The Switching Cost Confession

This morning at Omnicom's investor day, a sentence was spoken that every marketing client in the industry should have heard. The room was full of investors. Not clients. And the same sentence means completely different things depending on which chair you're sitting in.

Omnicom opened their entire presentation with RealID, Acxiom's identity platform, the centrepiece (alongside achieving media scale) of the IPG acquisition rationale. Two claims stood out.

First: RealID is more trusted because of the breadth of enterprises that use it: banks, financial institutions, healthcare companies, not just advertisers. The implication being that if regulated industries trust it with sensitive consumer data, marketers should too.

Second (and this is the one that stopped me) when clients use RealID, it "increases switching costs." Wow.

To an investor, that sentence is music. It means revenue is protected. Sticky. Recurring. That clients who build their marketing infrastructure around this platform will find leaving expensive, slow, and painful. That the billions spent acquiring IPG wasn't just buying billings, it was buying a mechanism that makes future billings harder to lose. The synergy story and the retention story become the same story.

To a client, that sentence means something else entirely.

Switching costs are not new in marketing services. Two have defined the industry for decades, and understanding what happened to them explains why a new one is being constructed now.

The first was relationships. The senior team you trusted. The account lead who knew your brand architecture better than most of your own people. The creative director who understood why you'd tried something and failed three years ago, and never made you explain it twice. The strategist whose judgment you'd tested enough to trust without full rationale. Real value. Real protection against churn. Clients stayed not because they were locked in, but because the people were irreplaceable and the thought of rebuilding those relationships with someone new felt genuinely costly.

But this switching cost is fragile. And integration waves are the most efficient destroyer of it. Every CMO watching the Omnicom restructuring, the agency brand eliminations, the 15,000 roles being removed, the senior talent choosing to leave rather than wait for their number, is doing a quiet calculation about the people they chose. How many are still there. How many will be in eighteen months.

The second switching cost was institutional knowledge. The agency holding your briefing history, your research archives, your campaign learnings across ten years, your brand guidelines through fifteen iterations, the tribal memory of decisions made and the context behind them. Moving agencies meant rebuilding context that took years to accumulate, running campaigns that underperformed while the new partner climbed the learning curve you'd already paid for.

Technology has dissolved both of these, in different ways and at different speeds.

The old switching costs were already eroding. The industry needed new ones. And so, as the old architecture weakened, a new one is being built directly into the data layer.

Today, Omnicom told investors about it directly. The sentence wasn't a slip. It was a strategic positioning statement, spoken to the right audience, at the right moment.

Arthur Sadoun, Publicis' 3rd CEO in 100 years, has never, to my knowledge, described Epsilon as a switching cost mechanism. Not once, in any investor call or industry presentation I have seen. In fact, I don't think he has even thought about it as an objective.

Publicis invested $4.4 billion in Epsilon in 2019. They integrated it now as CoreID, the foundational identity layer beneath everything the holding company does. They built it into creative, into media buying, into measurement. They developed outcome-based pricing models on top of it, because controlling the full data stack meant they could price on results rather than time. They positioned it, consistently and coherently, as something they built because genuinely understanding a client's customers better than anyone else is how you earn the right to stay.

The switching cost is a byproduct of doing irreplaceable work. Not the objective. The logic runs: if we know your customers better than you do, if our data makes your marketing materially better, if our platform is woven into how your business measures commercial outcomes, then you won't want to leave. Not because you can't. Because the work we do together is worth more than the disruption of leaving.

Today, Omnicom inverted that framing. They told investors that the architecture they're building, RealID integrated into Omni, connected to media buying, to attribution, to campaign performance, will make it harder for clients to leave. That's a legitimate commercial strategy. Building genuine platform dependency through data integration has worked for every SaaS business that has ever achieved scale. The formula is clean and well understood: acquire users, deliver value, generate recurring revenue, raise switching costs, defend the base. Repeat. Salesforce built a trillion-dollar market capitalisation on exactly this logic. Adobe bought its way into it. The platforms have perfected it.

The question for the marketing services industry is whether clients are users or partners. And whether that distinction matters when the contract comes up for renewal. Because the answer to both questions is yes, but the power to act on it depends entirely on knowing which tier you're in before the conversation starts.

The honest purpose of this newsletter has never been to tell you what to think. It's to give you the analytical framework to ask the right questions before someone else's framework is imposed on you. So here are the three that matter.

Where do your customer IDs actually live? In your own data infrastructure or in your agency's platform, enriched and processed through their architecture? This is not a technology question to delegate to your data team. It is a strategic question that belongs in the CMO's remit, because the answer determines what your next agency transition actually costs — in time, in money, in marketing performance degradation, and in the months of rebuilding that will happen before anyone in your organisation publicly admits the transition is harder than expected.

What happens to your measurement baseline if you change partners? Can you take your attribution history with you? Your audience segments, your lookalike models, your performance benchmarks, the data that tells you which channels actually drove revenue versus which ones claimed to? If the honest answer is "it depends on how the contract is written," the practical answer, in most cases, is no. Contracts written at signing rarely survive the full commercial reality of departure. The data portability clause that looked reassuring in the legal review looks very different when a migration actually begins.

Who enriched your data, and what did you provide them to do it? If your agency improved your match rates by combining your first-party data with their identity graph, your dataset is now more valuable, and more capable, than it was before the relationship began. Congratulations on the improved performance. You may also have materially increased what it costs to leave, because the version of your data that now drives your best results exists partly inside their architecture.

Investor days are carefully designed to reassure. Synergies on track. Platform differentiated. Revenue well-protected. Every word calibrated for people who want certainty about future cash flows and downside protection on a significant capital position.

"Switching costs" was one of the most honest things Omnicom said today. Not because the strategy is wrong, building genuine lock-in through data infrastructure is a rational, defensible commercial response to a market in which relationships and institutional knowledge no longer provide adequate protection against churn. The old moats are crumbling. New ones need to be built. That's real.

But the framing revealed the orientation. When a company describes its client data architecture to investors as a mechanism that increases the cost of client departure, the investor is the primary audience and the client is the asset being described. That is a legitimate investor relations strategy. It is also a signal worth noting if you happen to be one of the assets.

The agencies that will win the next decade are the ones whose clients never feel the need to ask those three questions. Not because they are prevented from asking. Because the work is genuinely irreplaceable, the data strategy has been built for client outcomes rather than client retention, and leaving never seriously enters the conversation.

There is a version of data integration built for client outcomes. And a version built for investor confidence.

Today, Omnicom told you which version they're building.

Brandflow is written by Justin Billingsley, who has spent his career on all three sides of the industry's table: senior client, global agency leader, technology founder. First published 12 March 2026 in the Brandflow newsletter on LinkedIn.