TL;DR
- A KPI is a metric tied to a goal, a target number, a time window and an owner; a number without those is an ordinary metric.
- SMART (Specific, Measurable, Achievable, Relevant, Time-bound) checks that a KPI is clear, but not that it measures the right thing.
- Pair a volume KPI such as lead count with a quality KPI such as the share of leads sales accepts, so volume cannot rise by lowering quality.
- In GA4, conversions are called key events; check each marked event against the agreed KPI definition before setting targets.
A KPI (Key Performance Indicator) is the headline measure you use to tell whether a business, team or campaign is moving toward an agreed goal. If the goal is "win more new customers", the KPI is the number tied directly to it, such as the number of qualified leads or the cost per lead. A number you can look at but that is not tied to a goal is just an ordinary metric.
What a KPI is, and how it differs from a metric and an OKR
A KPI is a metric chosen because it reflects a core business goal, not every number a system reports. GA4, Google Ads and Search Console report hundreds of metrics, but only a few deserve to be called KPIs. The simplest test is this: if the number got worse for two weeks in a row, would you change a decision? If the answer is no, it is not yet a KPI.
KPI versus metric
A metric is any number you can measure, such as clicks, visitors or bounce rate. A KPI is the small subset of metrics promoted because it has a goal, a target number, a time window and an owner. "Number of clicks" is a metric. "Cost per lead at or below the agreed ceiling by the end of this quarter" is a KPI, because it has a ceiling, a period and a person responsible.
KPI versus OKR
OKR stands for Objectives and Key Results. It is a goal-setting framework, and the Key Result is the number used to decide whether the Objective was met, so a Key Result is one way of using a measure inside an OKR. In practice, a KPI is usually tracked continuously to show the health of ongoing work, such as cost per lead staying under a ceiling, while a Key Result is usually set to move from a starting point to a target within a cycle such as a quarter. The two coexist. A team can use KPIs as the baseline gauges and promote a few of them into Key Results when it wants to push them.
Example: where an ordinary metric ends and a KPI begins
The table below compares numbers common in online marketing and shows which usually serve as supporting metrics and which are better tied directly to a business goal. The grouping is analysis of the principle, not a fixed rule, because the same number can be a KPI for one business and a supporting metric for another.
| Number | Usually treated as | Mechanical reason |
|---|---|---|
| Ad impressions | Supporting metric | Shows reach volume, but not whether anyone was interested or bought |
| Cost per qualified lead | KPI | Ties spend to a result the sales team can actually use |
| Landing page conversion rate | Page-level KPI | Shows how well the page turns visitors into leads |
| ROAS | KPI if real order values are recorded | Compares recorded sales with the ad spend paid |
| Page follower count | Usually a supporting metric | Counts people who followed; does not guarantee they saw posts or bought |
The SMART method for setting a KPI
SMART is a checklist for testing whether a KPI is clear enough to use. Each letter poses one question, and if you cannot answer it, the KPI is still vague.
- Specific: state exactly what is measured, for example "leads from the contact form", not just "enquiries".
- Measurable: there is a data source and a formula that anyone can reopen and get the same number.
- Achievable: the target comes from the account's own history, not from a figure heard about another industry.
- Relevant: the number connects to revenue or to a goal leadership cares about.
- Time-bound: it says which period is measured, such as monthly or by the end of the quarter.
A SMART KPI written out
The numbers below are illustrative only, to show the writing pattern, and are not an industry benchmark: "Reduce the cost per qualified lead from 1,000 baht to 800 baht by the end of this quarter, counting only leads the sales team confirms meet the criteria, measured from GA4 and sales records, owned by the head of the ads team." That sentence answers every letter, and the reader knows at once where to look up the number.
The limit of SMART
SMART makes a KPI clear but does not guarantee the target is right. A KPI can satisfy every SMART letter and still measure the wrong thing, for example "leads per month" defined precisely while most of those leads never become customers. Test every KPI against a business outcome as well.
