By Amit Jain · curated with Vinod Kumar Jain · All Frontier Global · 2026-07-05
A marketing plan is the document that says who you are talking to, what you are saying to them, where you are saying it, what it costs, and what it produced. It is not a strategy for the whole business, and it is not a script for a sales team. It is the plan for demand — creating it, capturing it, and proving that the money spent on it earned its keep.
| Document | What it owns | The question it answers |
|---|---|---|
| Marketing plan | Audience, message, channels, marketing budget, campaign calendar, marketing measurement | Who are we talking to, what are we saying, where, for how much, and did it work? |
| Go-to-market strategy | Product-market fit, pricing, packaging, distribution model, the sales motion, and how marketing and sales economics fit together for a launch or a market entry | How does this product reach this market and make money doing it? |
| Sales plan | Quota, territory, pipeline targets, sales process and stages, compensation, sales headcount | How much will the sales team sell, to whom, and how are they organised to do it? |
| Business plan | The whole enterprise: operations, finance, funding, hiring, legal structure, risk | Is this a viable business, and what does it need to become one? |
Before the six parts, a boundary, because the four documents above get confused with each other constantly and the confusion wastes real time in planning meetings.
A go-to-market strategy is the bigger document that decides how a product reaches a market and makes money doing it. It sets pricing and packaging, chooses a distribution model — direct sales, self-serve, channel partners, a marketplace — and decides how sales and marketing divide the work of getting a customer from unaware to paying. The marketing plan sits inside a GTM strategy as one of its instruments. It does not set price. It does not choose whether the company sells through a direct sales force or a self-serve checkout. It executes the demand-generation piece of whatever the GTM strategy has already decided.
A sales plan governs the sales team once a lead exists: quotas, territories, the stages a deal moves through, compensation, and headcount. Marketing hands leads or demand to sales; the sales plan is what happens to them after that handoff. A marketing plan can and should say what a qualified lead looks like and what happens at the handoff, because that boundary is contested more often than any other in the whole document. It should not attempt to set sales quotas or design a territory map — that is a different document with a different owner.
A business plan is the whole enterprise, including the parts that have nothing to do with acquiring customers: the operating model, the finance function, funding rounds, legal structure, hiring across every department, and enterprise-level risk. A marketing plan is a contributor to the business plan's revenue assumptions, not a substitute for it. If a business plan asks "is this company viable", a marketing plan only ever answers a narrower question: "will this spending, on this audience, with this message, produce enough demand at an acceptable cost."
The rest of this page stays inside that narrower question. Where GTM or sales-team management would normally get pages of their own, they get a paragraph here, because a marketing plan that tries to also be a GTM strategy and a sales plan ends up too vague to hold anyone accountable to any of the three.
Before anything else, the plan has to say what it is trying to achieve, in language specific enough that someone can tell, a quarter later, whether it happened. Most plans skip this and go straight to channels, which is why so many of them cannot be judged afterwards.
A business objective is stated in the language of the business: grow revenue in a segment, launch in a new market, improve retention, raise the average deal size. It belongs to the whole company and marketing is only one of the levers that moves it. A marketing objective translates that into something marketing can actually be held to: generate a defined volume of qualified pipeline, grow awareness among a named buyer group, shift the mix of new customers toward a higher-value segment, launch a product to a defined audience by a date. A channel metric is one level down again: rank for a set of search terms, grow an email list to a size, hit a cost per lead on a paid channel, publish a volume of content.
The failure mode running through most weak plans is collapsing these three layers into one. A plan that states its objective as "increase website traffic by a stated percentage" has written a channel metric and called it a business objective. Traffic can rise while pipeline falls, if the traffic is the wrong audience or the wrong intent. The fix is not complicated — it is discipline: write the business objective first, in the CFO's language; write the marketing objective second, in a form that could plausibly move that business objective; and only then choose channel metrics that would, if hit, produce the marketing objective. Each layer should be traceable to the one above it, and a reader should be able to ask "so what" at any layer and get an answer that points upward, not sideways.
Before setting a target for the period, the plan needs an honest account of where things stand across four areas: the market, the customer, the competitor set and the internal capability of the marketing function itself. The market analysis covers what is changing in the category — demand trends, new entrants, regulatory shifts, technology changes that alter how buyers search or decide. The customer analysis is not the same as the persona work covered in part two; here it is about aggregate behaviour — how buying cycles have lengthened or shortened, what channels the audience actually uses now versus a year ago, what has changed in how they research a purchase.
The competitor analysis should go beyond a feature comparison table. What message are competitors leading with, what channels are they visibly spending on, where are they weak, and where have they recently changed tack. A useful competitor section says what a competitor's spending pattern implies about what is working for them, not just what they are doing.
The internal capability audit is the part most plans omit and the one that saves the most later grief. It asks: what can the team actually execute, given its current headcount, skills and tools. A plan that assumes a small team can run search, content, social, events, PR, partnerships and lifecycle email simultaneously at a high standard is not a plan, it is a wish list. Naming the gap — we can do three of these well and the rest poorly, so we should choose the three — is more useful than pretending otherwise.
