The SaaS marketing framework

Most marketing frameworks fail in the same way. They describe a shape, nobody owns a piece of it, and four months later the deck sits in a Notion folder while the team argues about starting a podcast. The model below has five layers, and every one of them ends in a document with a date and a name on it. If you cannot point at that document, the layer is not done.

The five layers, and the artefact each one produces

Market, message, motion, channels, measurement. That order is not decorative: each layer sets the constraints for the one under it, so you cannot score a channel before you have settled the motion, and you cannot write a message before you know who you are excluding.

Layer	Question it answers	Artefact it produces	Owner at Series A
1. Market	Who are we for, and who are we refusing	ICP document with a disqualification list	Head of marketing, co-signed by sales
2. Offer and message	What do we sell, and why this over the alternative	Message hierarchy, one page per segment	Product marketing
3. Motion	How does a buyer actually reach paid	Motion memo: self serve, sales led or hybrid	Founder or CRO
4. Channels and campaigns	Where do we show up, and what does a fair test cost	Channel plan with test budgets and kill dates	Demand generation lead
5. Measurement	What do we believe happens next quarter, and why	Marketing model, a spreadsheet not a dashboard	Marketing operations

Channels are the fun layer, which is why so many teams start there. The result is a well-run paid program aimed at a buyer who was never going to renew. Everything in SaaS marketing sits on top of these five questions: tactics turn over every eighteen months, the questions do not.

Layer one: market definition, and the ICP document that ends it

The artefact is a two page ICP document listing the firmographic, technographic, behavioural and situational filters an account has to clear, plus the segments you are deliberately refusing. It is a spending filter, not a poster.

Build it from data you already have rather than a workshop. Pull the last 40 closed won accounts and the last 20 churned ones into one sheet with columns for employee count at signup, industry, the tool they replaced, who signed the contract, how they first arrived, and whether they were still paying at month twelve. The pattern is usually visible in an afternoon, and it is usually narrower than the story the team tells itself.

Two companies selling to the same job title can have completely different ICPs. Clay’s buyer is a go to market operations person who is comfortable writing conditional logic and enrichment waterfalls. Attio’s buyer is more often a founder or revenue operations lead replacing a CRM that got too heavy. Same revenue team, different account shape, different qualification bar, and therefore different channels three layers down. Our ICP template for SaaS has the exact field list if you want to skip the blank page.

Situational triggers deserve their own line in the document. Funding rounds, leadership hires, failed audits, new site openings and competitor price changes each create a window where budget and urgency exist in the same month, and an account list filtered on a trigger converts several times better than the same list filtered on company size alone.

The finishing test is blunt. If the document does not exclude a segment that currently pays you, it is not finished. Every company has a segment that closes easily and churns at month five, and naming it in writing is the single highest return hour in this entire framework.

Layer two: offer and message, and the hierarchy that ends it

The artefact is a message hierarchy: one positioning sentence, three claims that support it, and a named proof asset under each claim. One page per segment, and no claim ships without proof attached.

Superhuman has run on the same idea for years, that the product is the fastest email experience made, and the proof underneath it is measurable keyboard-first interaction rather than adjectives. Linear sells speed and opinionated defaults to product engineering teams, and the proof is the product itself, which is why their marketing leans so hard on the interface. Vanta sells time to a compliance certificate, and the proof is continuous control monitoring plus an auditor network. Each of those is a hierarchy, not a tagline.

Pricing and packaging belong in this layer, which is why the old 4 Ps of SaaS marketing still earn a place in a modern model. Change a price point and you have changed the message, the qualification bar and the viable channel set, all at once.

The hierarchy has a second job people forget: it is the brief for everybody else. A freelance writer, an agency, a new SDR and a conference booth all pull from the same three claims, and consistency across those surfaces is what makes a small company feel established. Six versions of the value proposition across six surfaces reads as a company that has not decided anything yet, and buyers notice that faster than marketers expect.

Write the hierarchy against the alternative the buyer actually considers, which is usually a spreadsheet, an internal build, or doing nothing. Naming a competitor when the real rival is inertia produces copy that answers a question nobody asked.

