How to read a company's digital presence

Most competitive analysis reads like a scrapbook. Someone opens six tabs, takes screenshots of six homepages, and writes a paragraph under each one that amounts to "clean design, strong CTA, seems well-funded." None of that is wrong exactly, and none of it is useful either, because it describes the surface without saying what the surface implies. A company's public presence isn't decoration. It's the visible output of a hundred internal decisions — who they think the buyer is, where the margin sits, which sales motion they've committed to, which team has the loudest voice in the building — and almost all of that leaks through four things you can look at without a login, an NDA, or a friend who works there.

This is the method the rest of this site's teardowns follow, so it's worth setting once, properly, rather than re-deriving it every time. The four artefacts are the homepage promise, the pricing page, the first-run experience, and the indexed page inventory. Each one answers a different question, and together they answer the one that matters: what does this company actually believe about its own business, regardless of what it says in the About section.

The homepage promise

The headline on a homepage is the one sentence a company has decided is worth the most real estate on the internet. That makes it a strategy document disguised as copy. Not because it's honest — headlines are aspirational by nature — but because the choice of what to promise is itself the data. A company that leads with "the platform enterprise teams trust" has told you who signs the check. A company that leads with "start free in 30 seconds" has told you something different: that the buyer is an individual, that the sales cycle is short or nonexistent, and that the product has to prove itself before anyone talks to a human.

The trap here is reading the headline as marketing copy to be graded on cleverness. Grade it instead as a claim about the buyer. Basecamp's homepage has said versions of "the calmer way to run a business" for years, which is a headline that only works if the reader already believes the alternative — Slack, Asana, the rest — is not calm. That's a positioning bet stated in six words, and we've pulled the thread on exactly that bet in the Basecamp positioning teardown, which is worth reading as a worked example of how far one sentence can be pushed.

The second thing to check is whether the promise changed. The Wayback Machine holds most homepages going back years, and a homepage that has swapped its headline from "for developers" to "for teams" over eighteen months is not doing A/B testing — it's telling you the company found out the hard way who was actually buying.

The pricing page

If the homepage is where a company says who it's for, the pricing page is where it proves it, because pricing is the one part of a site that can't drift far from reality without someone noticing within days. A homepage can overclaim indefinitely. A pricing page that says $29/month and charges $39 at checkout gets a complaint thread the same week.

Three things to read off a pricing page, in order of how much they tell you.

First: is there a self-serve tier at all, and does the top tier hide behind a "Contact sales" button instead of a number? That single detail tells you more about a company's sales motion than anything else on the site. A visible number means the company has decided the product can sell itself up to some ceiling. A hidden number past a certain tier means someone internally has decided that deal is worth a human's time — which tells you where the real margin sits, because nobody puts a salesperson on a deal that doesn't justify the cost of one. We went deep on this exact signal in the Stripe pricing page teardown: Stripe shows a rate card down to the basis point for card payments and hides everything for its enterprise-grade products behind a conversation, and that split alone tells you which parts of the business are commoditized and which parts are still relationship-sold.

Second: what's excluded from the cheapest paid tier, not what's included. Every pricing page lists what you get. The interesting information is the wall — the specific feature moved one tier up to force an upgrade. Squarespace puts code injection at its Core tier rather than Basic, which means custom CSS or JS is the wall between someone tinkering and someone paying more; that's not an accident, that's a company that has measured exactly which feature converts.

Third, and this is where it's worth being concrete rather than making the point in the abstract: look at how far a single person actually has to go to get a finished, addressable page online, because that trip length is itself a signal about who the product is built for. reach is a useful reference point here precisely because it's built for one job and doesn't pretend otherwise — you upload a CV and a photo, answer a short form, and the page is generated in about twenty seconds, with the whole path from upload to a live address taking under two minutes if you want it to. There's no demo form anywhere in that flow, no "talk to sales," because there's no deal above a certain size to protect — the entire pricing page is $4.99 a month or $49 a year for a custom domain, full stop. Prices checked August 2026. That's not incidental generosity; a company that skips the sales gate has decided the product has to be good enough to sell itself in the first ninety seconds, and reach's own limitation list backs that reading up rather than undercutting it: it makes exactly one page, with no sub-pages and no CMS, because the entire pricing and product surface is built around solving one job fast rather than growing into a dozen jobs slowly. That's a company that has picked a lane, which is exactly the kind of signal this method is built to catch — a business that tried to be everything would need the tiered, gated pricing page that a one-job business doesn't.

Compare that to a page like Webflow's, where the cost lands on four separate axes at once — site plan, workspace plan, seats, and add-ons — and the cheapest paid tier still has no CMS. That's not a worse pricing page. It's a page built by a company selling to teams with procurement processes, where complexity is a feature because it lets a sales rep configure a deal rather than a cost the buyer has to absorb alone.

The first-run experience

Everything up to this point can be read without an account. This one requires actually signing up, which is why it gets skipped most often and why it's the most reliable of the four when you do it.