SWOT earns its bad reputation honestly: most versions of it are a brainstorm dumped onto four quadrants with no connection to what happens next. Used properly, each entry in the strengths and weaknesses columns should be internal and specific to the marketing function — not "we have a great team" but "we have in-house design capacity but no dedicated lifecycle marketer" — and each entry in opportunities and threats should be external and time-bound — not "the market is growing" but "a named competitor has just stopped supporting a legacy product line, creating a window for switchers this year."
The part that makes SWOT useful is the step almost everyone skips: pairing entries across quadrants into implications. A strength paired with an opportunity suggests where to lean in. A weakness paired with a threat suggests where the plan is exposed and needs either investment or a decision to accept the risk. If the SWOT does not produce at least two or three sentences that begin "therefore, this plan should," it has not done its job and can be cut from the document — a SWOT that only describes and never implies is decoration.
The situation analysis and the SWOT are only as good as what feeds them, and the inputs worth gathering are more varied than a single customer survey. Structured customer interviews are the highest-fidelity source but the slowest to run; they surface the actual words a buyer uses to describe a problem, which matters enormously for message-writing later. Surveys scale further than interviews but are only as good as their sample and their questions — a survey sent to existing customers will systematically miss the reasons prospects who did not buy walked away.
Search demand data — what people are typing into search engines, and how that volume has moved over time — is a relatively unfiltered look at what a market is actually asking about, as distinct from what a company wishes it was asking about. Social listening does something similar for informal conversation: complaints, comparisons, and the language people use unprompted, away from a company's own channels. Sales-call notes and win/loss interviews are underused by marketing teams that treat sales as a separate department rather than a research asset — a salesperson who has just lost a deal to a competitor knows exactly why, if someone asks them soon enough after the loss for the answer to still be sharp. Support tickets reveal what already-paying customers struggle with, which is often a better guide to real product weaknesses than anything a prospect will admit to before buying. Web and product analytics close the loop by showing what people actually do, as opposed to what they say they would do in an interview.
None of these sources is sufficient alone. A plan built only on analytics tends to over-index on the behaviour of people who already found the product, missing everyone who bounced off it earlier. A plan built only on interviews tends to over-index on whoever was willing to spend forty minutes on a call, which is not a random sample of the market. The discipline is triangulation: treat a finding as real once it shows up independently in at least two of these sources, and treat a single-source finding as a hypothesis to check rather than a fact to build on.
Once the plan knows what it is trying to achieve and what the situation actually is, it has to decide who it is speaking to and what it is going to say. This is the part most likely to be inherited unchanged from last year's deck, which is usually a mistake, because audiences and their language move faster than plans do.
Segmentation is the act of dividing a market into groups that behave differently enough that they need different messages, different channels, or a different offer. The classic dimensions are firmographic or demographic (size, industry, role, age, location), behavioural (how they use a product category, how often they buy), and needs-based (what problem they are actually trying to solve). Needs-based segmentation is generally the most useful and the hardest to build, because it requires research rather than a spreadsheet of attributes that happen to be easy to query from a database.
Targeting is the decision, made after segmentation, about which of those segments the plan will actually pursue this period, and — just as importantly — which it will deliberately not pursue. A plan that targets everyone has not targeted anyone; the message has to be vague enough to apply to every segment at once, which makes it persuasive to none of them. Naming the segments the plan is walking away from this period, and saying why, is often the single most useful sentence in the whole targeting section, because it gives the team permission to say no to requests that fall outside it.
A persona is a compact description of a target buyer or user, built to make the audience feel concrete enough that a writer or designer can picture a real person rather than an abstraction. Done well, a persona is a research artefact: it is built from the interviews and win/loss conversations described in part one, and it captures the language the audience actually uses, the trigger that starts them looking for a solution, and the objections that stop them buying.
Done badly, a persona is invented rather than researched — someone in a workshop names "Marketing Mary" and gives her a stock photo, a made-up age, and a list of attributes nobody checked against an actual customer. The tell is usually specificity in the wrong places: a persona document that knows a fictional buyer's favourite coffee order but cannot state, in the buyer's own words, what problem sent them looking for a product like this one has been invented, not researched.
The second common failure is that personas get built once, presented once, and then never opened again. A persona that lives in a slide nobody consults while writing copy is not doing any work; it exists to be checked against, and if the team writing this quarter's landing page has not looked at it, it might as well not exist. The fix is unglamorous: keep personas short enough to be read in under a minute, and build the habit of opening the document as a first step before writing anything customer-facing, not as a retrospective justification for copy already written.
Jobs-to-be-done thinking asks a different question than demographic segmentation: not "who is this person" but "what progress are they trying to make, and what did they hire this product to do for them." The framing is useful precisely because it cuts across demographic categories — two buyers of wildly different age, role and company size can be hiring the same product to do the same job, and a persona built purely on demographics would have separated them into different segments unnecessarily.