Common online marketing KPIs by funnel stage
An orderly way to organise KPIs is by funnel stage, because each stage answers a different question and numbers from different stages should not be compared head to head.
Awareness
The question at this stage is whether the people who should see you have seen you. Usable numbers include impressions, reach and impression share for search ads. These describe volume, not quality, so pair them with a number from the next stage. If impressions rise but clicks do not, the ads are being shown to people who are not interested.
Consideration
The question is whether people who saw you were interested enough to come and look. Common numbers are click-through rate (CTR), cost per click (CPC), the engagement rate of sessions in GA4, and the share of visitors who view an important page such as services or pricing. This group shows whether the ad message and the landing page match what people searched for.
Conversion
The question is whether interested people did what the business wants. Most headline KPIs live here: the number of conversions (called key events in GA4), conversion rate, cost per lead or cost per acquisition (CPA), and ROAS for businesses with online sales. The important discipline is defining what counts as a conversion, because counting taps on a call button and counting leads the sales team has confirmed produce very different numbers.
Retention
The question is whether the customers you won come back. Useful numbers are repeat purchase rate, retention rate and customer lifetime value. Many businesses do not track this stage and reward the team on a low cost per lead alone, which can pull in low-quality leads without anyone noticing.
| Funnel stage | Example KPIs | Question answered |
|---|---|---|
| Awareness | Impressions, reach, impression share | Has the target audience seen us? |
| Consideration | CTR, CPC, session engagement rate | Did those who saw us come and look? |
| Conversion | Conversions, conversion rate, cost per lead, ROAS | Did interested people act on the goal? |
| Retention | Repeat purchase rate, retention rate, customer lifetime value | Did the customers we won come back? |
Leading versus lagging indicators
A lagging indicator reports a result after the event, such as monthly revenue or the number of new customers. A leading indicator moves first and helps anticipate the result, such as branded search volume or the conversion rate of a landing page under test. They play different roles. Lagging tells you whether it worked but arrives late. Leading gives an earlier signal but does not guarantee the final result will follow.
An example from advertising
If a business sells a service that takes several weeks to close, revenue from a campaign appears late. In that window the team should watch leading indicators such as the number of qualified leads and the rate at which those leads book a call with sales, then return to lagging indicators like revenue when the sales cycle ends. Watching only revenue means problems are fixed late. Watching only leading indicators means adjusting campaigns to numbers that have not yet turned into money.
How to choose a pair
A workable method is to pick one lagging indicator that matches the business goal, then find one or two leading indicators the team controls directly. After that, check the history: when the leading indicator moved, did the lagging one follow? If they never move together, the leading indicator is only a metric that looks good, not a reliable predictor.
Setting KPIs for an agency or marketing team without fooling yourself
The most common risk in setting KPIs is choosing numbers that look good easily but are unrelated to revenue, known as vanity metrics. Examples are impressions, follower counts and likes. These numbers are not wrong in themselves, but when they are used to judge performance, the team or agency has an incentive to push them up even when that does not help the business.
Principles that guard against self-deception
- Tie the KPI to the end result: if the business wants customers, count qualified leads or real orders, not every form submission.
- Use paired numbers: pair a volume KPI with a quality KPI, such as lead count with the share of leads the sales team accepts, so volume cannot rise by lowering quality.
- Agree definitions before starting: write down what a conversion means, which system measures it and who checks it.
- Set a baseline before a target: look at the account's own history before choosing a target number, otherwise the target is a guess.
- Separate what each party controls: an agency controls ad settings and content but not the sales team's closing, so assign each KPI to the party that owns it.
When a KPI decides pay or a contract
If a KPI is tied to compensation or contract renewal, first check how easily the number can be shifted or flattered. For example, cost per lead can fall by adding brand ads for people already searching for the business name, which pulls in cheap leads without creating new demand. Reporting brand and non-brand campaigns separately gives a truer picture.
Limits worth stating plainly
No KPI set fits every business, and there is no correct target number for everyone. The right target depends on profit per sale, the length of the sales cycle and the seasonality of that business. If someone offers a "standard" number without asking about those, be cautious.
