Google Ads Is Recording Conversions. Why Are You Still Losing Money?

You look inside the account and the conversion count is rising. The cost per conversion appears acceptable. Still, the sales team reports low-quality enquiries, or the shop's bank balance paints a very different picture.

Before you alter keywords or bids, identify the reason for the mismatch. If the report is measuring the wrong result, neither change will solve the problem.

First, find out what the conversion actually means

Go into the campaign's conversion goals and review each action that makes up the total. Note what activates every action, where the data originates, and whether it stands for a purchase, a lead, or an earlier step in the journey.

A click on a contact button, a completed form and a paid order each describe a different stage. If they are rolled into one total, judging customer acquisition becomes much harder.

Google separates primary conversion actions from secondary ones. Primary actions may be used for bidding when the campaign relies on their linked goal; secondary actions usually act as observation signals rather than bidding inputs. One key exception applies: if an action is part of a custom goal and that custom goal is assigned, then that action is used for bidding even when it is marked secondary. Do not rely on the label next to the action alone; check the campaign's real goal setup. Google explains these settings.

In a sales-focused campaign, ask whether the bidding goals match the result you are trying to achieve. Earlier-stage interactions may still be valuable for reporting, but making them bidding signals requires a deliberate choice and a clear justification. Google's conversion-goal documentation explains how account-default and campaign-specific goals work.

Test one real journey before trusting the total

Run the action yourself using a controlled test lead or order that can be identified later. Make sure a form is only recorded as successful after submission completes. For purchases, confirm that the correct amount, currency and unique transaction ID are sent. Then return to the confirmation page and check whether the tag fires again.

For purchase events, Google can deduplicate two entries with the same transaction ID inside the same conversion action. That ID has to belong uniquely to that transaction. If a fixed ID is reused, sales may be undercounted. This does not imply that two separate purchase conversion actions will automatically deduplicate against one another. Google's transaction-ID guide covers the implementation.

Also check whether the same commercial result is being counted in more than one place, for example by a Google Ads purchase tag and an imported Analytics purchase event. Decide which action should drive bidding and which, if any, should exist only for comparison.

The counting option is a different setting altogether. “One” records one conversion for each ad interaction tied to that action; it does not mean one distinct person across the whole business. “Every” can count several purchases after a single interaction. Pick the setting that suits the action, and rely on order IDs to stop duplicate purchase entries. The difference is explained in Google's counting options.

A cheap lead can be an expensive customer

Take this hypothetical service business as an example. These numbers are only there to show the arithmetic; they are not AdPilot results or sector benchmarks.

The stated cost per enquiry is £30. The advertising cost per customer works out at £400. Three customers contribute £900 before advertising, which means the campaign ends £300 down after the £1,200 ad spend, and that is before fixed overheads.

Here, contribution means revenue minus the direct and variable expenses involved in delivering the work. The £300 figure is only meaningful if those costs are included.

Treat the enquiries as one group and allow follow-up time. Track which ones were unsuitable, which were lost, and which became customers. A weak close rate may point to poor targeting, an unattractive offer, or weak sales follow-up. On its own, the number does not reveal which one.

Google supports qualified-lead and converted-lead goals for more advanced outcomes. A qualified lead has been assessed beyond the first enquiry; a converted lead has reached a later stage defined by the business. If those outcomes are sent back to Google, keep the stage definitions consistent and confirm the imported records. Do not simply rename every submitted form as qualified.

A 3× return on ad spend can still lose money

Now imagine a hypothetical online shop generating £3,000 in net sales from £1,000 of ad spend. That gives a revenue ROAS of 3×, or 300%.

Assume £2,250 is absorbed by product costs, fulfilment, payment fees and other variable expenses. That leaves £750 before advertising and a £250 deficit after advertising, before any fixed overheads. In this example, revenue already excludes tax and is net of refunds; refunds are not counted a second time in costs.

If the contribution margin before advertising is 25%, revenue ROAS needs to hit 4× just to pay for advertising in this simplified example: 1 ÷ 0.25 = 4. More would be needed for fixed overheads and any target profit. This is an accounting threshold, not a bidding recommendation or a guarantee that Google can reach it.

When comparing the store and the ad account, keep the revenue basis consistent. Review discounts, tax, cancellations and returns. Google offers conversion adjustments so values can be restated or eligible conversions can be retracted after the original event. A refund in the shop does not automatically mean the ad report has changed.

If the business is willing to take an initial loss in order to gain repeat customers, record the expected repeat contribution and the payback period. Build that expectation from observed customer behaviour, not from hoped-for lifetime value. Attributed sales also do not show how many sales were actually caused by the advertising; incrementality is a separate issue.

What to inspect across the six campaign types

The financial check does not change, but the evidence you can examine and the right goal do vary. Use the table below as a starting point, not as a statement that every format should be measured by immediate sales.

Selecting the campaign format is part of configuration, yet it does not resolve the measurement issue. Google's campaign-type guide outlines the distinct roles campaigns can fill. Our campaign-building guide explains the structure beneath the ads.

Give delayed conversions time to arrive

Spending from the recent period may already be finished even though the related conversions are still on the way. Google notes that conversion delay can make newer performance look worse than older results.

Choose a review period that matches your real sales cycle. Compare like with like for leads or orders, and verify the reporting-date basis before reconciling systems. Do not expect exact day-by-day parity between reports governed by different attribution rules.

Waiting for late-arriving sales is not the same as overlooking a confirmed error. A broken form, an incorrect purchase value, or an unsuitable bidding goal should be fixed as soon as it is discovered.

Decide what to change from the evidence

When measurement is incorrect, fix it before relying on those numbers to argue for higher spend. If the leads are real but poor quality, review the searches, targeting, offer, and qualification process. If customers are buying yet contribution is too low, look at product mix, costs, and acquisition economics before pushing for more conversion volume.

AdPilot's Optimize walkthrough demonstrates how account data can surface campaign issues such as wasted search terms and weak ad copy. That may help you respond to a problem. It cannot replace the order records, margins, and sales outcomes required to assess profit. For irrelevant searches, see our negative-keyword guide; for recurring checks, use the Google Ads audit checklist.

Often, the first worthwhile improvement is a better definition of success. Confirm what is being counted, tie it to the customer outcome, and work out what is left after costs. After that, select the campaign adjustment that solves the issue you identified.

Common questions

Does a Google Ads conversion mean a sale?

Only when the conversion action is set up to measure a sale. Check the action and its trigger; the total may also include enquiries or other configured events.

Is a higher ROAS always better for the business?

ROAS is based on the values that are reported. Revenue ROAS does not deduct fulfilment, product, or advertising costs. Compare it with contribution, overall volume, and the business's objectives.

Should I pause a campaign as soon as it appears unprofitable?

First determine whether the outcome is caused by a tracking problem, unfinished sales results, or a real performance gap. Spending that is confirmed to be unaffordable requires quick action; a new report by itself may not yet reflect the final result. Before you continue an experiment, set an affordable test budget and define a clear decision rule.