The practical output of jobs-to-be-done work is a small number of job statements — the situation, the motivation, the desired outcome — that can be tested against real interview transcripts rather than invented in a room. Where it earns its keep in a marketing plan is in message-writing: a message built around the job the customer is trying to get done tends to land better than one built around the product's feature list, because it starts from the buyer's situation rather than the seller's inventory.
A message hierarchy has three layers and they should be built in order, because each layer constrains the one below it. The top layer is a positioning statement: a single, internally-facing sentence stating who the product is for, what category it competes in, what it does differently, and why that difference matters to the buyer. It is not customer-facing copy — it is a compass the rest of the messaging points back to, and it should survive largely unchanged for the length of the planning period even as the copy built from it changes weekly.
The second layer is proof: the specific evidence that backs the positioning claim — case studies, data the company can stand behind, product capabilities, customer quotes, comparisons. Proof has to be built before copy is written, because copy without proof behind it is just assertion, and buyers who have seen enough marketing to be sceptical will notice the gap.
The third layer is the copy itself: the headlines, the ad text, the email subject lines, the sales one-liners, all of which should trace back through proof to the positioning statement. When copy across different channels feels inconsistent — different claims, different emphasis, a different sense of what the product is for — the usual cause is not a copywriting problem, it is a missing or unclear positioning statement that different writers have each filled in differently. Fixing the top of the hierarchy fixes the copy everywhere at once; editing individual pieces of copy without fixing the hierarchy just moves the inconsistency around.
Brand voice is the set of choices that make copy recognisably the company's, independent of what it is saying this week: vocabulary it uses and avoids, sentence length and rhythm, how formal or informal it is, whether it uses humour and how much, how it handles its own mistakes in public. A useful voice guide gives paired examples — this is on-voice, this similar sentence is not, and here is why — rather than adjectives alone, because adjectives like "friendly" or "confident" mean different things to every writer who reads them.
Tone is voice adjusted for a specific moment: the same brand voice reads differently in a product outage notice than in a celebratory launch email, and a voice guide that does not distinguish the two will produce copy that is either flippant when it should be serious or stiff when it should be warm. The practical test of a voice guide is whether a new writer, handed nothing but the guide and no other context, produces copy an existing team member would recognise as on-brand; if it fails that test, the guide is too abstract to be useful.
Message testing means checking, before a message goes into wide circulation, whether it actually lands with the intended audience — and it is different from A/B testing a headline on a live page, though the two are related. The cheapest form is qualitative: showing draft messages to a handful of people from the target audience and asking them to explain, in their own words, what the message is claiming and whether they believe it. This surfaces confusion and disbelief far faster than any quantitative test, because a message that a real prospect cannot paraphrase accurately is not going to work at scale regardless of the click-through rate it eventually produces.
Quantitative testing — running two or more versions of a message against real traffic and comparing a response metric — is useful for optimising a message that has already cleared the qualitative bar, but it has limits worth naming plainly. It needs enough volume to produce a statistically meaningful result, which many channels and many companies simply do not have for every message they want to test. It also only tells you which of the variants tested performed better, not whether a completely different message — one nobody wrote — would have beaten both. Message testing works best as a two-stage process: qualitative checking to rule out the message that does not land at all, then quantitative refinement, at whatever volume is actually available, to sharpen a message that has already cleared that first bar.
This is the part of the plan most people mean when they say "marketing plan" — the list of channels and what happens in each one. It is also the part most prone to becoming a wish list of everything the team could do rather than a reasoned choice of what it will do.
The 4Ps — product, price, place, promotion — is the oldest framework for describing a marketing mix, and it remains useful as a checklist for making sure a plan has not silently assumed away decisions that belong to it: what is being sold, at what price, through what channel to market, and how it is promoted. Its limit is that it was built for physical goods sold through retail, and it says almost nothing about the two areas that dominate most modern marketing plans: the people delivering a service and the digital channels through which most demand is now created and captured.
The 7Ps extends it with people, process and physical evidence — relevant wherever a service or an experience is part of what is sold, which is most B2B software and most consumer services. Even extended, though, both frameworks describe a static mix rather than a sequence of buyer behaviour, which is why most working plans now organise the channel section around owned, earned and paid, or around the funnel stage a channel primarily serves, and keep the Ps as a background checklist rather than the plan's spine.
Owned media is anything the company controls outright — its website, its email list, its product itself as a marketing surface, its community if it runs one. Owned media is durable: once built, it keeps working with only maintenance cost, and it is not subject to a platform's changing rules or rising prices. Earned media is attention the company did not pay for directly — press coverage, word of mouth, organic social sharing, reviews. Earned media is the hardest to plan reliably because it depends on other people's editorial or personal judgement, but it tends to carry more credibility with an audience than either owned or paid, because it was not the company saying it about itself.
