A Signup at the Root Says You Sell One Thing

We built a self-serve signup flow — plan selection, verification, payment — and put it at /sign-up.
It worked. It was also wrong, and the reason is worth more than the fix: a URL is a claim about what a company sells. A signup at the root of the domain says "there is one account here, and this is how you create it". For a company with one product that is accurate. For a company with a catalogue of services and several distinct products, it quietly misrepresents the business.
Worse, it strands the other products. Once /sign-up belongs to one of them, the question "where does
a customer buy the second one?" has no good answer.
The symptoms, before anyone names the cause
This kind of mistake announces itself as a set of small awkward problems:
- the signup page needs a heading explaining which product you are signing up for
- navigation cannot decide whether to link "Sign up" globally or only from one section
- the product's own pricing page needs a link out to a page that lives nowhere near it
- analytics shows a signup funnel that is disconnected from the product pages feeding it
- a second product gets
/sign-up-2, or/register, or a modal, because the obvious name is taken
Each is individually small enough to patch. Together they are the information architecture telling you it is wrong.
The fix: put the purchase where the product is
The flow moved to live under the product it sells:
/infrastructure/telecommunication/ucaas/pricing ← the table, and the entry point
/infrastructure/telecommunication/ucaas/sign-up ← the flow, under the product
And the pricing table became the only way in. Each tier's call to action deep-links into the signup with the selection already made:
/infrastructure/telecommunication/ucaas/sign-up?plan=pro&cycle=annual&seats=25
That single change fixed every symptom at once. The page no longer has to explain which product it is — the URL and the breadcrumb already did. Navigation does not need a global signup link, because signup is a step inside a product journey. Analytics shows a funnel that runs from the product page to the purchase, which is the funnel that actually exists.
Carry the selection, do not restate it
If someone has configured a plan on the pricing page, the signup form should open on that configuration — not ask again. Deep-link the parameters and validate them on arrival:
const tier = params.get("plan");
if (isMarketingTier(tier)) setPlan(PLAN_SLUG[tier]); // allow-list, not a cast
if (params.get("seats")) setSeats(clampSeats(params.get("seats")));
Validate rather than trust. A hand-typed ?plan=enterprise should select nothing if Enterprise is
sales-only, instead of silently putting a quote-only tier into a checkout.
Leave the old URL alone if something else owns it
Ours had a legacy page at /sign-up serving a different purpose. We restored it exactly as it was,
kept it out of the sitemap, and left it noindex. Reusing a URL because it is convenient is how the
original mistake happens in the first place.
The general rule
Ask what each URL asserts about the business, independent of what the page contains:
| URL | Claim |
|---|---|
/sign-up | There is one account for this company, and this creates it |
/products/x/sign-up | This is how you buy product X |
/pricing | The company has one price list |
/products/x/pricing | This is what product X costs |
/contact | There is one way in |
/products/x/get-started | This is how you start a conversation about X |
Neither column is universally right. A single-product company should absolutely own /pricing — that
is accurate and it is the strongest possible URL for the term. A multi-product company that takes
/pricing for one product has told every visitor and every crawler something false about its
catalogue, and has nowhere to put the second price list.
The test is simply whether the claim is true. If you sell several things, the root level describes the company; the products describe themselves.
Doing this without damaging what already ranks
Moving a page that has history is a real risk. We moved a page with none — the flow was days old — which made it cheap. If you are relocating something established:
- 301 the old URL to the new one, permanently, and leave it in place indefinitely
- update every internal link to point at the new URL directly, not at the redirect
- keep the canonical on the new URL only
- update the sitemap in the same deploy
- expect a settling period and do not interpret the first fortnight as a verdict
And if the page in question is converting well where it is, weigh that honestly against the structural tidiness. An accurate URL is worth real money over years. It is not worth breaking a working funnel in a week. The right time to fix this is before the page has history — which, in practice, means thinking about it on the day you create the route.


