Running a brand agency, a web agency and a paid-acquisition agency in parallel often costs more than a single integrated setup, not because of what each one charges, but because of three costs that appear on no invoice: coordination, broken attribution and lost context. These costs are not on any quote, and that is precisely why you pay them without ever seeing them.
The heaviest of the three is measurable. Observational attribution, the method every siloed vendor leans on, overstates a channel’s real effect by a factor of roughly 3 to 13, according to 663 large-scale experiments analysed by Gordon, Moakler and Zettelmeyer. Three inflated local metrics do not add up to one honest result. This article prices the hidden costs of a fragmented vendor stack, and it is just as clear about when keeping separate specialists is the right call.
Is stacking three agencies actually the safer choice?
On the surface, yes. Handing the brand to a brand specialist, the website to a web specialist and the ads to a paid-acquisition specialist looks like the most rational decision available. You pick the best on each line, you compare quotes discipline by discipline, and every single choice defends itself in isolation.
That is the trap. This reasoning treats three interdependent workstreams as if they were independent. It optimises each part and never optimises the assembly. In B2B, the brand feeds acquisition, the website converts what acquisition brings in, and acquisition surfaces the objections the brand has to answer. The parts talk to each other, or they should.
When they do not, a cost appears. It does not replace the price of the work. It stacks on top, quietly.
Why does the coordination cost explode without showing up?
Coordination cost explodes because the number of links to maintain between people grows far faster than the number of people. Adding a vendor does not add one link. It adds one for every vendor already in the room.
The number of communication channels in a group follows a simple rule. For n people, there are n(n-1)/2 possible links. Three vendors create three links. Six vendors create fifteen. Twelve would create sixty-six. The principle is old: it is the logic behind Brooks’s law, stated by Fred Brooks in 1975, which holds that every person added to a project raises its coordination burden disproportionately.
No link is free. You have to align a brief, settle a disagreement, forward a file, re-explain context. At three vendors it stays manageable. At five or six, coordination becomes a full job in its own right, one nobody was hired to do and one you end up doing yourself.
This is not a study of agencies. It is a structural principle, observed anywhere distinct parties have to produce a shared result. Transaction cost economics, from Ronald Coase to Oliver Williamson, Nobel laureate in 2009, describes exactly this: coordinating several external parties creates costs the market never puts on a bill.
The cost of context, the quietest of all
On top of coordination sits a cognitive cost that is easy to ignore. Every time you leave your own job to arbitrate between two vendors, you interrupt yourself. And the interruption has a measured price: it takes an average of about 23 minutes to return to a complex task after a distraction, according to research by Gloria Mark at UC Irvine.
Those 23 minutes are not the length of the interruption itself. They are the time it takes to rebuild the thread of what you were doing before. Multiply that by the number of round trips that coordinating several vendors demands, and you have measured a cost that falls first on the most valuable person in the company: you.
The real problem: nobody measures the whole chain
The heaviest cost is neither coordination nor lost time. It is broken attribution. In silos, each vendor optimises the metric it answers for, and nobody optimises the business result. The ad agency optimises cost per click, the web agency optimises the landing page conversion rate, the brand agency optimises awareness. Each one wins on its own number. The revenue belongs to no one.
The problem is not only organisational, it is mathematical. Attribution, the practice of crediting a channel for a sale, is notoriously unreliable when it rests on observation rather than experiment. Analysing 663 large-scale experiments, Gordon, Moakler and Zettelmeyer found that non-experimental methods overstated the true advertising lift by a wide margin, roughly 3 to 13 times depending on the stage of the funnel and the method used.
The authors are blunt about how far this goes. Even with rich user-level data and hundreds of experiments to check against, they report being “unable to reliably estimate an ad campaign’s causal effect.” In other words, the number each vendor presents in the monthly review is not a small approximation. It can be an overstatement by an order of magnitude.
Add three overstatements together and you get a picture where every channel appears to work while the overall result stalls. Nobody is lying. The setup simply produces a sum of local truths that never adds up to one global truth. We unpack the same mechanism in our piece on why GA4 and Meta conversions never match: three tools, three totals, none of which reconcile.
The brand-consistency number you should stop quoting
There is one statistic that gets wheeled out to justify a unified setup, and it deserves to be retired. Consistent branding, the claim goes, lifts revenue by 23%. It circulates in countless decks as proof that everything should sit under one roof.
That number is not evidence. It comes from Lucidpress, now Marq, and its State of Brand Consistency reports, a self-reported survey of marketers rather than an audited study. It even circulates in two incompatible versions, 23% and 33%, which is enough to disqualify its precision.
We could have used it, because it points our way. That is exactly why we set it aside. Building an argument on a fragile figure weakens the whole argument. Brand consistency has real economic value, but it is proven by the logic of memory and repetition, not by a percentage nobody can trace. The same discipline runs through our analysis of whether branding actually lowers CAC: the real mechanism is more solid than the easy number.
When multiple agencies are the right call
Integrated is not always better. Claiming otherwise would be a sales pitch, not an analysis. A single integrated setup becomes more economical in one specific case: when growth is run as a system, with workstreams that depend heavily on one another and one shared business result to optimise.
Outside that case, a dedicated specialist is often the more rational choice. Here is when consolidation adds nothing:
- The need is one-off and self-contained. Redesigning a logo, producing a batch of visuals, fixing a specific technical bug: none of these has a strong dependency on the rest of your marketing.
- The skill is rare and narrow. Some highly specialised expertise is easier to find in a top independent than in a generalist structure.
- You already have the referee in-house. A company with a strong marketing lead can coordinate several specialists itself, because it owns the overall result and absorbs the coordination cost knowingly.
So multiple vendors are not a problem by nature. They become a problem when interdependent work is treated as if it were independent, with nobody holding the full chain. What never works is not having several experts. It is having several experts and no referee. We weigh the same trade-off in our comparison of in-house versus agency B2B marketing.
Is consolidation worth it?
It is worth it when interdependence is real, and the waste of a fragmented setup shows up beyond agency fees. Even in tooling, fragmentation quietly burns money: Gartner found that marketers use a shrinking share of what they buy, with martech stack utilisation falling from 58% in 2020 to 42% in 2022 and 33% in 2023, according to MarTech’s reporting on the survey. Sprawl is expensive whether the sprawl is vendors or software.
An integrated setup does not remove the disciplines. It removes the costly joints between them. Brand, website and acquisition stay three distinct expertises, but they share one strategy, one message and, above all, one result metric. The question “who is right?” disappears, because one party answers for the revenue.
Concretely, three costs collapse. Coordination, because the links no longer cross company boundaries. Attribution, because the chain is measured end to end instead of measured three times, locally and inflated. Context, because you stop being the liaison between vendors who do not talk to each other. That is the logic behind our B2B growth firm model: not doing more things, but running the same things as one system.
In short
- Check interdependence before you split the work. If your marketing workstreams feed each other, handing them to separate agencies creates coordination and attribution costs that appear on no quote.
- Look for the referee, not just the expert. Multiple vendors work when someone owns the overall business result. Without a referee, three excellent specialists produce a sum of local truths, not a result.
- Keep the specialist for the isolated job. A one-off, self-contained need is served well by a dedicated specialist. Integration earns its keep when growth is run as a system, and not before.
Not sure whether to stack vendors or unify your setup? Book a diagnostic and we will map, with your own numbers, where your current joints are costing you the most.