Paid media is anything bought outright — search ads, social ads, sponsorships, display. Paid is the most controllable and the most immediately scalable: turn the spend up, and — within the limits of the audience available on that platform — volume goes up with it, roughly in proportion, until the available audience is exhausted or the cost per result rises past what the plan can tolerate. The trade-off across the three is speed versus durability: paid buys speed but stops the moment spending stops; owned takes longer to build but keeps compounding after the initial investment; earned is unpredictable but disproportionately trusted when it lands. A mature plan uses all three deliberately rather than defaulting to whichever one is easiest to turn on this quarter, which in most organisations is paid.
Search splits into two very different disciplines that share a results page. SEO is slow-building and durable — content and site structure earn rankings that keep producing traffic long after the work is done, but it can take months to show results and offers no direct lever to pull for an urgent shortfall this week. Paid search is immediate and controllable — turn a campaign on and traffic arrives within hours — but the moment spend stops, so does the traffic, and it gets more expensive as competitors bid the same terms up.
Content marketing — articles, guides, video, tools — is a demand-creation and trust-building channel more than a demand-capture one; it works over a longer horizon and its returns compound with SEO, but it is genuinely difficult to attribute cleanly to a specific sale, which makes it perennially vulnerable to budget cuts from anyone judging channels purely on short-run attribution.
Social media splits similarly to search into organic and paid, and organic reach on most platforms has become unreliable enough that it should not be planned as a primary demand-capture channel; it remains useful for brand presence, community-building and distributing content that was created for another purpose. Email and lifecycle marketing is one of the few channels the company owns outright end to end, and it is unusually good at nurturing an audience that already knows the company toward a decision, though it depends entirely on a list that has to be built and kept clean elsewhere.
PR and communications shapes how the company is described by people who are not the company — journalists, analysts, commentators — and it is slow to build relationships for and impossible to fully control the output of, but the credibility it produces is hard to buy any other way. Events and field marketing — from an owned conference to a shared trade-show booth — are expensive per contact but produce a depth of engagement and a speed of relationship-building that most digital channels cannot match, which is why they persist in B2B categories with long, considered sales cycles despite the cost.
Community — a space where customers or prospects talk to each other, not just to the company — is slow to build and does not scale by simply spending more money on it, but a genuine community becomes a durable, compounding asset that is very hard for a competitor to replicate quickly. Influencer and creator marketing borrows an audience's existing trust in a third party, which can move quickly, but the company does not control the message the way it does in its own paid ads, and a mismatch between the creator's usual audience and the company's actual target segment wastes the spend even when the content itself performs well by its own metrics.
Partnerships and co-marketing share the cost and the audience with another company, which extends reach, but success depends on the partner's cooperation and incentives lining up with your own, which is not always the case even when both sides say it is at the outset. Direct mail and other offline formats are unfashionable and comparatively expensive per contact, but in categories where every competitor has moved entirely to digital, a well-targeted physical piece can cut through simply by being unusual — the same logic that made digital channels attractive when print was crowded now sometimes runs in reverse.
Demand capture means reaching people who already know they have the problem and are actively looking for a solution — paid search on a high-intent term, a comparison page, a well-optimised product page. Demand creation means reaching people who have not yet framed their situation as a problem this product could solve, and making the case that they should — content that reframes a familiar frustration, a category-defining campaign, thought leadership that introduces a new way of thinking about an old task.
A plan made entirely of capture channels looks efficient in the short run — every pound spent has a traceable path to a lead, which makes it easy to defend in a budget review — but it has a structural ceiling: it can only ever reach the pool of people who already know they want this, and that pool does not grow on its own. Over several periods, a capture-only plan tends to plateau or decline as the addressable pool of already-aware buyers is worked through, and the cost of capturing the remaining, harder-to-reach share of that pool rises. Demand creation is what grows the pool that capture later draws from, but it is slower, harder to attribute, and easier to cut in a budget review precisely because its payoff shows up later and less traceably. A plan needs both, in a mix appropriate to the category's buying cycle length and the company's tolerance for spending against a slower, less certain return — and the balance is a judgement call the plan should state and defend explicitly, not a default that happens by not choosing.
A plan without a budget attached to specific line items is a set of intentions, not a plan. This part covers how the number gets built, who spends it, and how the year's activity is sequenced.
A bottom-up budget is built by listing the activities the plan intends to run — the campaigns, the channels, the headcount, the tools — costing each one, and summing the total. Its strength is that every pound in the budget is tied to a specific, named activity someone can be asked about; its weakness is that it tends to grow to whatever size the team can imagine spending, unconstrained by what the business can actually afford or what return is plausible.
A top-down budget starts from a constraint — often a share of projected revenue, or a fixed allocation set by finance — and works backward to what activity that number can fund. Its strength is fiscal discipline; its weakness is that the resulting number has no necessary relationship to what the objectives in part one actually require, and a top-down number handed down without reference to the plan's ambitions can quietly force the objectives to shrink without anyone deciding that on purpose.
