Most early teams try five channels at once: a few blog posts, some LinkedIn ads, a cold email sequence, an Instagram account and a Product Hunt launch. Six weeks later nothing has clearly worked and nothing has clearly failed, because nothing got enough effort to tell. Pick one channel that fits your price point, your buyer and the demand that already exists. Run it for six weeks against rules you wrote down beforehand. Then scale it or kill it.
Five channels at once means five weak signals
Every channel has a learning curve. Paid platforms need a minimum volume of conversions before their algorithms settle; Meta says an ad set usually leaves its learning phase after about 50 results in a week. Outbound needs dozens of conversations before you know which message works. SEO needs months of publishing before rankings show a trend.
A ₹3,00,000 monthly budget split five ways gives each channel ₹60,000, which may not buy enough results on any of them to get past the learning stage. Time splits the same way. One founder across five channels is a day a week each.
The book Traction argues for focus with its Bullseye framework, as summarised in these book notes: brainstorm widely, run cheap tests on a few promising channels, then put everything into the single one that works. The rest of this post is about getting that first choice right.
Start with channel-model fit
Money is the first filter. A channel has a typical cost to acquire a customer, and your business model has a typical amount each customer is worth. If the first is bigger than the second, no amount of optimisation will save it.
Reforge’s essay on the four fits puts it directly: “channels are determined by your model.” In its examples, outbound sales needs high contract values to support it, mid-cost channels suit medium contract values, and virality suits a freemium model with low price points.
A rough map follows. The boundaries are loose and move by market and country.
| What a customer is worth in year-one gross profit | Channels that usually fit | Channels that usually do not |
|---|---|---|
| Under ₹5,000 / £50 | SEO and content, referral and virality, marketplaces and app stores, community | Outbound sales, LinkedIn ads, events |
| ₹5,000 to ₹1,00,000 / £50 to £1,000 | Paid search, paid social, content, partnerships, inside sales for larger accounts | Field sales, heavy account-based marketing |
| Above ₹1,00,000 / £1,000 | Founder-led sales, outbound, LinkedIn ads, account-based marketing, events | Pure self-serve channels as the only route |
No idea yet what a customer is worth? Estimate it. The LTV, CAC and payback lesson shows how to work it out from early data, and the paid acquisition channel fit lesson goes further into matching paid channels to models.
Two more filters: the buyer and the demand
Who buys, and where they already are
A channel works when your buyer is already there with the right intent.
Consumers and solo professionals decide alone and fast, so channels that reach many individuals cheaply fit them: social, search, creators, app stores. A business buying for a team has to convince more than one person, which favours channels that allow a conversation, such as founder-led sales, outbound, webinars and partnerships.
Then look at where they spend attention. A developer is easier to reach on GitHub, in documentation searches and in technical communities than on Instagram. A small restaurant owner in Pune is easier to reach on WhatsApp and Instagram than on LinkedIn.
Is there search demand already?
This decides whether you capture demand that exists or have to create it.
Check in Google’s Keyword Planner how many people search for the problem you solve and for the category name. Google’s help notes that average monthly searches are historical, averaged over 12 months by default, for the locations you choose, and that its forecasts factor in bids, budget and seasonality.
If people already search in decent numbers (“invoice software for freelancers”), search is a strong first channel: paid search for speed, SEO for compounding. Demand spread across thousands of similar long-tail queries can suit programmatic SEO.
If you are naming a new category, search will not carry you. You need channels that create demand: outbound, content pushed through social, creators, community, partnerships.
Score your shortlist
Take three to five candidate channels that pass the money filter and score each from 1 to 5 on the criteria below. Multiply by the weight and add up.
| Criterion | Weight | What a 5 looks like |
|---|---|---|
| Channel-model fit | 3 | Expected cost per customer well under a year of gross profit |
| Buyer presence and intent | 2 | Your buyer is there, in a buying frame of mind |
| Speed to signal | 2 | You will know within six weeks |
| Cost to test | 1 | A real test fits your budget and team |
| Your team’s edge | 1 | Someone on the team already knows how to run it |
| Scalability | 1 | Room to grow 5 times if it works |
A clinic scheduling tool, scored
Take a B2B scheduling tool for clinics priced at ₹2,500 a month, with year-one gross profit per customer of roughly ₹24,000. Clinic owners search for “clinic appointment software” in modest numbers, and one of the founders has sold to clinics before.
