Fragmentation is not reach
27th August, 2026
There’s a reason supermarkets charge brands for shelf space.
More facings mean more visibility. More stores mean more chances to be seen, considered and bought. For most of marketing’s history, distribution was a game of occupying space: another shop, another billboard, another reseller, another point at which you might intersect with somebody at exactly the right moment.
The web removed the shelf, but we kept its logic. Worse, for a while the web rewarded us for doing so. Businesses are still carrying the architectural and organisational baggage.
More pixels, more profit
Domains are cheap. Pages are effectively infinite. Search engines made visibility look like real estate: every result was another place you might occupy. If one website could rank for a query, perhaps two could. Buy a competitor, keep its website, and now you own another route into the market.
There was enough truth in this to make the strategy seductive. So large organisations accumulated websites: the corporate site, product site, business division, consumer brand, professional version of the consumer brand, acquired company nobody quite knew what to do with, campaign microsite which somehow became permanent, regional sites, innovation hub, content platform, and the separate site for whatever the CEO was particularly excited about in 2018.
Each came with a plausible story about reach. More brands meant more audiences. More domains meant more chances to rank. More content meant more keywords. More sites meant more links between those sites, which meant more authority to move around.
More shelves. More products. More chances to win.
Except search stopped behaving like a supermarket.
Google has little incentive to fill a result set with several versions of the same company saying roughly the same thing. An ambiguous search might warrant a shop, a review, a news story, a video, a guide and an official source. Those are different answers to different possible needs. Five websites owned by the same organisation, with overlapping propositions and broadly interchangeable content, are five candidates for the same answer.
AI accelerates that shift. A search results page still has room for several visible winners. A generative system can retrieve from a much larger pool and compress what it finds into a few paragraphs. An agent may go further and choose a product, provider or service on your behalf. Not every journey will collapse to one winner, obviously, but the direction is clear: the source pool gets larger while the visible answer gets smaller.
At that point, there is not much shelf space left to acquire. What matters is whether, amongst all the available evidence, you are the clearest and strongest thing to select. Another place to appear is useless if the system has no reason to choose another version of you.
Splitting one proposition across several websites can divide its reputation, links, content, recognition, investment and evidence across weaker versions of itself. Each must become distinctive, authoritative and useful enough to survive independently, whilst customers and machines are left to work out how they relate.
Sometimes the separation earns its cost. First, though, we should be honest about the bill.
The fragmentation tax
A new website needs hosting, templates, analytics, governance, maintenance, security, accessibility, performance work, tracking, reporting, content, SEO, legal review and somebody who remembers how the bloody thing works when it breaks.
Those are merely the visible line items. The more damaging cost is dilution.
The same engineers now support another platform. The same content team feeds another publishing schedule. The same PR function promotes another brand. The same SEO team inherits another set of templates, technical problems and search opportunities. None of those resources multiplied when the domain was registered. They were divided.
This is the fragmentation tax: every new boundary strands some proportion of the organisation’s existing capability on the other side.
A sufficiently valuable property can pay for the extra machinery. The problem is the long tail: sites which never receive enough investment to become excellent, but remain important enough that nobody will kill them.
So they decay slowly. Templates age. Content drifts. Tracking differs slightly from the corporate standard. Product descriptions stop matching the main site. Navigation develops peculiar local rules. The backlog fills with problems which would already have been solved on a shared platform.
Then the organisation builds a design system to make the sites look related, data feeds to keep product information aligned, cross-domain analytics to reconstruct the customer journey, single sign-on to disguise the joins, and shared navigation to make the estate feel less fragmented.
We separate things, then spend money teaching them to behave as though they were connected.
Consolidation compounds
Fragmentation taxes every change. Consolidation makes improvements compound.
Fix a template once and thousands of pages get better. Improve performance, accessibility, structured data or analytics in one place and more of the estate benefits. One author profile can accumulate years of evidence instead of being recreated across five sites. One taxonomy can connect products, expertise, support and editorial content instead of forcing each property to maintain its own little model of the world.
Put crudely: 2 + 2 = 5, while 1 + 1 + 1 + 1 = 3. A connected system creates value in the joins. A fragmented one pays to recreate them.
The same finite team gets more leverage. The optimisation surface shrinks while the reach of each change grows. That is not merely a saving. It is compounding advantage.
This is not a promise that moving everything onto one domain will magically improve rankings. Consolidations can destroy useful distinctions, botch migrations and create a vast corporate landfill with excellent canonicalisation. The advantage comes from reducing needless boundaries while preserving the differences which matter. The implementation still has to be competent.
Then the politics arrive. Separate properties acquire separate owners, targets and incentives. A business unit wants visibility. A regional team wants control. An acquired company wants to preserve its identity. Nobody wants their site to become a subsection of somebody else’s, because ownership of a website feels suspiciously like ownership of importance.
Architectural boundaries harden into organisational ones. Consolidation becomes a negotiation over status, budgets and territory. The estate persists not because it serves customers, but because it accurately represents where power sits inside the company.
Sprawling web estates are rarely just collections of websites. They are maps of the compromises that created them.