Illustrative arithmetic: if a bottom-up build of named activities totals £420,000 and a top-down allocation of a share of projected revenue comes to £310,000, the gap of £110,000 is not a rounding error to smooth over — it is a decision the plan has to make explicit, either by cutting named activities down to the top-down number, by making the case upward for the larger number against a specific objective it funds, or by phasing some of the bottom-up activity into a later period. The two approaches rarely agree on the first pass, and that disagreement is useful precisely because it forces the conversation about which objectives the money can actually buy, rather than letting an unreconciled gap sit quietly in a spreadsheet until it surfaces as an overspend or an unmet target.
Fixed spend continues regardless of results this month — salaries, retainers, software subscriptions, a annual sponsorship already committed to. Variable spend can be turned up or down in response to what is working — most paid media, most freelance production, most one-off campaign costs. The ratio between the two determines how much genuine flexibility the plan has mid-period: a budget that is nearly all fixed cannot respond to a channel that stops working or a competitor that suddenly moves, while a budget that is almost entirely variable offers flexibility but usually means the team is understaffed relative to what it is trying to execute, since headcount is close to the purest form of fixed cost there is.
An in-house team accumulates institutional knowledge about the product, the customer and the brand voice that is expensive to replicate elsewhere, and it is available for the unglamorous, ongoing work — the newsletter that goes out every week regardless of how interesting this week's news is — that agencies are usually not economical for. An agency brings specialist skill the company does not need full-time, a breadth of pattern recognition from working across multiple clients, and the ability to scale a specific effort up quickly for a defined period without a permanent hiring commitment; the trade-offs are less product context, a relationship that has to be actively managed rather than simply directed, and a cost structure that assumes ongoing retained work rather than one-off bursts.
Freelance sits between the two: closer to an agency's flexibility, without an agency's account-management overhead, but without an agency's bench of backup capacity if a particular freelancer is unavailable or turns out to be a poor fit. Most plans of any size end up as a blend, and the plan should say explicitly which activities sit in which category and why, rather than leaving the resourcing model implicit in a set of ad hoc hiring and contracting decisions made separately from the plan.
The calendar is where the plan becomes concrete in time: what runs when, what depends on what, and where the unavoidable seasonal or industry-event anchors sit that the rest of the year has to be planned around. A calendar built only in the heads of the people running each channel produces the two classic failure patterns — two big campaigns launching in the same week and competing for the same audience's attention, and a long dead stretch where nothing is scheduled because everyone assumed someone else had it covered.
A campaign brief, for each individually planned campaign on that calendar, should state: the objective it serves (traced back, as in part one, to a marketing objective and from there to a business objective), the target audience and the specific message it is built on, the channels it will run through, the budget allocated to it, the timeline including any dependencies on other teams, who owns it, and the metric that will be used to judge it afterwards. A brief missing any of these tends to produce a campaign that drifts in scope as it is built, because there is no written reference to say what was actually agreed.
Most organisations of any size need some review step between a piece of marketing being written and it going live — a compliance check in a regulated industry, a legal review of claims that could be challenged, a brand check for consistency, or simply a second set of eyes from a senior stakeholder. The plan should state what needs review, by whom, and with what lead time, because an unplanned review step is one of the most common causes of a campaign missing its launch date. The trade-off to manage deliberately is that too light a review process risks a costly public mistake — an unsubstantiated claim, a legally exposed statement — while too heavy a process slows every piece of marketing down enough that the team stops attempting timely, reactive work altogether. Where the right balance sits depends on the industry and the size of the potential misstep, and a plan should say plainly which failure mode it is more worried about this period and design the review step accordingly.
Always-on activity runs continuously regardless of the calendar — ongoing paid search, the regular content cadence, the standing email programme, community moderation. Campaign activity is time-bound and built around a specific push — a launch, a seasonal moment, a defined promotional window. The two need different planning treatment: always-on work benefits from steady, incremental optimisation over a long period and suffers when it is treated as an afterthought squeezed in around campaign work, while campaign work benefits from a concentrated planning effort and a hard deadline and suffers when it never actually gets that concentrated attention because the team is permanently absorbed in always-on maintenance. The split between the two, stated as a rough share of budget and of the team's time, is worth making explicit in the plan rather than leaving it to whatever happens to be loudest in a given week.
A plan that cannot be measured cannot be defended, adjusted, or learned from. This part is the one most plans treat as an afterthought, and it is the one most responsible for whether a marketing function keeps its budget the following year.
The funnel is a simplification — real buyer journeys loop, stall and skip stages far more than the tidy diagram suggests — but it remains a useful organising device for assigning metrics to a stage. At the top sits awareness and reach: how many people in the target audience have encountered the brand at all. Below that, engagement: whether people who encountered it did anything — read further, watched, returned. Below that, lead generation or demand capture: whether engagement converted into an identifiable prospect. Below that, the handoff to sales and eventually to a closed deal or a purchase, and beneath the whole thing, retention and expansion, which is a marketing concern as much as a product or customer-success one wherever repeat purchase or renewal matters to the business.
