Over the past year, I’ve been sitting in two kinds of meetings. In the first, the team assures me everything is “under control”: Klaviyo, two freelancers, and a self-taught generalist. He has used Zapier and ChatGPT to wire together something they call an integration. For one feature a more serious platform includes in the licence, they maintain a Frankenstein of five systems. Nobody knows any more which segment refreshes where, or why the unsubscribe numbers never match. But it works. More or less.
In the second meeting, the decision was made before I sat down. Salesforce. They don’t yet know exactly what the company needs - but Salesforce will know. That mindset comes with its own saying from the seventies: nobody ever got fired for buying IBM. It was never an IBM ad, just an unwritten rule of corporate purchasing. Logos change; the rule stays. And to be fair, it doesn’t work only for Salesforce. Even today, it takes a brave CFO in a large company not to buy ERP from SAP.
Two meetings, the same mistake with opposite signs. One picks the tool out of love, the other out of fear. Neither picks by criteria.
Before everyone became an expert on ChatGPT and Claude, they were experts on Klaviyo. Before Klaviyo, on Facebook ads. Now it’s vibecoding’s turn: one Friday afternoon, we’ll build ourselves what thousands of engineers have spent decades developing.
Every generation of tools produces a generation of self-appointed experts. They defend the tool their market value is built on. That’s not mockery, that’s how the labour market works. But you don’t pick a platform to validate someone’s expertise. You pick it because you’ll have to make money with it for the next five years.
Now, something in defence of the first camp. The Frankenstein wasn’t born out of stupidity. It was born out of growth.
You chose Klaviyo correctly - back then. It was fast, affordable, and built for online retail. Its biography is public: in the IPO prospectus, it disclosed that 77.5% of recurring revenue came from customers selling on Shopify. A platform with that biography excels in a world of one brand, one market, one catalogue. Then you grew: physical stores, a second market, a loyalty programme, a wholesale channel. Each move added another prosthetic. Today you’re not paying for a platform - you’re paying for prosthetics and the people holding them together.
The licence isn’t the problem. Klaviyo costs $20 a month at 500 contacts, $720 at 50,000, and $2,300 at 250,000. Since February 2025, every contact the platform could message counts - including the ones you once imported and never write to. Even at the top of that curve, that’s $27,600 a year. Small change next to what two external contractors and internal patching cost you. The most expensive platform isn’t the one with the highest licence fee. It’s the one you’ve outgrown and keep on life support. The only thing more expensive is the one you bought and never did anything serious with. But it’s the right brand in the garage. That counts too.
Salesforce, meanwhile, is not a bad platform. It’s a very good platform for a company with an IT team, a budget, and the discipline of a large system. Marketing Cloud Engagement starts at $1,250 a month, the corporate edition costs $4,200, and enterprise costs whatever the quote says. A full enterprise deployment with every module exceeds a million dollars a year at list price. If you know that and need that, it’s a legitimate decision.
The problem starts when the decision isn’t a conclusion but an insurance policy. “Salesforce will know” isn’t a strategy - it’s shifting responsibility onto a logo. In sixteen years of FrodX, we’ve also migrated companies off Salesforce. Not because the platform was bad, but because the company was wrong for it. Too few people, too few processes, too few reasons for that level of complexity. The logo carried no responsibility. The signature on the contract did.
Which brings us to the point - to the people between the extremes, the ones honestly asking: “Which one is best?”
Wrong question. In a demo, Emarsys, Klaviyo, Bloomreach and Salesforce Marketing Cloud show you practically the same thing: segmentation, automation, personalisation, every channel, AI. An RFP with two hundred rows of requirements ends in a draw - all four have seen it a hundred times before. Gartner measured last year that companies actively use 49% of the martech capability they buy. Only 15% of organisations hit their goals and a positive return with it. Half the boxes you tick in that RFP will never get used. Winning the demo tells you who has the better demo. Nothing else.
So, the first test: instead of a table, write five scenarios from your own business. Concrete ones: a customer buys online, returns in-store, goes quiet for three months - what happens automatically, and who in the company sees it? Then ask each vendor to build those five scenarios live - with the people who would actually work on your project. A table asks what the platform can do. A scenario shows how much effort it takes to do it.
Second test: read the platform’s biography, because it tells you who it was written for. Klaviyo is the autobiography of the Shopify world. It’s moving upmarket: the number of customers paying over a million dollars a year doubled last year. But the default assumptions stay where it was born. Bloomreach Engagement is the former Exponea, built by data engineers in Slovakia and acquired by Bloomreach in January 2021. A CDP at the core, enormous freedom - but it demands a data discipline someone at your company has to own. Emarsys was born in Vienna for multichannel retailers. SAP bought it in November 2020 and renamed it SAP Engagement Cloud in February 2026. It assumes retail logic and an implementation partner, not Friday experiments. Salesforce Marketing Cloud is a collection of acquisitions glued into an enterprise offer - with all the power and complexity that brings.
None of these biographies is wrong - the question is which one matches yours. Sharper still: who at your company will run this in month thirteen, when the implementation team leaves? A marketer, a data engineer, or an external developer? That question eliminates half the candidates faster than any table.
The third test is uncomfortable, so almost nobody asks it: what does leaving cost? Every platform has an entry price in the offer and an exit price written nowhere: migrating segments, automations, history, consents. At FrodX, practically every third project last year and this year has been a replatforming. Most move up: the customer base has become too valuable to run on tools the company outgrew long ago. A minority moves down from the big brands - towards simplification, lower cost, and people who can actually run the system. Whoever knows the exit price in advance reads offers differently: less excited by the first-year discount, more interested in five years of growth.
The fourth test is the partner - and here I put my cards on the table. I sell SAP Engagement Cloud and HubSpot. I’m biased, and there’s no point pretending otherwise.
But bias has arithmetic. A consultant who lives off a single platform has exactly one answer to every question you ask. I have at least two - and enough replatformings behind me to know the price of the wrong one. I don’t sell Bloomreach, yet I’ll still mention it to a company with the right profile. A lost deal costs me less than a failed project with my name on it. An implementation partner doesn’t carry the risk at purchase; they carry it at implementation. That’s why a serious partner optimises for fit, not for their own logo. Some clients have even reversed the order: first they chose who to grow with, and only then the tool.
When you choose an advisor, ask three things. How many platforms do they know from the inside? How many migrations have they done? And what would they advise you against? If the answer to the last one is nothing, they aren’t advising. They’re selling.
“Which platform is best” is a question for demo day - and on demo day, all four win. The real questions are four others: your scenarios, the match of biographies, month thirteen, and the price of exit. Platforms don’t answer those. You do.
The Frankenstein and the believer have one thing in common: both made their decision without asking a single one.
igor.pauletic@frodx.com
P.S. HubSpot is deliberately absent from this comparison. It’s different terrain - CRM first, marketing second - and it would bend the comparison more than it would clarify it. Another time.