Don't save at the wrong end! Why a website relaunch costs you online visibility

In short: The most expensive mistake in a website relaunch, a domain consolidation or a merger does not happen on launch day. It happens months earlier — the moment a decision is made without taking stock. An analysis of 892 domain migrations found that organic traffic took an average of 523 days to return to pre-migration levels, and that 17% of sites had not recovered even after 1,000 days (Search Engine Journal). Whatever gets saved on preparation comes back several times over as correction.

Why does a relaunch cost visibility when everything is supposed to get better?

Because a new website is built for people, while visibility is attached to addresses. Every indexed URL carries signals built up over years: rankings, external links, usage signals, mentions in AI answers. Those signals do not stick to the brand — they stick to the specific address. Replace the address without passing the value on cleanly, and the value disappears, no matter how much better the new page looks.

Google’s official site-move documentation names exactly the points where things break: every old URL belongs redirected individually to its new counterpart, not in bulk to the homepage. You should keep control of the old domain for at least 180 days. And you should not make several major changes at once — rebuilding structure, content and navigation in one go makes it impossible, for users and search engines alike, to trace what caused a later drop.

The underlying pattern is always the same: the relaunch changes more than was recorded beforehand. You can only protect what you know about.

What does skipping the preparation actually cost?

Considerably more than the preparation itself — and that is not industry folklore but decades-old documentation. In Software Engineering Economics, Barry Boehm showed that the cost of correcting a defect rises with every project phase: what costs almost nothing during requirements costs a multiple once the system is in operation. The famous 100:1 ratios come from large projects; on smaller efforts the range is closer to 5:1. The direction is identical in both cases.

Translated to a web relaunch: a missing redirect costs ten minutes before launch. After launch it costs an analysis, a fix and a re-indexing — with weeks in between during which the page cannot be found. A forgotten noindex left over from staging costs one glance at a checklist beforehand and a month of visibility afterwards.

This is where the real price question sits. The cheaper quote or the lower hourly rate is rarely the cheaper total. What gets saved during the concept phase reappears during implementation as extra work — and in operation as lost inquiries. The difference: the saving is written in the quote, the cost is written nowhere.

Which mistakes show up in almost every project?

Surprisingly few, and they repeat. Drawn from an inventory we carried out for a corporate group with six web presences across four countries, plus the patterns from comparable projects:

• The starting point was never measured. Without a zero point, nobody can say after launch whether anything was lost. The discussion becomes a matter of opinion.
• Content exists but cannot be reached. In that project, one brand had fully optimized landing pages that were not linked internally anywhere. Another had dozens of job listings with no internal link at all. It is rarely the substance that is missing — it is access to it.
• The technology blocks what the rules permit. Four of the six domains allowed every crawler in robots.txt — and still served them a captcha wall instead of the page, Googlebot included. The rule and the actual response had drifted apart, unnoticed for months.
• Legal leftovers from templates. One live site carried another company’s complete legal notice, another carried their terms and conditions, and on a third the terms page consisted of placeholder text. All residue from a shared template origin.
• Functional errors at the most expensive spot. A homepage’s main call to action led to an error page. Every prospect who followed it was lost — and demonstrably nobody had noticed.
• Foreign link profiles. On several domains, backlinks appeared with identical sales anchor text, some added just days before the measurement. Whether bought or planted from outside cannot be determined externally — but the consequences differ substantially.
• Wrong country signals and inconsistent multilingual setups. A Swiss domain was sending search engines the country signal for Germany. Three sites handled multiple languages with a standard tool, three not at all.
• Load times nobody had measured. On mobile, the values ranged from several seconds to more than half a minute until the main content appeared — on a site that promises applying by smartphone.

Not one item on this list is exotic. All of them were found within days. And none would have been cheaper to fix after the relaunch.

Why is the sequence more expensive than the structure?