Each stage has metrics that belong to it, and one of the most common measurement mistakes is reading a metric from one stage as if it answered a question that belongs to a different stage — treating an awareness metric like impressions as if it were a leading indicator of revenue, when the two are separated by several stages each with their own conversion rate and their own reasons to fail independently of the stage above.
Attribution is the attempt to credit a specific channel or touchpoint with a specific outcome, and every method for doing it involves a real trade-off rather than a simple accuracy ranking. Last-click attribution credits whichever channel a customer interacted with immediately before converting; it is simple and universally available in basic analytics tools, but it systematically overcredits the channels that happen to sit last in a journey — often direct traffic or branded search — and undercredits the channels that did the earlier work of creating awareness in the first place. First-click attribution does the reverse, crediting the first touchpoint and undercrediting whatever closed the deal.
Multi-touch attribution tries to split credit across every touchpoint in a journey according to some weighting rule, which is more honest about the fact that most real purchases involve several touches, but it depends on being able to observe and stitch together all of those touches for the same person across devices and over time, which is increasingly difficult as privacy rules and browser changes restrict cross-site tracking — and the weighting rule itself is usually a modelling choice rather than an observed fact, which means two organisations using "multi-touch attribution" can mean two different specific models.
Incrementality testing takes a different approach entirely: rather than attributing credit within existing activity, it holds a channel back for a defined group — a geographic region, a customer segment — and compares outcomes against a group where the channel ran as normal, to see what the channel actually added on top of what would have happened anyway. It is the most rigorous way to answer whether a channel is causing results rather than merely correlating with people who were going to convert regardless, but it requires a large enough audience to split meaningfully, a willingness to deliberately withhold marketing from part of that audience for the test period, and enough patience to let the test run to a valid read.
Media mix modelling takes a statistical approach at a higher level, using historical spend and outcome data across all channels together to estimate each channel's contribution, without needing to track individual users at all — which makes it more resilient to privacy changes than multi-touch attribution, but it needs a substantial history of spend variation to produce a reliable estimate, and it works at an aggregate level that cannot say much about which specific campaign or creative within a channel drove the result.
The honest position, which practitioners genuinely disagree on, is that no single method is correct and each answers a different question: last-click and first-click are cheap, always-available approximations useful mainly for day-to-day optimisation of a single channel; multi-touch is more complete where the tracking holds up; incrementality testing is the most trustworthy for the specific, narrow question of a channel's true causal effect, at the cost of time and a willingness to run a controlled experiment; and media mix modelling is most useful at the aggregate, budget-allocation level, where individual-user tracking either isn't available or isn't the question being asked. A mature plan uses more than one, matched to the decision it is informing, rather than picking one model and treating its output as ground truth.
A lagging indicator confirms what already happened — revenue closed, deals won, a quarter's total pipeline generated. It is the number that ultimately matters, but by the time it is available, the activity that produced it is finished, which means it arrives too late to change the activity that produced it. A leading indicator moves earlier in the process and gives an earlier signal of whether the lagging indicator is likely to land where it needs to — content engagement rates, lead volume in the early weeks of a campaign, email list growth rate, search rankings on target terms.
The discipline is choosing leading indicators that have actually been shown, in this specific business's own history, to precede the lagging outcome the plan cares about — not indicators that sound plausible in the abstract. A leading indicator chosen without that check can move in a reassuring direction while the lagging outcome it was meant to predict fails to follow, which erodes trust in the whole measurement system once the gap becomes visible.
Different audiences need different reporting frequency and different levels of detail. A channel owner running paid search benefits from a daily or weekly view granular enough to catch a problem quickly. A marketing leader needs a less frequent, more aggregated view that connects channel performance to the objectives from part one. An executive or board audience needs an even less frequent view focused almost entirely on the business objective and a small number of marketing objectives, with channel-level detail available on request but not presented by default.
A dashboard is a live surface of numbers; a decision is a specific choice about what to do differently because of what the numbers show. The two are not the same thing, and a great deal of reporting effort goes into building dashboards that get looked at without ever producing a decision, because nobody has been assigned the job of asking, on a fixed cadence, "given this number, what are we going to change." A reporting cadence that does not end in a named decision-maker looking at the numbers and choosing an action is theatre, however well the dashboard is designed.
The written plan is a starting position, not a fixed script. This part covers how the plan gets adjusted in flight, and the recurring points of friction that come up every period regardless of how well the plan was written.
Most marketing functions plan annually but operate on something closer to a quarterly rhythm, because a year is too long a horizon to hold fixed given how quickly channels, competitors and the market itself move, and a week is too short to allow anything meaningful to compound. A quarterly cycle typically opens with a review of the previous quarter against its objectives, moves to a re-forecast of the current quarter's targets in light of that review, sets or adjusts the coming quarter's campaign calendar, and reallocates budget between channels based on what the measurement in part five has actually shown, rather than what was originally guessed months earlier. Treating the annual plan as the fixed reference and the quarterly cycle as the mechanism for keeping that plan honest, rather than treating the quarterly cycle as an excuse to abandon the annual objectives whenever something more exciting comes up, is the discipline that keeps a plan from drifting into whatever the most recent idea happened to be.
