How to build a marketing strategy from scratch.
Most marketing strategy decks fail because they start at the wrong end. They open with channels and budgets and conclude with a customer who feels suspiciously like the founder. A strategy that survives contact with reality starts somewhere else. This is the order of operations, with worked examples.
Strategy is the discipline of saying no. Almost any marketing tactic can produce some result somewhere, which is why the absence of strategy looks so much like having one. The five steps below give you a thinking pattern that protects against doing the most expensive version of nothing.
Why most strategy documents fail
Walk into the average mid-market company and ask to see the marketing strategy. You will get a deck with sections titled "Goals," "Channels," "Budget," and "KPIs." Somewhere on slide three there is a persona named Sarah who is 34 and likes podcasts. The document is internally coherent and externally useless, because every section was written without rigorous answers to two earlier questions: who exactly are we selling to, and why would they choose us over the alternative.
The five steps below put the questions in the order they actually need to be answered. Skip step one and steps two through five are theater.
Step one: the ideal customer profile (ICP), drawn from real data
The ICP is a precise description of the customer who is easiest for you to win, fastest for you to onboard, longest to retain, and highest to expand. It is not a persona. It is not aspirational. It is a description of a real type of buyer who, looking at your last 24 months of revenue, accounts for a disproportionate share of your healthy customers.
To draft a useful ICP, do this in one afternoon. Pull your customer list. Sort it by revenue contribution, retention length, and net promoter score. Look at the top quartile. Find the three or four characteristics that show up disproportionately: industry, company size, geography, tech stack, buying trigger, the title of the person who signed. Those characteristics are your ICP.
To make this concrete, imagine two software companies selling the same product (a scheduling tool) with two different ICPs.
- Say, for illustration, that Company A's ICP is solo practitioners in healthcare. Channel strategy: Google search ads on high-intent terms, SEO content targeting "best appointment scheduling" queries, partnerships with EHR vendors. Sales cycle: a 14-day free trial with no human contact. Budget split: 70 percent paid acquisition, 20 percent SEO, 10 percent partnerships.
- Company B's ICP: hospital systems with 200+ employees. Channel strategy: account-based marketing to a list of 400 named targets, content marketing for procurement committees, a small in-person events program at industry conferences. Sales cycle: 9 to 14 months with three or four stakeholders. Budget split: 30 percent content, 40 percent sales-led ABM, 20 percent events, 10 percent paid social on LinkedIn.
Same product. Same competitive landscape. Almost nothing in common at the strategic level. The ICP picks the channels, the cycle picks the cost structure, the cost structure picks the team you need to hire. Get the ICP wrong and every downstream decision is wasted effort.
Show us the ICP page. If it has three sentences and a photo of a stock-photo woman with a laptop, the rest of the deck is fiction. Send the team back to the customer data and start over.
Step two: positioning, in seven words
Positioning is the answer to a single question: what category are you in, and why are you the best choice in that category for this specific buyer. April Dunford's framework, laid out in her book Obviously Awesome and in her quickstart guide to positioning, is a useful public reference on this. The whole exercise should produce a one-page document.
A good positioning statement names the category clearly, names the buyer specifically, lists two or three differentiated values (not features), and identifies the alternative the buyer would otherwise choose. If your positioning could just as easily belong to your three nearest competitors, it is not positioning. It is a slogan.
Step three: channel selection, driven by the ICP and the cycle
Channels are not chosen because they are trendy or because the agency you just hired specializes in them. They are chosen because they intersect with where your ICP spends time and how your buying cycle plays out. The math is mechanical.
- Short cycle, high search intent: SEO and paid search dominate. Email supports retention.
- Short cycle, low search intent: Paid social and content marketing drive demand creation. Email captures and nurtures.
- Long cycle, named accounts: Account-based marketing, sales-led outbound, content marketing, and a small events program. SEO is a long-term play, not a near-term driver.
- Long cycle, broad market: Content marketing, organic search, email nurture, and selective paid social to fill the pipeline.
The mistake is buying every channel and underfunding all of them. An early-stage budget usually does better funding two channels well than five poorly. Add a third only when the first two are producing and you have evidence the third will too.
Step four: budget allocation, with realistic ranges
A useful budget is not a wishlist. It is a forecast tied to unit economics. In The CMO Survey's 2026 edition, marketing budgets averaged 9.0 percent of company revenues, the lowest share in several years. Inside that envelope, the allocation across channels is what the strategy actually decides.
Put each new dollar where it will return the most, not where last year's budget happened to go. Most teams add their marketing budget line by line based on what they did before. Strategy is the discipline of saying "we are doing less of that and more of this."
The CMO Survey publishes marketing budgets as a share of revenue each year. Read it before you write your own.
Step five: measurement, decided before launch
A common failure in marketing strategy is deferring the measurement question until after the campaign launches. By then, the tracking is wrong, the baseline is missing, and nobody can tell what worked. The measurement framework gets designed at the same time as the channel plan, not after.
Three measurement layers belong in every strategy. The leading indicators (impressions, clicks, traffic, signups) that tell you whether the activity is happening. The lagging indicators (leads, opportunities, closed revenue) that tell you whether the activity is producing business. The diagnostic metrics (cost per lead by channel, conversion rate by stage, payback period) that tell you what to change when something is not working.
In our experience, six metrics is about right. More than six and nobody reads it. Less than four and you cannot diagnose what is happening.
While you are here
Want a straight answer about your own site?
Tell a founder what you are working on. You will get an honest read on what is worth doing first and what to skip, whether or not you hire us.
Not ready? Keep reading
Putting it together: what a finished strategy looks like
A complete marketing strategy fits on about ten pages. The ICP gets two pages. Positioning gets one. Channel selection and reasoning gets two. Budget allocation gets one. Measurement framework gets two. The remaining pages cover the team you need, the timeline, and the assumptions that, if wrong, would cause you to revisit the whole document.
That last section matters most, and strategy documents often skip it. What would have to be true for this plan to fail? What signals would tell you it is failing? When will you check? Naming the failure conditions out loud is the only thing that converts a strategy from a document into a decision.
The honest disclaimer
A strategy is a hypothesis, not a forecast. It is your best current guess at what will work, written down in enough detail that future you can tell whether the guess was right. The strategies that produce results are the ones that get revisited every quarter and rewritten every year. The ones that get filed and forgotten produce nothing regardless of how good they were on the day they were finished.
The five-step pattern above will not guarantee any outcome. What it will do is force you to answer the hard questions in the right order,
