Home/Learn/Performance marketing

Find the paid channels that pay back

Performance marketing uses paid ads to bring in leads and customers, then judges the spend against those returns. Before choosing search, LinkedIn or remarketing, compare your company's economics with estimated click costs. Identify the constraint that could prevent the channel from paying for itself, and design the first test around it.

Updated November 2026

The campaign proposal arrives before anyone has worked out what a customer can cost to acquire. You have a product and customers, but no marketing lead. Between an agency pitch and ad-platform credits, a small test sounds reasonable. You need a way to decide what that test should answer.

Software searches can consume a test budget with relatively few clicks. Before buying them, work out what would have to happen after a visitor arrives. The estimate gives you a starting decision; the campaign tests it. That means examining the economics, the page and the lead count together.

Work out what a click is worth before buying

Compare the money a customer brings in with the cost of the clicks needed to win that customer. This gives the campaign a condition to meet before it receives a budget.

Start with what a customer pays and how often a lead becomes a customer. Then examine the searches that could bring those leads to the site. A click has value only through the chance that it produces a customer; its affordable cost depends on that whole path.

A click, a lead and a customer are three different events. The estimate needs to connect all three; assuming that every click becomes a lead would make the campaign look more affordable than the business can justify.

The lead-to-customer rate describes part of the path. The landing page still has to turn a visitor into a lead. Where that rate is unknown, the assessment needs to show the assumption being tested. A promising estimate is a reason to investigate a channel, with a limit on what the test can spend.

Search intent can change the arithmetic considerably. In keyword research for one software startup, estimated software-search clicks cost $60-114. Estimated clicks for audit preparation were about $10. These figures were estimates from keyword research.

The audit-preparation group raises a different question from the software group. Its lower estimated click cost helps only if those searchers could buy from the company. They would need a page about preparing an audit, with an offer and action that follow from that problem.

A useful assessment therefore ends with a verdict and the constraint behind it. It also gives a scorecard and a rough campaign shape: which searches belong together, what page they need and what would make the test worth continuing. The founder can then discuss whether this search group, page and spending limit make a worthwhile first test.

  • Calculate the customer value. Use what customers pay and how leads become customers to establish what acquisition can cost.
  • Separate searches by intent. Compare estimated click costs for each buyer problem before selecting the campaign's first group.
  • Identify the limiting assumption. Put the assumption that could overturn the verdict beside the proposed campaign, where it can guide the test.

Clicks teach little until the landing page asks for one action

A visitor needs a clear action after reading the ad's promise. Without it, spending can produce visits while leaving the campaign's central question unanswered.

The action should follow from the search. If the ad concerns audit preparation, the page needs to explain the help available for preparing an audit. A general homepage may require the visitor to find that connection themselves. The test then mixes interest in the problem with the visitor's ability to find the relevant offer on the site.

Separate searches by intent and give each group its own ad and landing page. The page can answer the particular question that brought the visitor there. Its next action gives the campaign a response to count, so the report can connect the search group with the leads it produced.

This also makes an unsuccessful test easier to examine. If an ad attracts clicks but the page produces no leads, read the page against the search. Does it answer the promised question? Can a visitor understand the offer and find the action without working through unrelated material?

A small budget cannot correct an unclear page. It can buy too few visits to distinguish weak demand from a poor explanation. Fixing the page first removes one avoidable source of uncertainty before the test begins.

The same preparation applies to the action's tracking. The campaign needs to record the step it asks visitors to take. Otherwise, the founder may see spend and clicks without knowing whether the people who arrived made an inquiry. The test should have a clear question, a relevant page and a countable answer.

  • Match the ad's promise. Keep the search, ad and page focused on the same buyer problem so visitors understand the connection.
  • Choose a clear action. Give interested visitors a specific next step that the campaign can count as a response.
  • Test the whole path. Follow the page through its action before buying visits, so a broken step does not consume the budget.

Price or search demand usually decides whether paid search can pay

For a product sold at a low monthly price, expensive software-search clicks can use up the money available to win a customer. Changing an ad headline cannot remove that price limit. The first decision is whether another search group could fit the economics.

The product's price sets a limit on the money available to acquire a customer. If the clicks needed to reach a sale cost more than that customer can return, the proposed search group needs to change. A cheaper intent group may offer a different route, provided the buyers still need the product.

Sometimes there is room in the economics but little demand for the proposed phrase. A recruiting-research firm's keyword plan found 210 monthly searches for its category term. The language its buyers used accounted for about 34,000 monthly searches. The figures measured the monthly search demand available to investigate.

The category term could have produced a very different test from the broader set of buyer questions. With little search activity, a campaign may have few opportunities to learn. Increasing the budget does not create people searching for that term.

Naming the constraint changes the next decision. A price problem calls for another intent group or channel. A demand problem calls for different buyer language, a wait, or a way to reach people where they already spend time. Each response follows from what is preventing the proposed campaign from working.

