Marketing Budget

7 min

How Much Should a B2B Company Spend to Test a New Paid Channel?

Size the test from the result you need to read, run it in phases, and set the cut rules before the money moves.

How Much Should a B2B Company Spend to Test a New Paid Channel?

Most B2B teams that say a new channel didn’t work ran a test that couldn’t have told them. The budget bought a few weeks and a handful of leads, the cost per lead looked high next to search, and the channel was cut. The question the test was meant to answer, whether this channel reaches buyers we can’t reach now, never got enough data to be answered either way.

We size channel tests the other way round. First we decide what the test has to show and which result proves it. Then we work out how many of those results it takes to trust the read. The budget is whatever buys that many, in a set number of weeks, with the rules for cutting or scaling written down before launch.

How much should a B2B company spend to test a new paid channel?

Enough to buy about 30 of the result you’ll judge the channel on, inside the test window: your target cost per result times 30, for each audience you test. Google Ads recommends judging automated bidding over periods with at least 30 conversions, such as a month or longer (Google Ads Help, 2026).

The arithmetic is simple once the result is chosen. If the channel’s job is to produce qualified leads and you can afford $400 for each one, a readable test needs about 30 of them, so about $12,000 for one audience. Two audiences need two cells. If that number is more than you can spend, test fewer audiences or pick an earlier result, such as qualified visits from target accounts, and say in advance that the test answers a smaller question. What doesn’t work is a test sized by habit, such as a flat few thousand dollars or a fixed share of the budget. It produces a number nobody can act on.

Choosing the result is where most tests go wrong.

What should a new channel test measure?

The result that matches the job you gave the channel. A channel meant to create demand gets judged on awareness lift, branded search or pipeline from the accounts it reached, not on cost per lead. For Slalom, connected TV was judged on awareness, and it drove a 6-point lift on 30% of spend.

Judged on leads alone, a connected TV line can look expensive, and a team reading only the lead report would cut it. Judged on the job it was given, Slalom’s did what it was there for. So write the job down before launch: “this channel is here to reach senior buyers who aren’t searching” leads to an awareness study or a branded search read; “this channel is here to produce sales conversations” leads to qualified leads or meetings. Then pick one primary result and at most two supporting ones. A test with six success measures will always find one that went up.

The result also sets how long the test needs to run.

How long should a B2B paid channel test run?

Long enough to reach your 30 results and to see whether they turn into pipeline, which in B2B usually means one quarter. Read cost per result after the first two to four weeks, and read pipeline from those results at the end. A test shorter than your platform’s learning period tells you about the algorithm, not the channel.

We plan tests in phases inside that quarter rather than as one block. The first phase runs the new channel broadly to get a baseline on formats, audiences and messages. The second cuts what underperformed and moves its money to what worked. The third puts the full test budget behind what the first two phases validated. Each phase ends with a read against rules set before launch, so cutting a format is a planned step, not a judgment made under pressure after a bad week. A phased test also costs less to stop: if the first phase shows the channel can’t reach your buyers at any price you would pay, you stop there.

Phase

The question it answers

What gets cut

1. Broad launch

Can this channel reach our buyers, and with which formats and messages?

Nothing yet; this phase sets the baseline

2. Cut and move

Which combinations produce the result at a cost we can live with?

Formats, audiences and messages below the cut line

3. Scale what worked

Does the result hold at full test budget, and does it turn into pipeline?

The channel itself, if it misses the bar set at launch

How do you keep a channel test readable?

Change one thing at a time where it matters, keep the rest of the program steady, and test the assumption you’re least sure of. In Slalom’s test, narrow account-based audiences raised costs without lifting conversion against broad targeting, so targeting moved broad. That finding came from testing audience width on purpose.

The common way to ruin a test is to change three things at once: a new channel, new creative and a new offer, launched the same week the sales team changes how it follows up. When the result comes in, nobody can say what caused it. Keep the creative and offer you already know work for the first phase, so the channel is the variable. Hold your other spend steady during the test, or at least log every change, so a jump in branded search can be traced to the right cause. And tell sales before launch which leads come from the test and what to do with them, because a test whose leads sit untouched for a week reads as a channel failure. Between 60% and 75% of US buy-side leaders say advanced measurement, incrementality testing included, falls short on rigor, timeliness, trust and efficiency (IAB, 2026). A test designed this way is one your finance team can trust.

When should you stop or scale a channel test?

Stop when the channel misses the cut line you set at launch after it has had its 30 results, or when phase one shows it can’t reach your buyers at a cost you would pay. Scale when it holds the line through phase three and its results turn into pipeline. Decide both lines before the first dollar moves.

Write the lines as numbers: the most you will pay per result, the share of results that must become pipeline, and the date of each read. Then hold to them in both directions. Teams tend to stop good channels early because the first month is the most expensive, and to keep weak ones running because cutting them feels like admitting a mistake. A channel that passes moves from the test line into the plan, with its own budget, sized from the budget structure the rest of the program uses. For how the channels fit together by sales motion, see how to split budget across Google, LinkedIn and Meta. Moving Parade’s Paid Media work runs every new channel this way.

One move: Before your next channel test, write three lines on one page: the result you’ll judge it on, your target cost per result times 30, and the cut line for each phase. If the budget line is more than you can spend, test fewer audiences or an earlier result rather than buying fewer results.

Frequently asked questions

Should a channel test come out of the current budget or new money?

Either works, as long as the test has its own line and its own measures. Taking it from an existing channel is often better, because the comparison is cleaner: you can see what the money did in each place. Taking it from the channel that looks cheapest, usually branded search, is the move to check twice.

Can you test more than one new channel at once?

Yes, if each channel has enough budget to reach its own 30 results and you read them separately. Slalom’s first phase ran six channels at once, out-of-home, trade publishers, Demandbase, LinkedIn, podcasts and connected TV, to get a baseline on all of them before cutting. A budget that can’t fund that should test one channel well.

GO DOWN THE RABBIT HOLE

Chat with this article. Or talk to a Moving Parade strategist.

Pick a question above, or bring your own.

Ready to build pipeline?

Ready to build pipeline?

Tell us where you are. We’ll tell you what we can do.