Why channel choice goes wrong
Three forces push businesses toward the wrong channels. Novelty: a platform is in the news, so it must be worth trying. Imitation: competitors are there, so we should be. Habit: we have always done this. None of them asks the only question that matters — what job would this channel do that nothing else in the plan does better? Channel sprawl is the result: a presence on six platforms, none of them run well, with no way to tell which produced customers.
The method below replaces those forces with four checks: demand, attention, job and economics.
Step one: locate existing demand
Demand that already exists is the cheapest to serve, so the first question is how much there is and where.
- Search demand. How many people search for the category, the problem and the brand each month, in your market? Keyword research tools, Google's own planner and Search Console (if a site exists) answer this. If the volume is meaningful, organic search (SEOImproving visibility in unpaid search results.) and paid search are candidates; the SEO traffic estimator converts rankings into expected visits.
- Marketplace and review demand. Do people look for the category on marketplaces, app stores, comparison or review sites? Presence there is a channel in its own right.
- Direct and referral demand. Existing customers, word of mouth, partners. Email and referral programmes serve this demand and are usually the cheapest channel per sale.
If search demand is small, the channel plan has to create demand rather than capture it, which changes the candidates entirely. The inbound vs outbound comparison explains the implications.
Step two: locate attention
Where does the audience spend time when they are not looking for you? This is where demand is created, and it is answered by audience research rather than keyword tools: which platforms, which formats (short video, long video, text, audio), which creators and publications, at which times. For a consumer audience the answer is usually one or two social or video platforms; for a professional audience it may be LinkedIn, industry publications, events or search for adjacent problems. The audience research guide covers how to find out rather than assume; the touchpoints guide maps where attention sits by journey stage.
Step three: assign each candidate a job
A channel without a job cannot be judged. Write the job in one sentence using the funnel stages:
- Create demand: reach people who do not yet know they need this — paid social, video, display, creators.
- Capture demand: be the answer when people look — organic and paid search, Shopping, marketplaces.
- Convert: turn interest into action — landing pages, retargeting, brand search, messaging.
- Retain: bring customers back — email, SMS, messaging, loyalty, community.
Then check coverage. A plan with two channels capturing demand and none creating it will exhaust the pool; a plan with no retention channel buys every sale twice. The channel mix guide covers how to balance the jobs; this step only checks that each one has an owner. For a given objective, the objectives library lists which channels are credible routes and which are not a fit.
Step four: check the economics
For each candidate, estimate the cost per result it would need to deliver and whether that is plausible.
- The exposure price of the channel (CPMCost per 1,000 impressions. or CPCAverage cost of each click.) and the response rates typical of the format give an expected cost per acquisition (CPACost of each conversion.). The CPA calculator does the arithmetic.
- The target CPA comes from the business: margin per sale or value per lead. If the expected CPA exceeds it, the channel is not viable at current conversion rates, regardless of how large the audience is.
- The minimum budget to learn anything matters too. A channel that needs a large monthly spend before its algorithms stabilise is a poor fit for a small test budget; the budget calculator turns a target into the spend required.
Channels paid with time rather than money (organic search, organic social, content) are checked the same way, with hours as the cost. When the budget can fund only one paid channel, the guide to marketing on a small budget sets out how to concentrate it and which of these time-paid channels to run alongside it.
Step five: score and shortlist
A simple scoring table makes the trade-offs visible and the decision defensible. Score each candidate from one to five on each criterion and weight the criteria by what the business needs most.
| Criterion | Question |
|---|---|
| Demand or attention | Is the audience demonstrably there, in the relevant moment? |
| Job fit | Does it fill a funnel stage the plan lacks? |
| Economics | Is the expected cost per result below the target? |
| Speed | How soon will it show results? |
| Capability | Can the team run it well, or would it be run badly? |
| Durability | Does it build an asset (list, rankings, following) or only rent reach? |
The shortlist is the top two or three, not the top ten. The limiting resource is rarely budget; it is the attention needed to run each channel well.
A regional tutoring service
Search demand for tutoring in the region is clear and seasonal, peaking before exam periods; parents also ask for recommendations in local community groups. The shortlist: paid and organic search to capture the seasonal demand, a presence in the community groups where recommendations happen, and email to re-enrol families each term. A short-video platform scores well on attention among students but poorly on job fit — students are not the buyers — and is left off the first plan, to be revisited if the business starts selling directly to older students.
Step six: test before committing
Run each shortlisted channel as a bounded test: a fixed budget or time, a single clear success metric agreed in advance, and a date on which the decision is made. The metric should be the channel's job — reach among the target audience for a demand-creation test, cost per acquisition for a capture test — not a generic engagement number. Tests fail for reasons other than the channel (weak creative, wrong landing page, tracking gaps), so a failed test deserves one diagnosis before the channel is dropped.
Step seven: prune and revisit
Channels that pass become part of the plan with a budget and a job; channels that fail are stopped, not left running at a token level "to keep a presence". A standing budget for these tests can be managed with the 70-20-10 budget rule, which sets written conditions for a channel to move from experiment to promising to proven, or to stop. Revisit the set when something changes: a new product, a new market, a platform's audience shifting, search demand growing to the point where capture needs more investment. The annual planning guide builds this review into the cycle.
Channel selection checklist
Before a channel enters the plan, every line should be a "yes" or a documented exception.
Revisiting the decision
The shortlist is a snapshot. Four events should reopen it. Search demand grows past what the current capture channels can absorb, visible in rising impression share at stable cost — add capacity before competitors do. A platform's audience shifts, visible in falling reach at constant spend or in research showing the segment has moved. A new product changes the jobs (a subscription adds retention to a business that only needed acquisition). And a test in the reserve produces a marginal return better than an incumbent channel's. Outside those triggers, the set stays fixed and the effort goes into running each channel well, which is where most of the return comes from. A channel run well for a year usually beats three channels run adequately for a quarter each.
When the whole programme produces too few enquiries, marketing not generating leads checks channel choice alongside positioning, forms, uncounted enquiries and timing before any channel is cut.
When a shortlisted channel has reached its review date without results, diagnosing a new channel that is not working separates a test that could not answer from a channel that genuinely failed.
When the channel under review is no longer a trial but an established part of the paid plan, the guide to stopping a paid advertising channel covers the holdout an exit should rest on, what leans on the channel and how to wind it down.
Common mistakes
- Choosing by audience size. A large audience with no intent and no fit produces impressions, not customers.
- Starting with the most complex channel. Programmatic and advanced automation reward scale and expertise; they are rarely the right first channel.
- Running a channel without owning the destination. Any channel sending people to a weak page or collecting no data is renting results.
- Keeping channels alive by habit. A presence that nobody manages damages more than it helps.
- Confusing the channel with the job. "Social" can create demand, convert or retain depending on how it is used; the job must be written down.
The channel overview summarises each channel on the three axes this method uses, and the digital channels expertise page shows how the author works across the main ones.