Example: KPIs for a service business (illustrative)
This example is entirely hypothetical, to show the logic. It is not from a real account and not a market average. Suppose a service business sells a high-priced service with a long decision time. The business goal is to increase the number of new customers coming from search ads.
- Primary KPI (lagging): new customers from search ads per quarter, with sales recording the source.
- Leading KPI: qualified leads per month, and cost per qualified lead.
- Quality KPI: the share of leads that sales can book a follow-up call with, to stop volume rising by lowering quality.
- Supporting metrics: CTR, CPC and impression share, used to trace causes when a leading KPI moves.
In this structure, if cost per lead falls but the share of leads that book a call falls with it, the team knows at once that the improvement came from lower quality, not better efficiency. That is why paired KPIs are needed.
Tracking KPIs in GA4 and a dashboard
Tracking KPIs continuously needs two things: correct data collection and a display people can read easily. They are separate. If the data is wrong at the source, a dashboard cannot fix it however attractive it is.
Set up conversions in GA4 to match the KPI
In GA4, the events a business treats as valuable must be marked as key events (formerly called conversions), such as a form submission, a call tap or a purchase. A common problem is minor events being marked, which inflates conversions, or important events not being configured at all. Before setting KPIs, open the key event list and check that each one matches the agreed definition. If you are moving from Universal Analytics or need to reorganise measurement, the Google Analytics 4 migration and setup service covers this.
Combine data from several sources
One KPI often needs data from several systems, such as ad cost from Google Ads, visits from GA4 and lead status from a CRM. If each system sits apart, teams copy numbers across by hand, which invites errors and leaves everyone's reports disagreeing. Consolidating data from several sources in one place lets everyone use the same set of numbers.
Reporting from ads and SEO
For advertising, numbers such as cost per lead and conversion rate should be pulled from both Google Ads and GA4 and compared, because the two systems count conversions under different rules. A small gap is normal, but a large one needs investigating. For a team that manages the account and agrees KPIs with you, see the Google Ads service, and for organic search see the SEO service, which uses KPIs such as clicks from search and leads that originate on web pages.
What this means for Thai marketers
This section is analysis, not a claim that any source studied Thailand. Many Thai businesses receive leads through several channels at once: web forms, call buttons and chat in LINE. If a KPI counts only forms, the number will be lower than reality, and if it counts every button tap without checking quality, it will be higher than reality.
What to check in your own account
- Write a short list of KPIs per team, with an owner and a data source for each.
- Check whether the channels customers really use, such as calls and LINE, are measured as key events yet.
- Pair volume KPIs with quality KPIs, and get lead status data back from the sales team.
- Report brand ads separately from non-brand ads.
- Review KPIs every quarter and remove any that never changed a decision.
If you want specialists to help define KPIs, set up measurement and build reports your team will use, you can talk to the Relevant Audience team through the related services above.
Frequently asked questions about KPIs
What is a KPI in the shortest possible terms?
A KPI is a headline measure tied to the goal of a business or team, with a target number, a time window and an owner. It differs from an ordinary metric, which is simply a number that can be measured.
How many KPIs should a team have?
There is no fixed number, but in principle fewer is clearer for decisions. With too many, the team cannot tell which to weight, and some will never be looked at.
How is a KPI different from an OKR?
A KPI is a gauge of the health of ongoing work that is tracked continuously, while an OKR is a goal-setting framework with an Objective and Key Results, usually set to move within a cycle. They can be used together.
What is a vanity metric and what should I do about it?
A vanity metric is a number that looks good but is not tied to revenue or a real goal, such as follower count or impressions. The fix is not to use it as the main measure of performance and to pair it always with a number that reflects the end result.
What tools do I need to track KPIs?
It depends on where the KPI data comes from. Most businesses start with GA4 for on-site behaviour and Google Ads for ad data, then use a dashboard to bring those numbers together. More important than the tool is agreeing the definition of each KPI first.