When different isn’t different enough
The awkward cases look different on an org chart and almost identical from the outside.
- Acme
- Acme Business
- Acme Enterprise
- Acme Professional
- Acme Services
- Acme Solutions
To the people who created them, these distinctions may feel profound. Each has a strategy deck, a target audience, some carefully differentiated messaging and perhaps even a different shade of blue. To everybody else, they are variations on Acme.
The closer two propositions are semantically, the more work their separation creates. Customers must understand which one applies to them. Search engines must determine whether they are different entities, different parts of the same entity, or several sources describing roughly the same thing. AI systems must reconcile their names, claims, products and relationships before deciding which one is relevant.
Audience segmentation is a repeat offender. Business customers may need different pricing, proof points, purchasing journeys, support and language. None of those things inherently requires another domain. They may simply require a good /business/ section.
The distinction that matters is whether this is genuinely a different thing, or merely another route through the same thing.
Naming makes it worse. Suffixes such as “Business”, “Pro”, “Enterprise”, “Solutions” and “Services” ask users to learn an internal taxonomy which is usually corporate soup outside the organisation. Where does “Professional” end and “Business” begin? Is “Enterprise” a larger version of “Business”, or a separate product? Is “Solutions” a product family, a consulting arm, or just the word somebody reached for after discovering that “Services” was already taken?
If you need a brand architecture diagram to explain the names, the names are not doing enough work.
Somebody searching for “Acme business login” does not care which division owns the account system. They do not want to reason about whether they need Acme Business, Acme Enterprise or the business section of Acme. They want the bloody login.
Every ambiguous boundary introduces a decision the customer did not ask to make. Increasingly, it introduces one a machine has to make too.
Semantic distance should usually determine architectural distance. Things which are genuinely far apart can benefit from stronger boundaries. Things which are close together usually benefit from being represented as parts of a coherent whole.
Regulation, geography, ownership, acquisition strategy and operational constraints can force different choices. Fine. The burden of proof should still increase with every new boundary. A URL path is considerably cheaper than a new brand.
The principle applies below the brand layer, too. A marketing page, login screen, dashboard, support centre and application may need different experiences and technology. They do not all need to compete as public discovery surfaces. Technical separation does not require search separation. Give people and machines one obvious public doorway, then route them appropriately once their intent is clear.
If customers recognise something independently, seek it out by name, understand why it exists and benefit from the separation, perhaps it has earned the boundary. If one team merely sells to companies with 500 employees and another to companies with 5,000, perhaps we can keep the domain registrar out of it.
Authority cannot be photocopied
Once several overlapping properties exist, some inevitably become stronger than others. That is usually when somebody suggests linking them all together.
The corporate site is strong. The acquired brand is weaker. The regional site needs help. So the stronger properties point towards the weaker ones, the weaker ones point back, and somewhere in the resulting web of crosslinks there is supposed to be a compounding effect.
Except nothing has been multiplied. Whatever reputation, links, recognition and trust the organisation has earned still has to come from somewhere outside the system. Moving signals between four websites does not create four times as much independent evidence.
Authority can be distributed. It cannot be photocopied.
Links between related properties can still be useful. They help discovery, explain relationships and signal importance. But several sites owned by the same organisation endorsing one another are not creating independent corroboration. It is the digital equivalent of moving money between four bank accounts and congratulating yourself on having quadrupled your wealth.
The architecture becomes circular: create several sites; weaken them by dividing attention and reputation; link them together to compensate; then treat those links as evidence that the division was worthwhile. You have built a dependency loop and mistaken it for strategy.
If Acme.com has spent twenty years accumulating links, mentions, navigational demand and recognition, launching AcmeBusiness.com does not duplicate that history. The organisation has split a proposition away from something already understood, then started an expensive programme to rebuild the authority it abandoned.
If the new brand genuinely needs independence, rebuilding reputation is part of the price. If the distinction is weak, all you have done is create another account which needs funding.
Nor is search a neutral leaderboard which simply stacks the ten strongest pages. It tries to assemble a useful set of different answers. A retailer, review, comparison, video and official source each contribute something. A cluster of sibling websites offering roughly the same proposition does not.
Sibling sites can appear together, and search systems will not always understand corporate relationships correctly. Still, owning more domains is not the same thing as owning more intent. Poorly differentiated properties may simply split the evidence which could have made one of them the obvious winner: the best links point to one, the strongest content lives on another, and the freshest product information sits somewhere else.
Machines can often reconstruct the relationship. Making them do so is still a strange strategy.
Stop asking how many properties you can make relevant to a query. Ask how many meaningfully different needs you can satisfy. Modern discovery rewards diversity of value, not diversity of domains.
From ranking to selection
AI makes the compression harsher because the interface no longer has to show a list of destinations at all.
Traditional search was relatively generous. A results page might include organic links, ads, shopping, maps, videos, news, snippets, forums and whatever else Google had decided to bolt on that week. There were lots of ways to appear, which made visibility feel spatial.