Keep that constraint visible when reviewing the campaign. A lower click cost is useful only if the resulting leads can become customers. More impressions are useful only if they come from searches the product can answer. The measure should stay connected to the reason the channel was selected.

  • Check the price limit. Compare the proposed acquisition cost with what the product can return before approving expensive search terms.
  • Read the demand behind words. Inspect the language buyers use so a narrow category label does not define the entire test.
  • Change the relevant assumption. Choose another intent or channel when the constraint makes the current campaign difficult to justify.

Check that your tracking counts leads before judging any ad

Before reacting to a weak campaign report, check that it records the leads the page receives. A missing inquiry changes the apparent performance. The person reviewing the budget needs to know whether the low count reflects the campaign or a fault in how the site measures it.

In October 2026, a review of a UK company's analytics uncovered a cookie-banner problem: 70-85% of visitors were invisible in the reports. The review found the issue in the first week of work. Before interpreting campaign performance, the reviewer had a concrete counting problem to investigate.

Other early site reviews found different failures. Analytics sent data to a dead project. An app had analytics while its marketing website had none. Another site had no analytics at all. Each failure could leave a founder trying to evaluate marketing from an incomplete record.

The practical check is to follow an actual lead through the page and into the report. If the action happens but the count does not appear where expected, investigate that gap before reading the campaign as a failure. A dashboard's existence does not establish that it counts the website's inquiries.

Counting also needs attention after launch. A campaign report is useful only while the page and its tracking still describe the same action. When the numbers change, the reviewer needs to consider both buyer behavior and the way that behavior was recorded.

Weekly paid-ad management uses tracked leads to tune budgets. Reliable counts make those decisions possible: the reviewer can inspect which searches brought inquiries and whether the spend still fits the original economics. When measurement fails, repairing it becomes the next job before another budget judgment.

  • Follow a real inquiry. Compare the action on the website with its recorded event to find gaps in the lead count.
  • Inspect missing visits. Check consent behavior and the analytics destination when the report stops reflecting activity on the site.
  • Review the count with spend. Read tracking problems alongside campaign performance so a reporting fault does not drive the budget decision.

Start with paid search only when buyers search for the fix

Google Search is a sensible first test when a buyer already knows the problem and searches for help. The company then needs an offer whose economics can carry those clicks.

This is why channel selection belongs after the initial assessment. Search can reach someone expressing a need, but that advantage depends on the words they use and the page they find. The proposed campaign should connect that need with an action the business can serve.

For a low-priced product, the affordable acquisition cost may rule out the obvious software terms. For a new category, the intended buyers may rarely search for the category at all. Either situation gives a reason to consider another route before paying to test the same limitation.

Search pages can answer related questions, LinkedIn can build familiarity, and outbound can start a conversation with selected buyers. These routes ask different questions of the market. Choosing among them should follow how buyers seek help and what the company can economically turn into a sale.

Google is often the first paid-search test in this work. Bing, LinkedIn ads and remarketing are other paid options, each requiring its own assessment. A promising result on one channel does not settle the economics of the next, because the audience and the route to a lead can differ.

Whichever paid channel is selected, the work continues after setup. Someone has to keep the page relevant, read the lead count and decide where the next budget goes. Buying campaign execution without that ongoing judgment leaves the founder with the same unanswered spending question, only later in the month.

  • Confirm an existing search. Find buyer questions that a relevant page can answer before making search the first paid channel.
  • Consider another route. Use the product's price and the available demand to decide whether another channel deserves attention first.
  • Include the weekly decisions. Assign the page, tracking and budget review together so the campaign can change as leads arrive.

Example: checking where the page's action sits before buying clicks

An August 2026 review of a compliance startup's site compared the position of its only form with how far visitors scrolled.

FieldRecorded finding
PageThe startup's site, with one contact form
Form position90% of page depth
Average scroll depth58.6%
ReadingThe average visitor stopped well above the form

Paid visits sent to that page would meet the same problem: the action sat below where a typical visitor stopped reading. Moving the ask, or giving a campaign its own page, comes before spending on clicks. The review records the finding; no campaign result is reported.

Launch the first test only after four checks

  • Record the customer economics

    Establish what customers pay and how often a lead becomes a customer, so the acquisition estimate starts with the business.

  • Compare estimates by intent

    Pull estimated click costs for separate search groups and select a buyer problem the company's numbers could support.

  • Prepare the page and count

    Give that intent its own landing page and next action, then check that actual leads appear in tracking.

  • Set the spending limits

    Agree the cap and stop rule before launching, and use the weekly lead review to decide whether spending continues.

Test whether performance marketing can pay for you

Talk to Tin about your customer economics, which searches could lead to profitable growth channels, and the first campaign worth testing. We can discuss the pages, tracking and spending decisions that test would need.

Book a call