Measure this visit?
Accept allows Google Analytics, Microsoft Clarity and PostHog, including session recording. Reject stops their recording. Cookieless counts and server request logs stay on. What we measure
Growth & Demand
Whole-funnel marketing for subscription software. Positioning, demand, trial activation and retention run by one team that also writes the code.
Updated
We run the whole funnel for a subscription software business. That means positioning and the offer, which channels get funded and in what order, the trial or demo path, activation and retention. It also means the reporting that ties all of it to pipeline, which is the number this engagement is measured on.
It does not take over single channels. Conversion rate optimization, web development and media buying each have their own page, linked above.
Why the whole system? Because every channel report can look fine while revenue sits where it sat three quarters ago. Paid hits its cost per lead, content is up on sessions and lifecycle email holds its open rate. The price of acquiring a customer only means something next to how long that customer stays. No channel report sees that far.
In a transactional business, the sale settles the question. In a subscription business it opens one. The same signup is profitable or ruinous depending on whether that account is still paying in month nine. Acquisition decisions cannot be evaluated inside the window a marketing report uses.
That has consequences all the way down, and they shape everything we do.
We interview your customers, read your win/loss notes, and write positioning from what buyers actually say. Category language copied from competitors is how four companies end up describing themselves identically and then competing on price.
We choose channels from the model, not from fashion. Some businesses are search constrained and some are trust constrained. Some sell to engineers who will never read a landing page, and who need documentation and a working free tier. The mix is an output of the diagnosis.
We look at the first session inside the product, find where new accounts stall, and get the fix specified and scheduled. Onboarding email, empty states, and the moment a trial user first gets something useful are all inside scope. None of them are reachable from a landing page.
One model of the funnel, reconciled against billing, with a single definition of a qualified lead and an activated account that every channel reports into. Client-side analytics under-reports for every visitor who blocks scripts or declines consent. A funnel model built on it misprices channels in a direction you cannot predict. What we report against is first-party and server-side. The instrumentation a particular experiment needs is built by whoever runs that experiment. What this engagement guarantees is that the channel reports and the billing system describe the same business.
Leads we cannot tie to revenue stay out of the readout. A channel we cannot measure does not get bought. A brand awareness retainer with no exit condition is not something we will sell you. A test does not stay alive because stopping it would read as an admission.
Two of those are better read as conditions, because they decide whether you should hire us at all. If the evidence says the constraint is retention, the recommendation is retention work. The acquisition version of this engagement is larger and far easier to sell. If what you need is a cost per acquisition quoted before we have seen your data, we are the wrong shop. A number invented in a pitch can only be met later by redefining what counts as a lead.
The same team writes the software and the marketing. A fix that turns out to be a product change does not need a second vendor, a second contract and a month of translation. That is the whole argument for buying both from one place. It is also the reason we are comfortable saying the honest thing about which half your problem is in.
We map every stage from first touch to renewal and find where revenue is actually lost. Adding traffic above a broken stage buys you a more expensive version of the same problem, so we refuse to start at the top.
Before any money moves we agree one definition of a qualified lead and of an activated account, and reconcile the funnel model against billing. Channels reporting on four incompatible definitions is how a business argues about dashboards instead of about revenue.
Each test states its hypothesis, its budget and the result that ends it, agreed before launch. Deciding what counts as failure after seeing the data is how teams keep funding channels that never worked.
A result that survives becomes work with a named owner and a date. A page test for the conversion team, a site build for the web developers, an in-product activation change for the engineers. A recommendation with nobody attached to it produces nothing.
The weekly readout is sourced pipeline and what changed it, including the tests that lost. Reporting impressions and sessions makes an engagement look healthy while the business does not move.
Media buying is one channel, bought and optimised as a channel, and it is a reasonable thing to hire on its own. This engagement owns the whole funnel, which means positioning, the mix of channels, the trial or demo experience, activation, and retention. If your paid account is the only thing that needs work, buy that page instead and keep your money.
Usually the connective tissue. A content person, a paid person and a lifecycle person can each be doing their job well. With no shared model of the funnel and no agreed definition of a qualified lead, each optimises their own metric and revenue stays flat. We can run the model and the measurement while your team keeps executing. And if the honest answer is that you should hire someone instead of retaining us, we will say that.
No. Any number we produced before seeing your data would be priced off your competitors and our own hopes. That is a sales artefact. It is not a forecast. What we commit to up front is the measurement system, the cadence of experiments and a decision date on each one. After a quarter of real data, a forecast has something underneath it, and you can hold us to that one.
Then we will tell you, and we will not start by buying traffic. Acquisition spend on a product people leave converts cash into a larger pool of unhappy former users. Retention work is inside the funnel we own, so this is not a referral elsewhere, but it does change what the first quarter looks like.
Probably not for this engagement. Whole-funnel work assumes there is a repeatable motion to scale. Before that, the useful work is talking to users and shipping product. A retainer aimed at demand generation will mostly produce evidence you already had. We would rather say that than take the contract.
Yes to both. Modelling the funnel needs read access to your analytics, your billing system and your CRM. The fixes this work produces are usually code, so the repository matters too. That holds even when the pull request comes from the conversion team or your own engineers. If access is impossible we can write specifications instead, but unimplemented recommendations are the usual reason these engagements produce nothing.
You do. Everything lives in your repository, your ad accounts, your analytics and your email platform, under your credentials. We document what we built as we build it. An agency that keeps the machinery hostage has an incentive problem, and you should assume it affects their advice.
Start a conversation