A generative system can retrieve from hundreds of sources and return three sentences. Put five overlapping properties into that process: one has the strongest brand recognition, one the best documentation, one most of the links, one the freshest product information, and one the detailed explanation of the service the user asked about.
In a conventional search journey, several might appear. In an AI-mediated journey, the system may simply decide what Acme offers, whether it is relevant, and whether it belongs in the answer. The carefully separated properties become ingredients in one judgement.
They can also disagree: different terminology, names, product descriptions, detail and claims about the same service. Five versions of what the company apparently is. The machine must do the integration work the organisation declined to do itself.
Agents make this more unforgiving. An answer can fudge ambiguity; an action eventually has to pick something. If a system is asked to compare products, request a quote, find availability, make a booking, call an API or purchase something, it has to decide which product, provider, catalogue, account or version of Acme to use.
That is a different failure mode from ranking third. You can still get traffic from position three. You cannot be the third service an agent chooses when it only needs one.
The real-estate metaphor finally breaks. The emerging web is not giving us smaller shelves. It is replacing the shelf with a decision.
One well-understood entity, with a clear proposition, coherent evidence, consistent data and obvious routes to action, is easier to choose than several overlapping entities carrying fragments of the same story. The old approach maximised opportunities to appear. The emerging one must maximise the probability of being selected.
Not every journey ends with one winner. AI systems can recommend several products, search engines can cite multiple sources, and agents can present options before acting. But thousands of sources can become an answer which mentions three; hundreds of suppliers can become a shortlist of five; five can become one action. That is winner-takes-most territory.
This will not stop at websites. The fragmentation has moved into the machine layer: tools, services, feeds, APIs and agents which overlap just enough to create ambiguity without creating useful choice.
There is no prize for making the decision harder.
There is a good version of fragmentation
Some boundaries earn their keep.
A company with genuinely different products, audiences or market positions may be better served by separate brands and experiences. Procter & Gamble does not need Tide, Gillette and Pampers to pretend they are one thing. A luxury product should not inherit the commercial framing of the budget version. A regulated financial service may need to behave very differently from the consumer app owned by the same parent.
Distance can create value. The mistake is turning every internal distinction into a boundary.
Large organisations have a habit of letting one decision drag the others behind it. Different messaging becomes a new domain. Autonomy becomes a new CMS. A recognisable acquisition survives indefinitely. A repositioning exercise becomes another platform. Before long, organisational structure has become web architecture.
A useful multi-brand estate is not a pile of independent islands with a footer linking them together; it is a system. Each property needs a clear reason to exist, a proposition distinctive enough to support its own identity, sufficient investment to build its own reputation, and relationships which are obvious to customers and machines.
The important bit is what stays shared. Several brands can share product data, identity systems, analytics, infrastructure, governance and content operations while presenting genuinely different customer experiences. The surface can vary without recreating the machinery underneath it.
Brand architecture, information architecture, discovery architecture and technical architecture do not need to mirror one another.
Most sprawl happens when they do. A new brand gets a new website and CMS, which needs its own integrations, templates, analytics and deployment process. Then the product data drifts, so another integration is built to reconnect the systems which were separated in the first place. A branding decision has quietly become an operating model.
The sane version is federated where difference creates value and consolidated everywhere else. A brand should earn its independence through meaningful difference, not inherit it because somebody once drew a box around it on an org chart.
Every boundary must earn its existence
I am not proposing one enormous corporate website. That would be a different kind of failure.
Some things deserve independence: a brand serving a distinct market, a product with its own identity, a business unit under different legal or commercial constraints, or an acquired company whose recognition is worth more than the tidiness of folding it into its parent.
But independence should be earned. The default question is not “why shouldn’t this have its own site?” It is: what gets materially better because this exists separately?
- Does it make the proposition materially clearer?
- Does it help customers understand what they are buying?
- Does it occupy a meaningfully different market position?
- Does it create a genuinely different kind of answer in search?
- Does it give machines a clearer entity, product or service to understand and select?
- Does it enable legal, commercial or operational behaviour which consolidation would genuinely constrain?
If the answer is no, the boundary is probably ornamental. And ornamental boundaries are expensive. They duplicate technology, governance, content, analytics, reputation-building and search effort, with more places for everything to drift.
If you cannot point to the return, you are paying the fragmentation tax for theatre.
Semantic distance should determine architectural distance. Not “what structure gives us the most SEO coverage?” Not “which team owns this?” Not “can we get another result in Google?” But “what is this thing, and does it deserve to be understood independently?”
The economics have inverted. Publishing space is infinite. Clarity, authority, recognition, reputation and consistency are not. Every arbitrary boundary spends some of those scarce resources and gives evidence, links, content and attention another place to leak.
When boundaries reflect real differences, that cost can buy useful specialisation. When they do not, you are paying to make your strongest signals weaker.
Consolidation is not a crusade for one website. It is a demand that every boundary justify the compounding advantage it gives up.
The goal is to maximise the strength, clarity and selectability of the things worth keeping, not the number of properties you can point at on a slide.
Fragmentation is not reach.
Sometimes it is just fragmentation.