| Channel | Fit (×3) | Buyer (×2) | Speed (×2) | Cost (×1) | Edge (×1) | Scale (×1) | Total |
|---|---|---|---|---|---|---|---|
| Founder-led sales to clinics | 4 (12) | 5 (10) | 5 (10) | 4 (4) | 5 (5) | 2 (2) | 43 |
| Paid search | 4 (12) | 4 (8) | 4 (8) | 3 (3) | 2 (2) | 3 (3) | 36 |
| SEO content | 5 (15) | 4 (8) | 1 (2) | 4 (4) | 2 (2) | 4 (4) | 35 |
| LinkedIn ads | 2 (6) | 2 (4) | 3 (6) | 2 (2) | 1 (1) | 3 (3) | 22 |
| 3 (9) | 2 (4) | 3 (6) | 3 (3) | 1 (1) | 4 (4) | 27 |
Founder-led sales comes out on top because the buyer is reachable, the team has an edge and the signal comes fast. Paid search is the obvious second. SEO scores well on fit but is too slow to go first, so run it as a background habit and keep the main test elsewhere.
These scores are judgement. What they buy you is a team arguing about the same six criteria instead of about favourite channels.
Design a six-week test
You need a hypothesis, a budget big enough to learn from and a decision date.
- Week 0, set-up. Write the hypothesis: “Founder-led outreach to 200 clinics in two cities will produce at least 8 paying customers at under ₹15,000 sales cost each.” Check that tracking works end to end, from first touch to payment, with consistent UTMs on every link. Our tracking audit tactic works as a checklist.
- Weeks 1 and 2, launch. Run the basic version without optimising, and collect enough volume to see a baseline.
- Weeks 3 and 4, improve. Change one big thing at a time (the message, the offer, the audience) and note what changed and when.
- Week 5, hold steady. No changes, so the last stretch of data is clean.
- Week 6, decide against the rules you set in week 0.
Size the budget to reach a meaningful volume. If you need around 50 conversions to judge a paid campaign and expect a cost per trial of ₹1,200, budget at least ₹60,000 for the testing period, plus room to iterate. Can’t afford that? Test a cheaper, earlier conversion (sign-up instead of purchase) or pick a different channel.
Kill, iterate or scale: set the rules first
Agree the thresholds before the test starts, so the week-6 decision isn’t a mood.
| Result at week 6 | Decision |
|---|---|
| Cost per customer within target, and improving | Scale: double the budget or effort, and watch whether cost per customer holds |
| Within 1.5 times target, with a clear lever left untried | Iterate: run four more weeks on that one lever |
| Above 1.5 times target and no clear lever left | Kill: stop, write down what you learned, move to the next channel on your list |
| Not enough volume to tell | Fix the test, not the channel: more budget, a cheaper conversion event, or a longer run |
Two warnings. Early results from a founder selling personally will beat what a hire gets, so discount them before you plan a team. And cost per customer usually climbs as spend goes up, because the most responsive buyers go first. Scale in steps and check the cost at each one.
What to ignore while you test
Ignoring channels is the hard part. While the test runs:
- Don’t open new social accounts “just to have a presence”.
- Don’t launch on directories and launch sites without a plan for the traffic.
- Don’t start a second paid platform because a peer said it worked for them.
- Do keep one low-effort habit that compounds, such as one useful article a fortnight, and don’t judge it yet.
When the second channel is due
Go back to the scoring table once the first channel works and you can see where it will stop scaling. The usual signs: each extra step of budget or effort brings customers at a noticeably higher cost than the last; you are already reaching most of the search demand, target accounts or community members that exist; or one platform’s policy change or price rise would cut your new customers by half or more.
Add the second channel while the first is still growing, so you have time to learn before you need it. Re-score the shortlist with what you now know, which is your real cost per customer, your real sales cycle and the skills the team has built.
Before the week-0 meeting
Have these on paper before anyone spends a rupee:
- Year-one gross profit per customer, even a rough figure
- Keyword Planner numbers for your problem and category in your main markets
- A scored shortlist of three to five channels that pass the money filter
- The hypothesis, budget and kill, iterate and scale thresholds for the top channel
- Tracking that works from first touch to payment
- The week-6 decision meeting, already in the calendar
Questions people ask
How many growth channels should an early-stage startup focus on?
One at a time, with at most one small exploratory test alongside it. Each channel needs enough volume and attention to learn from, and splitting a small budget or a small team across several channels usually leaves all of them below the point where results mean anything.
How long should I test a new acquisition channel?
Six weeks is a practical minimum for most paid and outbound channels: one week to set up tracking, four weeks of running and improving, and one week to read results. SEO and content need longer, so judge them on leading indicators such as rankings and impressions at six weeks.
What is channel-model fit?
It is whether a channel's cost to acquire a customer fits what your business model earns from that customer. Expensive channels such as outbound sales need high contract values; cheap, high-volume channels such as virality or SEO suit low price points.
When should I add a second channel?
When the first one is working and you can see where it will stop scaling, for example when cost per customer rises as you add budget, or when you have captured most of the search demand. Start the second channel while the first is still growing, before it plateaus.
