Different Thinking: Why Cloudways Paid More for a Better Customer, on Purpose

Key Takeaways

• Most paid search accounts are built to chase the cheapest possible conversion. Cloudways, the managed cloud hosting platform later acquired by DigitalOcean for $350 million, built its account structure to chase the most valuable one instead, even when that meant paying more to acquire it.
• Working with agency Bind Media, Cloudways fed more than ten LTV-weighted customer personas directly into Google and Microsoft Ads, with bid adjustments set by country, so the bidding algorithm itself was optimising for long-term value rather than a single blended cost-per-acquisition target.
• The result was a 66% rise in customer value and a 78% rise in conversions simultaneously, evidence that optimising for quality and optimising for volume are not automatically in tension, if the underlying data feeding the ad platform is built correctly.
• The deeper lesson isn’t about paid search specifically. It’s about which number a growth team lets the algorithm see. If the only signal an ad platform gets is “did this become a customer,” it will happily fill your funnel with customers worth nothing, at speed.

There’s a quiet assumption baked into almost every paid acquisition account in existence, one so common that most growth teams never think to question it: a conversion is a conversion. Someone signs up, the pixel fires, the platform’s bidding algorithm notes that this particular combination of audience, creative and keyword produced a result, and it goes looking for more of the same. It’s a reasonable assumption for a lot of businesses. It is a genuinely bad one for a business whose customers vary wildly in how much they’re actually worth, which describes most SaaS and infrastructure companies, and describes Cloudways, the managed cloud hosting platform, with unusual precision.

The assumption everyone else was making

Cloudways sells managed hosting to a spectrum of customers that runs from solo developers spinning up a single WordPress site to agencies managing dozens of client servers, from one country’s price-sensitive freelancers to another’s well-funded digital shops. A signup from the first group and a signup from the second group look identical to a standard conversion pixel. They are not remotely identical in what they’re worth to the business over the following year. Yet the default behaviour of a paid search account, left to its own devices, treats them as interchangeable, because “interchangeable” is what the data feed tells it they are. Optimise purely for conversion volume under that setup, and you will, with total algorithmic confidence, spend your budget filling the funnel with the cheapest signups you can find, which are very often also the least valuable ones.

This is the trap that catches a lot of growth teams, and it rarely announces itself as a mistake. Cost per acquisition goes down, the dashboard turns green, the monthly report writes itself. The damage shows up later, diffusely, in a customer base that churns faster and spends less than it should, by which point nobody traces it back to a bidding decision made eighteen months earlier.

The different thinking: make the algorithm see what you see

Cloudways, working with growth agency Bind Media, made a specific, unglamorous engineering decision instead: rather than trying to fix this downstream, with better lead scoring or a sales team taught to prioritise, they fixed it at the source. Bind Media restructured Cloudways’ paid search accounts and built a custom conversion-data endpoint that fed more than ten LTV-weighted customer personas directly into Google and Microsoft Ads, with bid adjustments set by country on top. In plain terms: instead of telling the ad platforms “this person converted,” the new setup told them, in effect, “this person converted, and based on everything we already know about customers like them, here’s roughly what they’re actually going to be worth.” The bidding algorithm, given that richer signal, could do what it’s actually good at, optimising toward a target, except now the target was long-term value instead of a single blended cost-per-signup number.

It’s worth sitting with how unglamorous this is as an idea. There’s no new channel, no clever creative insight, no viral growth loop. It’s a data plumbing decision: make sure the number the algorithm is chasing is the number you actually care about. Most growth teams never get around to this, not because it’s technically difficult exactly, but because it requires admitting that the metric you’ve been reporting on, raw conversions, was never the real goal, just the easiest thing to measure.

Why “more expensive but better” beat “cheap and more”

The results argue for themselves in a way that should unsettle anyone who assumes quality and volume trade off against each other. Customer value rose 66%, and conversions rose 78%, at the same time. Conversion rate improved 20% on top of that. If the standard assumption about acquisition marketing were correct, that chasing higher-value customers should mean chasing fewer, more expensive, harder-to-close ones, at least one of those numbers should have moved in the opposite direction. None of them did. That’s the tell that the underlying model was wrong, not just under-optimised: Cloudways wasn’t trading value for volume or volume for value, because the entire premise of a fixed trade-off assumed the algorithm couldn’t tell the difference between customers in the first place. Once it could, it stopped needing to choose.

Joakim Holmquist, Marketing Director at DigitalOcean, put his own scepticism about agencies on record before crediting the result: “Working with Bind Media has been a refreshing experience. They managed Cloudways’ paid media and broke my recurring experience about agencies lacking expertise and ownership. Bind Media’s in-depth knowledge of the paid media landscape to drive SaaS growth and strong sense of ownership helped us achieve record numbers.” It’s a telling detail that the compliment is aimed less at any single tactic and more at ownership, the sense that someone was actually accountable for whether the underlying number being optimised was the right one, rather than just the easiest one to report against a target.

Why this mattered more than a marketing metric

Cloudways wasn’t running this program in a vacuum. The company, based in Malta, was scaling toward what became a $350 million acquisition by DigitalOcean, announced in August 2022 and completed that September, at a valuation reported at roughly 6.7 times Cloudways’ annual revenue, with the company generating north of $52 million in revenue that fiscal year and a three-year compound annual growth rate above 50 percent. A business being evaluated on the durability and quality of its growth, not just its top-line size, has a very direct incentive to make sure its acquisition engine is actually producing customers who stick around and spend, rather than a headline signup number that looks impressive in a board deck and evaporates on renewal. The LTV-based account restructuring wasn’t a marketing-team side project. It was, in a fairly literal sense, protecting the story the business needed to tell about itself.

What this means if your growth stack still only knows “converted”

The specific mechanics here, LTV-weighted personas, custom conversion endpoints, country-level bid adjustments, are Cloudways-and-Bind-Media-specific. The underlying principle generalises to almost any founder running paid acquisition with more than one kind of customer, which is to say, almost every founder running paid acquisition. Ask a blunt question about your own account: does your ad platform currently know the difference between your best customer and your worst one, or does it only know that both of them, at some point, converted? If the honest answer is the second one, you are not running a growth engine optimised for your business. You’re running one optimised for whatever is cheapest to acquire, and trusting that cheap and good turn out to be the same thing purely by coincidence. Cloudways’ bet, and its result, was that they usually aren’t, and that the fix isn’t a smarter campaign. It’s better data, fed earlier, to an algorithm that was capable of using it the whole time.