An experiment is a deliberately time-boxed, budget-boxed test of a new channel, message or tactic, run specifically to generate a decision about whether to invest further, not simply to try something new for its own sake. Sizing an experiment means setting, in advance, the budget it gets, the duration it runs for, and — critically — the result that would count as success and the result that would count as failure, all decided before the experiment starts rather than argued about afterward once the actual numbers are in and everyone has a stake in a particular interpretation of them.
Illustrative arithmetic: an experiment budgeted at £8,000 over six weeks, with a stated success threshold of at least 40 qualified leads at a cost of no more than £200 each, gives a clear read at the end regardless of which way it goes — 55 leads at £145 each is an unambiguous pass, 12 leads at £667 each is an unambiguous fail, and the team can move to the next decision without relitigating what the goalposts were supposed to be. An experiment sized without that upfront threshold nearly always produces an ambiguous result that different stakeholders read in opposite directions, because everyone quietly supplies their own bar for success after seeing the number.
A channel deserves to be killed when it has been given a fair, adequately funded and adequately long test against a pre-agreed threshold and has failed to clear it — not simply when it has had one disappointing week, and not simply when it has become unfashionable to discuss in the industry generally. The two mistakes run in opposite directions and are both common: killing a channel too early, before it has had time to compound (this is especially damaging for durable, owned-media channels like SEO or content, which are structurally slow to show results and get cut just before they would have started paying off); and keeping a channel alive too long out of sunk-cost attachment to past success it is no longer producing, or because no one wants to be the one to say the number everyone privately suspects out loud. The discipline that avoids both is the same one that makes experiments useful: set the threshold and the timeline before running the test, and honour the result once it arrives, whichever direction it points.
Some of what marketing builds keeps producing value after the specific activity that built it has finished — search rankings earned by content already published, an email list already collected, a community that keeps talking to itself, a brand that keeps being recognised. These compounding assets are genuinely owned: nobody can switch them off, raise their price unilaterally, or change the rules governing how they perform overnight. Other channels are rented: a paid platform's ad inventory, a social platform's algorithmic reach, an affiliate network's placement — all of which stop the moment the company stops paying for them or the platform changes its terms, and none of which the company controls.
A healthy plan invests in both, but is explicit with itself about which is which, because a portfolio that is entirely rented is fragile to platform changes entirely outside the company's control, while a portfolio that is entirely compounding-asset-focused is slow to respond to an urgent, near-term shortfall. The proportion that makes sense depends on the company's stage and risk tolerance, and it is worth stating in the plan rather than letting it be an accident of which channels happened to be easiest to start with.
Marketing usually owns, or shares ownership of, the company's public voice in a moment of crisis — a product failure, a public complaint that goes wide, a mistake the company itself made. Preparation matters more than improvisation here: knowing in advance who has authority to approve a public statement, what channels a response goes out through, and what the company's general posture is (acknowledge quickly and specifically, versus wait for full facts before saying anything) saves crucial time when a real situation is actually unfolding and everyone involved is under pressure. The plan does not need to script every possible crisis, but it should name who owns the decision and what the first hour of response looks like, because the first hour is usually the one where the most avoidable damage is done.
Almost every organisation that has both a marketing and a sales function eventually has some version of the same argument: marketing believes it is generating strong leads that sales is failing to follow up on properly, and sales believes marketing is generating a high volume of unqualified leads dressed up in a good-looking report. The dispute is not really about a specific lead; it is a disagreement in the underlying definitions — what counts as a marketing-qualified lead (MQL) is not the same as what counts as a sales-qualified lead (SQL), and if the two teams have not written down and agreed the criteria for each stage and for the handoff between them, both sides will keep score using their own private definition and both will be able to produce a report that looks bad for the other.
The concept of an MQL is itself genuinely contested among practitioners — some argue any behavioural scoring threshold produces too many false positives to be worth the argument it generates and that the label should be scrapped in favour of a much narrower, sales-defined qualification bar applied earlier; others argue that without some marketing-owned qualification stage, sales ends up either drowning in raw, unfiltered leads or marketing gets no credit at all for demand that eventually converts much later than a single touch would suggest. Both positions have real merit depending on the sales cycle length and deal size in question — a long, high-value enterprise cycle behaves differently from a short-cycle, low-value transaction — and a plan should state which position it is taking for this business and why, rather than importing a definition wholesale from a different company's playbook.
Whatever position the plan takes, the practical fix that resolves most of the friction, regardless of which side of the MQL debate the organisation lands on, is the same: a written, jointly agreed definition of what qualifies at each stage, a documented service-level agreement for how quickly sales follows up on a qualified lead, and a shared view of the data — usually a shared system of record — so a dispute is a data question with a checkable answer rather than a matter of two people's differing recollections of a conversation that happened weeks ago.