Layer three: motion, the decision most teams skip

The artefact is a one page memo saying how a buyer reaches paid: self serve, sales assisted, sales led, or an explicit hybrid with a named handoff threshold. Most teams never write it down, then spend two years running a sales led cost structure on top of a self serve price point.

The economics of software as a service set the boundary. Recurring revenue means acquisition cost is repaid out of a margin stream over months, not recovered from one invoice, so the annual contract value decides how much human contact you can afford per deal. As a rough working rule: under about $5K annual contract value, a human in every deal does not survive contact with the payback math. Above roughly $30K, no human in the deal usually means no deal at all. Between those, you are running a hybrid, designed or accidental.

Hybrid is the honest answer for most B2B products, and it is also the hardest to operate, because it needs a written threshold. Slack and Figma both grew bottom up and then bolted an enterprise sales layer on top of accounts that had already adopted the product. That works. What does not work is an undeclared hybrid, where sales chases every self serve signup and marketing keeps optimising for trial volume that sales then ignores. The mid market SaaS marketing playbook covers the threshold question in more depth, because mid market is where undeclared hybrids do the most damage.

Layer four: channels and campaigns, where the plan meets the bank balance

The artefact is a channel plan with six columns: channel, the hypothesis in one sentence, the test budget, the time to first reliable signal, the kill date, and the owner. Anything that cannot fill those six columns is an idea, not a channel.

That number is the reason this layer is hard. Most of the buying process happens with nobody from your company in the room, so your channels have to work unattended: a comparison page that answers the objection you would have handled on a call, a review profile that survives a procurement check, a template someone finds while doing the job your product does. If you want raw material, SaaS marketing ideas is a working list, but run each one through the six columns before it gets budget.

The discipline here is subtraction. Two channels funded properly beat six funded at a quarter each, every time, and the arithmetic is not close. A test that cannot reach about 30 conversions inside its window has not produced evidence, it has produced anecdotes, and anecdotes are how teams keep zombie channels alive for three quarters.

Campaigns sit inside channels rather than beside them, and enforcing that distinction keeps the plan readable. A channel is a durable route to a buyer with its own economics. A campaign is a time-boxed set of assets pointed down one or more channels, with a start date and an end date. Teams that treat every campaign as a channel end up with a fourteen row plan, none of whose rows has a cost per opportunity attached to it.

One rule keeps this layer honest: a channel cannot enter the plan without a named owner who has time for it. A two person marketing team can operate two channels properly. Adding a third does not add a third more output, it subtracts from the first two, because briefing, creative review and reporting each take a fixed slice of somebody’s week regardless of how small the budget is.

Layer five: measurement, and the model that ends it

The artefact is a marketing model: a spreadsheet with inputs on the left, assumed conversion rates in the middle, and a forecast on the right. Not a dashboard. Dashboards report the past, and the job of this layer is to state what you believe will happen and then be wrong in a way you can learn from.

Keep it to about fifteen rows. Sessions, signups, qualified accounts, opportunities created, opportunities won, average contract value, and the conversion rate between each pair. When the quarter closes, put actuals next to assumptions in the same view and write one line explaining each gap. That single habit does more for forecasting accuracy than any attribution tool you can buy, and the SaaS marketing plan template is built around exactly this structure.

The handoff between marketing and sales gets measured here too, which is why the sales and marketing SLA template belongs in layer five rather than in a separate process document. An SLA without a number in the model is a promise, and promises do not show up in a quarterly review.

The model is also where you settle the argument about what marketing is accountable for. Pick one of marketing sourced or marketing influenced pipeline as the headline number, define it in the model, and report the other as a secondary line. Running both as headlines gives you two figures that disagree, and a quarterly review where the disagreement becomes the topic.

Accuracy is not the goal in the first two quarters. A model that forecasts pipeline within 25% and explains its own misses beats one that hits the number by accident, because the explanation improves the next forecast. Write the one line gap explanation even when the quarter went well, since a beat you cannot account for is the same problem as a miss you cannot account for.