Because the wrong sequence doubles the effort — and that decision gets made whether it is spoken out loud or not. Centralize existing structures before the concept and content are settled, and you migrate what will shortly be replaced. The migration effort is incurred twice.

That does not mean everything has to wait for the concept. A three-way split works:

1. What runs immediately, independent of everything else: legal defects, functional errors, crawler access, unregistered brand domains. None of these are touched by a structural or agency decision. They cost hours, and their only prerequisite is access.
2. What can be centralized right away because the relaunch does not replace it: hosting, security, update regime, backups, domain and DNS control, license management, governance. This layer sits beneath the sites and is not invalidated by new websites.
3. What waits for the concept because it would otherwise be built twice: page structures, templates, navigation, language logic, content.

That distinction settles the “centralize first or design first” argument with: both, on different layers.

What else happens when several companies come together?

A layer appears that a single relaunch does not have: the question of who is actually who online. After a merger the group exists on paper first. Online it exists only once the relationship between the companies is visible and machine-readable.

In the project described, exactly one link was measurable between six corporate domains — and it led from a Christmas invitation to a PDF. None of the sites marked up the corporate relationship in structured data, and no legal notice named the parent company. Even so, AI systems assigned the group affiliation correctly in four out of five test answers — citing LinkedIn, the commercial register, press portals and the website of the law firm that had handled the acquisition. Asked the other way round, starting from a subsidiary brand and without naming the group, the rate collapsed; for the smallest brand, to a third.

That is the direction that matters in the market: a buyer checks the subsidiary, not the holding company. And it is an unstable state — it rests on press coverage, and press coverage ages.

The second finding of the same kind concerns authority. The strongest signals sat with the subsidiaries, not with the newly registered parent brand. Merging onto the parent domain would have moved established authority in the wrong direction and bought twelve to eighteen months of rebuilding — for a result achievable without merging domains at all, through linking, structured markup and legal notices. That calculation is only possible if you measured first.

And the third, most awkward one: two actively operated sites did not appear in the agreed brand architecture at all. You do not find that in a workshop. You find it by comparing the strategy document against the actual inventory.

Which analyses belong before the decision? The checklist

This list is the actual core. It applies to relaunches, to consolidations and to portfolios with many individual sites — the scope scales, the sequence does not.

Phase 0 · Take stock before anything is decided

• Site and domain inventory: every domain, subdomain, country variant, microsite and forgotten campaign page — with owner, registrar, expiry date, host, CMS and responsible person
• Visibility balance per site, including at least twelve months of history: which terms rank, and do they describe your actual business?
• Traffic and conversion contribution per site and per URL cluster, organic separated from other channels
• Authority and link profile balance per domain, including spam share and the origin of suspicious links
• Content inventory with value assessment: what pulls, what is dead, what exists but cannot be reached
• Technical inventory: indexability, orphaned pages, redirect chains, structured data, mobile load times, accessibility
• Check what crawlers actually receive — not just robots.txt, but what the server really returns to individual user agents
• AI visibility: is the brand named in AI answers, which sources back it up, and which addresses do those sources point to?
• Local visibility per location and company, consistency of name, address and phone number across directories
• Brand space: how often is each brand searched by name? That determines whether a brand may be retired
• Legal and organizational: trademark rights, available and third-party brand domains, legal notices and policies, licenses, contracts, notice periods, access credentials

Phase 1 · Target picture

• Target architecture with alternatives costed out: independent domains, central platform or hybrid — each with effort, risk and rationale
• Reconciliation with brand strategy: does the planned structure support the agreed brand architecture, or quietly override it?
• A search-intent map instead of an org chart: navigation follows the user’s question, not the reporting lines
• Responsibility matrix of market × topic field × brand, so your own brands do not later crowd each other out
• Language and country matrix per brand, settled before anything is built

Phase 2 · Delta and risk

• URL-by-URL mapping from old to new, including deliberate “this will be deleted” decisions
• Redirect concept: one-to-one, permanent, no chains, no bulk redirect to the homepage
• Risk list with likelihood, severity and countermeasure
• The forgotten side theatres: email addresses, tracking and consent, campaign links, QR codes in print, inbound links from partners

