How to Know If Your Ads Are Actually Working (Beyond Clicks and Likes)

Measure the result your business actually needs

Clicks, impressions and likes describe activity. They can help diagnose a campaign, but they do not establish whether advertising is commercially worthwhile.

A campaign can attract enthusiastic engagement without generating sales. Another can produce fewer clicks but bring in customers who fit your business, pay promptly and return. The difference becomes visible only when advertising data connects to real customer outcomes.

Start by defining success in operational terms. For a consultancy, it might be a qualified enquiry that becomes a paying client. For a retailer, it might be a completed order that remains profitable after fulfilment and returns.

Build a measurement chain

Map the journey from advertisement to business outcome:

• Ad interaction or visit.
• Enquiry, booking or purchase.
• Qualification or attendance, where relevant.
• New customer or completed order.
• Revenue, refunds and contribution.

Assign each stage a clear definition and owner. A qualified lead might require the right geography, a relevant need, an achievable budget and a realistic purchase timeline. Your definition should match your business rather than a generic marketing template.

For lead generation, use a CRM or a consistently maintained spreadsheet to record source, campaign, lead date, qualification, sale date and value. For ecommerce, connect order identifiers to transaction events and reconcile them against the order system.

Know what the core metrics mean

CPL: cost per lead

CPL equals advertising spend divided by the number of leads. State whether you mean all enquiries or only qualified leads.

Hypothetical example: spending 2,000 currency units produces 100 enquiries. CPL is 2,000 ÷ 100 = 20. If only 25 enquiries qualify, qualified CPL is 2,000 ÷ 25 = 80.

Both calculations are correct, but they answer different questions. A lower raw CPL is not an improvement if qualification deteriorates enough to increase customer acquisition costs.

CPA: cost per acquisition or action

CPA equals spend divided by the specified conversion count. Because platforms may use “acquisition” or “action” differently, name the outcome every time.

Continuing the hypothetical example, if ten enquiries become customers, media-only customer CPA is 2,000 ÷ 10 = 200. A platform reporting a 20 CPA for form submissions is not contradicting that calculation; it is counting a different event.

For fuller acquisition economics, include applicable agency fees, creative costs and other acquisition expenses. Keep this fully loaded figure separate from media-only CPA so comparisons remain consistent.

ROAS: return on advertising spend

ROAS equals attributed revenue divided by advertising spend.

Hypothetical example: 8,000 in attributed revenue from 2,000 in ad spend gives a ROAS of 4, often written as 4:1 or 4x. That is revenue efficiency, not profit.

If the contribution margin before advertising is 30%, those sales contribute 2,400 before media cost. Subtracting 2,000 leaves 400 before other overheads and acquisition expenses. Under these simplified assumptions, break-even ROAS on media alone is 1 ÷ 0.30, approximately 3.33.

Use a margin definition that includes the relevant variable costs, such as goods, fulfilment and payment fees. Account for refunds consistently, and do not treat speculative lifetime revenue as cash already earned.

MER: marketing efficiency ratio

MER is commonly calculated as total business revenue divided by total marketing spend. Definitions vary, so label your denominator explicitly. If you include only media spend, say so rather than comparing it with someone else's all-marketing MER.

Hypothetical example: 30,000 total revenue divided by 6,000 total marketing expenditure gives a MER of 5. This provides a blended business view, including revenue that individual ad platforms cannot confidently attribute.

MER does not establish causation. Revenue may change because of seasonality, price increases, repeat customers, stock availability or offline activity. Use it alongside channel evidence, not as proof that every campaign works.

Make the tracking trustworthy

Before interpreting a report, test the measurement setup:

• Complete a real test enquiry or permitted test purchase.
• Check that the intended conversion fires once, not on every refresh.
• Confirm transaction values and currencies.
• Exclude obvious internal tests from business reporting.
• Use consistent campaign naming and URL parameters.
• Check cross-domain journeys and booking tools where applicable.
• Deduplicate browser and server events if both are implemented.
• Collect and share data in line with consent requirements and applicable law.

Server-side tracking is not a bypass for privacy obligations, and no implementation captures every customer perfectly. Document known gaps so the team does not confuse missing data with missing demand.

Understand attribution before adding numbers together

Different platforms may claim the same customer under different attribution windows and rules. One might count a conversion after an ad view, while another attributes it to a later click.

Keep three reporting views: platform results for optimisation, analytics for cross-channel journeys, and business records for actual sales and margin. They will not always agree. Investigate material differences rather than forcing artificial consistency.

For longer sales cycles, compare lead cohorts after enough time has passed. This month's ad spend divided by this month's closed sales can be misleading when those customers first enquired months earlier.

Turn the report into decisions

Review spend, qualified leads, new customers, revenue and contribution on a regular schedule. Keep diagnostic metrics such as click-through rate below those business outcomes.

If clicks are relevant but enquiries are weak, inspect the offer and landing page. If enquiries are plentiful but qualification is poor, review targeting, creative promises and form criteria. If qualified leads rarely close, examine pricing, response speed and the sales process.

Increase spend when economics are acceptable and capacity exists, while monitoring whether marginal results worsen. For larger decisions, consider a well-designed holdout or geographic experiment where feasible. Attribution describes credit; controlled testing can provide stronger evidence about additional business generated.