Choose channels around buying intent, margins, budget and the team available to run them. Fund what has a clear job, cap experiments and park activity you cannot execute or measure. Platform-attributed sales are not separate piles of revenue you can add together.
You do not need a presence on every channel. You need enough suitable ways for customers to discover the business, evaluate it and act—and the capacity to maintain them.
Before choosing, write down the customer, offer, buying cycle, available creative, operating capacity and acquisition economics. A channel with abundant reach may still be a poor fit for your current constraints.
Compare the seven jobs
| Channel | A useful job | Reason to delay |
|---|---|---|
| Organic search | Answer existing questions and buying intent | Important pages or business facts are incomplete |
| Paid search | Test demand expressed in queries | Conversion tracking or landing pages fail |
| Shopping | Present an eligible product catalog | Product data, availability or margins are unreliable |
| Paid social | Test offers and creative with an audience | No usable creative or workable test budget |
| Organic social | Maintain a useful public conversation | Nobody can produce or respond consistently |
| Help a permission-based audience return | Consent, list quality or the offer is unresolved | |
| Partnerships and referrals | Reach relevant buyers through real relationships | The proposed audience or terms do not fit |
These are decision prompts, not universal channel rankings. Content supports several of these channels; publishing articles is not a distribution plan by itself.
Work out the commercial constraint
Use actual contribution and customer-acquisition costs where available. If the business cannot reconcile orders and spend, repair the measurement before making precise claims about which channel is profitable.
The order economics guide provides a worked model. For service businesses, agree on what counts as a qualified inquiry and track its progression. A cheap form submission is not necessarily a useful lead.

Worked choice: a small reporting consultancy
Imagine a consultancy with a long sales cycle, limited creative capacity and several clear customer questions. It could begin with useful service pages, supporting articles, existing professional relationships and a permission-based follow-up process.
A paid-search test might follow once the landing page and inquiry tracking work. Launching daily video campaigns at the same time could exceed production capacity. This is a hypothetical prioritization, not a claim that those channels always outperform video.
For a product store with strong visual material and repeat purchases, the sequence could differ. The offer and resources should change the decision.
Give a test a decision rule
Specify the audience, offer, spend or time limit, conversion definition, review period and stop conditions. Include the delay between initial contact and purchase. Write down what you will do if the evidence is inconclusive.
Use Google's conversion tracking overview and the GA4 attribution documentation to understand the reports you compare. Platform attribution allocates credit under a model; it is not proof that the channel created every credited sale.
Review the whole business response
Compare channel delivery with valid inquiries, orders, contribution and team workload. Avoid adding overlapping platform conversions as unique customers. Retain an unresolved category when a source cannot identify the journey.
Download the campaign test brief. Keep a channel when its role and evidence justify the effort; repair or pause it when they do not. “We have always posted there” is not a measurement method.
When the decision depends on reconciling reports, business intelligence can help establish the measures and source records first. A channel brief needs an outcome the business can recognise.
Sources
- conversion tracking overviewsupport.google.com
- GA4 attribution documentationsupport.google.com