Phase 3 · Action catalogue

• Prioritized by effort and impact, not by the order of items in the quote
• Timeline, ownership and dependencies per measure
• A clear split: before launch, at launch, after launch
• Acceptance criteria with target values that belong in the contract — regardless of who implements

Phase 4 · Measurement framework

• Freeze the baseline before launch: rankings, traffic, conversions, links, AI mentions — with tool, date and source, so every measurement is repeatable
• Fixed checkpoints after launch, tightly spaced in the first two weeks
• Abort and rollback criteria, in writing
• Ownership of the follow-up phase — the mistakes surface long after the implementation partners have left

This is exactly where weooo works: inventory, target picture, action catalogue and measurement framework — evidence-based, with documented measurements. We do not have to handle the implementation for that, and in many projects we deliberately do not. Whoever runs the analysis without depending on the build budget can also recommend building less.

How do you spot a quote that gets expensive later?

By what is missing from it. Four questions for any provider before you sign:

• How will the current state be recorded before launch? If there is no concrete answer, there will be no yardstick afterwards.
• What does the redirect concept look like, and who builds the URL mapping? If it only comes into being “at go-live”, it comes too late.
• Which acceptance criteria apply, with which target values? Load time, accessibility, indexability and structured data belong in the contract with numbers, not in the rework.
• What happens in the first 90 days after launch, and who pays for it? Without a follow-up phase, nobody is there when the effects become visible.

A quote that covers those four points is rarely the cheapest. It is usually the one with fewer invoices attached to it later.

When is this kind of analysis overkill?

It can be, and that deserves saying plainly. For a single site with a few dozen pages, no meaningful organic share and no established link profile, a full inventory is out of proportion. There, a clean URL mapping, a baseline and an indexability check are enough — one or two days.

The opposite failure exists too: analysis as postponement. If you have not built anything after the third round of concepts, you do not have a knowledge problem but a decision problem. Good preparation ends in decisions, not in further questions.

The threshold where full preparation pays off can be named: more than one site, more than one language or market, established rankings and links, or a noticeable share of inquiries from organic search. If more than one applies, the analysis is the cheapest line item in the entire project.

Conclusion

A relaunch is not a design project. It is a move of assets you cannot see. Fail to take stock of them beforehand and you cannot protect them — and you notice the loss only when the inquiries stop and nobody can say why. The figures are unambiguous: more than a year and a half to recover on average, and nearly one site in five never recovering at all. Against that stands preparation that costs days and produces the basis for every later decision. Saving at the wrong end means exactly this: the saving is written in the quote, the bill arrives later.

Facing a relaunch, a consolidation of several sites or a merger? Talk to us — we do the inventory and the target picture before anything gets built. Whoever wants to can implement it afterwards.

Sources

• Google Search Central, “Site moves with URL changes” — official guidance on site migrations: page-by-page redirects instead of a bulk redirect to the homepage, retaining control of the old domain for at least 180 days, no simultaneous major changes to structure, content and navigation, partial move as a test on large sites.
• Search Engine Journal, analysis of 892 domain migrations — an average of 523 days to regain pre-migration organic traffic; 17% of sites not recovered even after 1,000 days; fastest recoveries where complete 1:1 redirects covered every indexed URL.
• Barry W. Boehm, Software Engineering Economics, 1981 — cost-of-correction curve across project phases; the frequently cited ratios of up to 100:1 come from large projects, while smaller efforts sit closer to 5:1.
• German Accessibility Strengthening Act implementing the European Accessibility Act, applicable since 28 June 2025 — accessibility requirements covering websites and customer portals, with exemptions for fundamental alteration of the service or disproportionate burden.
• Own inventory analysis of a corporate group with six web presences across four countries, 2026 — reproduced anonymized, without names and without exact measurements.