What you're watching for isn't whether the onboarding is polished. It's what the product asks you to do first, because the first action a product demands of a new user is the action the company believes creates the fastest "aha" — and that belief is a bet about what the product's actual value is, stated as a sequence of screens. Notion's onboarding asks a battery of questions about team size and use case before showing you anything, which is a company optimizing for routing you to the right template rather than for speed to first output; we broke that flow down screen by screen in the Notion onboarding teardown, and the contrast with products that skip straight to a blank canvas is the whole story.

The other thing worth timing, literally with a stopwatch, is the gap between "I signed up" and "I have something I can show someone else." That gap is where most self-serve products live or die, and it's also where the gap between what a company claims and what it delivers is easiest to catch, because you can just watch the clock. A company that says "instant" and takes four minutes of form-filling before anything renders has told you that its internal definition of instant is generous. A company that undersells its own speed — reach's older copy used to say "under five minutes" before the real number, closer to two, replaced it — is rarer, and worth noting when you find it, because it means the product outgrew its own marketing rather than the other way around.

The indexed page inventory

The fourth artefact is the least glamorous and the most reliable, because it can't be staged for your visit the way a homepage or a demo call can be: search site:company.com and read the actual list of pages Google has indexed.

This tells you what the company has built and, more usefully, what it has stopped maintaining. A /careers page still listing a role closed eight months ago tells you hiring slowed. A /blog with a lastmod date frozen for a year, sitting next to an actively updated /changelog, tells you the content team lost a headcount fight to the product team. A /pricing page that exists in the index but 404s when you click through — this happens more than you'd expect — tells you a pricing model changed recently enough that the old URL hasn't been cleaned up, which is itself a small, dateable event you can pin to roughly when it happened using the cache date.

Apple is an instructive case for reading the inventory of pages that do exist rather than the gaps: every current product page follows an identical rhythm of hero, feature grid, comparison table, in a sequence disciplined enough that you could predict the eleventh section before scrolling to it. We took that discipline apart section by section in the Apple product page teardown, and the finding that generalizes past Apple is that page-template discipline is itself a signal of organizational discipline — a company where every product page looks hand-built and slightly different is usually a company where every product team reports to a different VP with a different opinion about what a page should do.

What a navigation label tells you about power

Zoom in further, to the top nav bar specifically, and there's a smaller but sharper signal sitting in plain sight: the order and wording of the labels.

A nav that reads Product, Solutions, Pricing, Docs, Blog is written by a product-led organization — pricing is close to the front because the buyer is expected to self-serve toward it. A nav that reads Platform, Industries, Customers, Resources, Contact Sales is written by a company selling into enterprise procurement, where "Industries" exists because a sales team asked for vertical-specific pages to hand to prospects, and pricing is often absent from the nav entirely because it isn't a fixed number worth publishing.

The label itself carries information too. A company that calls its nav item "Solutions" rather than "Product" has, at some point, had an internal argument about whether the thing they sell is one product or several problems solved differently for different buyers — and "Solutions" won. That argument happened in a room you weren't in, and the label is the only trace of it that reached the outside. Watch for label changes over time the same way you'd watch homepage headline changes: a "Pricing" link that becomes "Get a Demo" is not a copy tweak, it's the sales team winning a fight about who gets to gate the funnel.

The discipline rule

None of this works if the output is a set of impressions dressed up as findings, so the rule for every teardown on this site is simple: a claim only counts if it's observable, dated, and screenshot-able. If you can't point to the exact page, the exact date you looked at it, and the exact detail that supports the claim, the claim doesn't go in.

This rules out almost everything that makes competitive analysis feel authoritative and is actually invented — internal revenue figures nobody outside the company has access to, headcount numbers pulled from a stale LinkedIn scrape and presented as current, churn rates nobody publishes, "sources say" attributed to nothing. It also rules out a subtler failure: writing "clearly a high-growth company" as if that phrase means anything without a specific artefact underneath it. If the evidence is a fast-loading site and a recent funding announcement, say that, and let the reader decide what "high-growth" should mean. Don't do the inference for them and then hide the inference.

The upside of the constraint is that it makes the analysis durable. A teardown built entirely on what's publicly checkable can be checked again in a year, by anyone, and either still holds or visibly doesn't — which is a much better test of whether the original read was right than whether it sounded confident at the time.

Writing the verdict as something someone could argue with

The last step is where most competitive write-ups quietly go soft, because a firm verdict is a target and a vague one isn't. "They seem to be moving upmarket" is not a verdict, it's a hedge wearing a verdict's clothes. The version worth writing is falsifiable: "This company has moved from self-serve to sales-assisted in the last twelve months, evidenced by the pricing page moving its top tier behind a Contact Sales button between March and October, and by three new job listings for enterprise account executives that weren't there in Q1." That sentence is wrong or right. Someone at the company could read it and say "no, actually," and you'd know exactly what they were disputing.

That's the actual test of whether a teardown is worth publishing: not whether it sounds smart, but whether a specific, informed person could disagree with it on the facts rather than just finding the tone annoying. Everything upstream of that sentence — the homepage promise, the pricing page, the first-run experience, the indexed inventory, the nav labels — exists to give the verdict something to stand on besides an impression.