The annual review closes the loop the quarterly rhythm keeps open through the year: a full look back at the business objectives set at the start of the period, a comparison against what was actually delivered, an honest account of which channels earned their budget and which did not, and a fresh situation analysis to carry into the next planning cycle rather than assuming the market held still for a year. Its most important output is not a scorecard but a short, specific list of what the next plan should do differently — which channels to invest further in, which to cut, which segments to add or drop, and what the objectives themselves should be revised to given what was actually learned. A plan that is rewritten from scratch every year without reference to what the last one taught wastes the one advantage a second attempt has over a first.
| Channel | Good at | Cannot do |
|---|---|---|
| SEO | Durable, compounding traffic; capturing existing search intent long-term | Produce fast results; guarantee ranking against a well-resourced competitor |
| Paid search | Immediate, controllable volume against known intent | Keep working once spend stops; stay cheap once competitors bid up the same terms |
| Content marketing | Building trust and authority; feeding SEO; educating an audience that isn't ready to buy yet | Show a clean, fast attribution path to revenue |
| Organic social | Brand presence; distributing existing content; community signalling | Deliver reliable reach at scale on most platforms today |
| Paid social | Targeting by interest and behaviour; visual and video formats; broader-funnel reach than search | Match the intent-specificity of paid search |
| Email and lifecycle | Nurturing an audience the company already owns; retention and repeat purchase | Reach anyone not already on the list |
| PR and communications | Third-party credibility; category and narrative shaping | Be scheduled or guaranteed; controlled in exact message |
| Events and field | Deep engagement; accelerating considered, high-value sales cycles | Scale cheaply per contact |
| Community | Durable peer trust; retention; product feedback | Be built quickly or bought outright |
| Influencers and creators | Borrowed audience trust; fast reach into a defined niche | Guarantee message control or audience-fit every time |
| Partnerships and co-marketing | Shared cost and shared audience reach | Succeed without the partner's genuine cooperation |
| Direct mail and offline | Cutting through in an all-digital category; memorability | Scale cheaply; be measured as precisely as digital |
| Method | Can tell you | Cannot tell you |
|---|---|---|
| Last-click | Which touchpoint immediately preceded conversion | Whether that touchpoint caused the conversion, or just happened to be last |
| First-click | Which touchpoint started the journey | What actually closed the deal, or whether the first touch mattered at all |
| Multi-touch | A weighted view of every observed touchpoint in a journey | Anything about touchpoints it could not observe or stitch together; whether its weighting rule reflects reality |
| Incrementality testing | What a channel truly added versus a held-out control | Fine-grained detail below the level the test was designed to split |
| Media mix modelling | Aggregate contribution of each channel across a long spend history | Which specific campaign or creative within a channel drove the result |
| Metric | What it measures | What a bad reading usually means |
|---|---|---|
| Reach / impressions | How many people were exposed to the message | Weak targeting or an under-funded distribution channel — rarely a message problem on its own |
| Engagement rate | Whether people who saw the message did anything with it | A message or creative that is not resonating with the audience it reached |
| Cost per lead | Spend divided by leads generated in a channel or campaign | Either the targeting is too broad or the offer is too weak to convert efficiently |
| Lead-to-opportunity rate | Share of leads sales accepts as genuinely qualified | A mismatch between marketing's and sales's definition of a qualified lead |
| Win rate | Share of opportunities that close as customers | Poor lead quality upstream, or a sales-process issue outside marketing's control — worth checking which |
| Customer acquisition cost | Total acquisition spend divided by customers acquired in a period | Either spend is inefficient or the funnel is leaking heavily somewhere between lead and close |
| Marketing-sourced pipeline | Value of open opportunities marketing activity is credited with generating | A channel mix skewed toward awareness with too little demand-capture activity underneath it |
| Retention / renewal rate | Share of customers who stay or renew over a period | Often a product or service issue, but can reflect a marketing-set expectation the product did not meet |
| Share of voice | The brand's visibility relative to named competitors in a category | Being outspent or out-published on the channels the audience actually pays attention to |
The cycle repeats every quarter inside the year the plan covers, and the annual review repeats it at the larger scale of the whole planning period.
A plan that has done the thinking in the six parts above still has to exist as a document someone can read, approve and be held to later. The following is the skeleton most working plans converge on, offered as a checklist for what to include, in roughly this order.
Developed by Amit Jain at allfrontierglobal.com
© 2026 All Frontier Global · Panchkula, Haryana, India
Developed by Amit Jain at allfrontierglobal.com · purposed.in · purposed · purposed2 · merchcomp.com · uuka.org
Hand-authored essays — perspectives and figures reflect their writing date; verify current rules with official sources.
A question, a correction, or something you'd like covered. It goes straight to his inbox — no list, no newsletter.