// growth
SaaS Marketing Agency for B2B Software Companies
The decision you are trying to make is probably framed as a channel question: fund more content, fund more paid, hire someone for lifecycle. It is worth checking whether that is the decision at all, because nothing in a single-channel report can settle it. Paid can be hitting its cost per lead. Content can be up on sessions quarter over quarter, and lifecycle email can hold a respectable open rate. Revenue sits where it sat three quarters ago. The price of acquiring a customer only means something next to how long that customer stays, and no channel report can see that far.
So the scope here is the whole system. Positioning and the offer, the mix of channels and the order they get funded in. The trial or demo path, activation, retention, and the reporting that ties it to pipeline.
The breadth is also the boundary, and it cuts both ways. Testing and rebuilding what happens after the click, along with the experiment apparatus behind it, is conversion rate optimization. Designing and building the marketing site itself is web development, over in the build hub. Running one paid channel properly is media buying. All three are linked at the top of this page, and this engagement does not take any of them over. It decides what those surfaces have to achieve and in what sequence. Then it holds the number they add up to.
Why a subscription funnel is a different object
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.
- The limiting stage is usually not the top. Traffic added above a leaking stage arrives, converts worse, and costs more. Finding the leak first is what separates spend that compounds from spend that evaporates.
- Two motions run at once. Self-serve signups and sales-assisted deals share a website, a price list and a reputation. They need different proof at different moments. Treating them as one funnel produces copy that serves neither.
- Activation is marketing. What happens in a new account’s first session decides whether the acquisition cost was ever recoverable. That work is product engineering. It sits inside this engagement, and it does not get thrown over a wall.
What whole-funnel actually means here
Positioning and the offer
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.
Demand
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.
Activation and retention
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.
Measurement
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.
What we turn down
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.
How this fits with the rest of the work
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.
What you get
- A funnel model of your actual business, from first touch to renewal, with the stage that is currently limiting revenue identified and evidenced
- A positioning and messaging document written from customer interviews and your own win/loss notes, not from competitor copy
- A quarterly plan of named experiments, each with a hypothesis, a cost, a decision date and the condition that kills it
- A channel mix with an order of play: which channel is funded this quarter, which is held back, and the evidence that would promote it
- A lifecycle map from signup through activation to renewal, naming each trigger, the message it sends and the stage it is accountable for
- One definition of a qualified lead and an activated account, shared by every channel, so the readout's numbers add up to the same business
- A weekly readout stating what moved, what did not, and which tests we stopped
- Handover documentation for every system we build, so the engagement can end without the machinery stopping
How it runs
- 01
Model the funnel before proposing a channel
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.
- 02
Make the numbers mean one thing
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.
- 03
Run experiments with kill criteria written first
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.
- 04
Route the winners to whoever ships them
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.
- 05
Report against pipeline, not activity
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.
Questions we get asked
- How is this different from your media buying page?
- 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.
- We already have a marketing team. What is left for you?
- 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.
- Can you commit to a CAC or a pipeline number up front?
- 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.
- What if churn is the real problem and acquisition is fine?
- 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.
- We are pre-product-market-fit. Should we hire you for this?
- 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.
- Do you need access to our codebase and our analytics?
- 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.
- Who owns the work if we stop the engagement?
- 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.