Route the symptom to a layer before you rebuild anything

When growth stalls, the instinct is to redo everything. Resist it. Almost every symptom points to one layer, and fixing the wrong layer costs a quarter you will not get back.

Symptom	Layer most likely at fault	The test that confirms it	First fix
Low traffic and low everything else	Four, channels	Your best channel has never had a properly funded test	Fund one channel to 30 conversions before adding a second
Traffic holds, signups do not follow	Two, message	Five customers describe the product differently from your homepage	Rewrite the hierarchy against the real alternative, then the page
Signups arrive, revenue does not	Three, motion, or one, market	Signups cluster in a segment sales cannot economically serve	Set a contract value threshold and write the motion memo
Revenue closes, churn hits by month four	One, market	Churned accounts share a firmographic pattern with each other	Add that segment to the disqualification list and stop paying for it
Sales calls marketing leads junk	One plus the SLA	Marketing and sales use different definitions of qualified	Agree one event per stage and publish both definitions
Everything works but nothing is forecastable	Five, measurement	Nobody can state last quarter assumptions from memory	Build the fifteen row model before buying any tooling

What this costs, and the three ways it breaks

Building the five artefacts properly takes about three weeks of calendar time for a team with clean closed won records, and closer to six weeks if the CRM has been neglected. That time comes out of campaigns that would otherwise be running, which is worth saying out loud before anyone starts.

The first failure is rot. An ICP document is usually wrong within two quarters of a price change, and a channel plan goes stale the moment a paid platform rewrites its auction rules. Dates on documents are the cheapest defence available, because an undated artefact gives nobody a reason to reopen it.

Diffuse ownership is the second failure. Five artefacts owned by a committee become five documents everybody references and nobody maintains. One name per layer, written at the top of the file, is the entire fix, and it works because it creates a person who feels something when a stale number gets quoted back at them.

Layer one carries a political cost, and that is the third failure. Somebody’s best logo lands on the refusal list, a rep points out that their largest deal last year came from an excluded segment, and the exclusion quietly disappears from the next version. Put an exception process in the document instead: excluded accounts can be worked with a director signature, and the exceptions get reviewed at quarter end. A framework nobody argues with is a framework nobody is using.

What changes at seed, Series A and Series C

The layers stay the same. Depth, ownership and cadence change, and getting that wrong is why seed teams drown in process while Series C teams run on founder intuition that expired years ago.

At Series C the framework mostly stops being about decisions and starts being about consistency: three segment teams, three ICP documents, one message hierarchy that has to hold them together. If you want to see all five layers filled in for a single company, the worked SaaS marketing strategy example runs the whole chain end to end, and SaaS marketing examples shows how public companies express each layer on their own sites.

Where AARRR, the bowtie and jobs to be done fit

They fit inside the layers rather than replacing them, and treating any one of them as a whole strategy is a reliable way to lose a year.

AARRR, the acquisition, activation, retention, referral and revenue model Dave McClure popularised around 2007, is a measurement lens. It belongs in layer five. It is good at telling you where users drop and silent on who you should have targeted.

The bowtie funnel from Winning by Design extends the same lens past the close into onboarding, adoption and expansion. Also layer five, and worth adopting if you sell an annual subscription with meaningful expansion revenue, because a classic funnel that stops at closed won will happily report a great quarter in a business that is quietly losing net revenue.

Jobs to be done sits in layer two, with message. It is a method for understanding the switch a buyer is making and the alternative they are firing, which makes it the best input to a message hierarchy and a poor input to a channel plan.

A fourth model comes up in every second planning meeting, so it is worth placing too. Pirate metrics, the bowtie, jobs to be done and the flywheel are not competing frameworks, they are lenses at different altitudes. The fastest way to end an unproductive framework debate is to ask which of the five layers the model in question produces an artefact for. If the answer is none, it is a thinking tool rather than an operating one, and thinking tools do not need a row in the plan.

Start with the two layers you can finish this month

Do not attempt all five at once. Build the two artefacts that unblock the rest, then work down.

Put a review date on each document before you close the file. Six months from now, the version of you that has to explain a missed quarter will want to know which assumption broke, and an undated artefact cannot answer that question.