic orders an ad order brings with it, which on the vanilla tub was about five for every ten ad orders. Halden had coasted on the standard side all spring. For Prime Day it would push.
Maren described the plan to her assistant the way she'd describe it to Theo:
The assistant built the agent, every fork included, the same kind Theo had drawn by hand in Part 2. Theo opened it on the canvas, checked the forks, read three days of simulated changes, and changed nothing. On the Monday before Prime Day, Maren switched it live.
The campaign agent saw the extra spend coming and raised budgets as campaigns ran dry. The dayparting agent kept shifting bids on the hour. The rank checks showed the vanilla tub climbing for "whey isolate", the search everyone in the category fights over, from 12th to 4th by the second day.
On the Friday after Prime Day, Theo switched the push off. The bid agent took back over at its normal 25% aim, and the account went back to its usual rhythm the next morning.
Halden's two Prime Days were its two best sales days ever: $9,800 across both, three times a normal day, and a third of the orders from customers who had never bought Halden before. In August, a good share of those customers came back for a second tub, from Halden's organic listing, without a click to pay for.
The report
Halden had taken a small investment from two angels the year before, and Maren sends them a short update every month. She used to dread the Amazon section. It was a screenshot of Seller Central and a paragraph of excuses.
In July she built June's update in Fluvi's report builder, from blocks: a performance summary, traffic, a chart of daily sales and spend, the top campaigns and products. It downloaded as a PDF with Halden's name on top.
One of the angels replied within the hour: What changed?
Maren wrote back: We stopped guessing.
Six months
In the first week of August, Priya put one slide on the conference room TV, in the same room where she'd shown the January numbers. This time she talked while it was up.
"January is our best month," Priya said. "Every year. June is usually one of our worst. This June we sold twenty-three percent more than January, and spent less on ads doing it. Every ad order now earns more than it costs, before we count a single organic sale."
She skipped to the last row. "And Theo is back to coaching three mornings a week."
That row was the one Maren cared about most. Halden hadn't hired anyone, or found a better agency, or a growth hacker with a secret. It had taken the judgement one person already had, the five lines Theo wrote in his car on a Saturday morning, and made it run on every keyword, every day, without him.
Theo still reads the logs. He reads them on Sunday evenings now, not at midnight on a Tuesday, and mostly to look for the next idea. The last one was audience boosts. The campaign agent can manage those too, and Halden's repeat-purchase customers look like exactly the kind of shoppers worth paying a bit more for. He's had it in simulation for a week.
He'll probably switch it live on Monday.
Previous: Part 3: Momentum. Start from the beginning: Part 1: The Leak.
Halden's story starts the same way yours can: with a free look at your own account, and nothing changing on Amazon until you say so. Run your free audit.
About this story: Halden and its people are fictional, the same brand you'll find on our PPC product page, and the numbers are illustrative. The screens are the real Fluvi app, running on Halden's demo account. Every feature in it works the way it's described.
