Service

Analytics and Measurement

Most marketing arguments are two tools counting differently. Measurement work is deciding what the number means before anybody has to defend it.

Book a Strategy Call

All services

Building a reporting dashboard from marketing data

Why do our analytics and ad platform numbers never match?

Because every tool counts its own way, and nearly every setup has a leak somewhere. Google Ads might say 120 conversions last month, Meta says 95, Google Analytics says 70 and your CRM shows 52 actual customers. Part of the gap is normal. Every platform credits itself for any sale it touched, counts on its own timeline and fills holes with modelling. Much of it, though, comes from broken or doubled tracking, a missing consent setup, or form fills counted as sales.

Our job is making those numbers trustworthy enough to bet money on. That means tracking the right actions, once, correctly, on every device. Respecting privacy law. Linking marketing data to what actually happened in the business.

The short answerGood measurement tells you which marketing actually produces customers. It needs a clean tracking setup, clear definitions of what counts as success, privacy-compliant consent, a link between website data and your CRM or sales records. And simple reporting focused on decisions. Without it, every other marketing decision is a guess.

Why platforms disagree

Reason.	What happens.	Normal or a problem?
Self-attribution.	Google Ads and Meta each claim credit for a sale that both touched.	Normal. That’s why totals across platforms exceed real sales.
Date of credit.	Ads platforms credit the date of the click. Analytics credits the date of the conversion.	Normal. Causes monthly differences.
Attribution windows.	Each platform uses different lookback periods and models.	Normal, but worth knowing.
Consent and blockers.	Visitors who decline cookies or use blockers aren’t fully tracked.	Normal. Modelled data fills some gaps.
Duplicate tags.	The same conversion fires twice from two tags or a page reload.	Problem.
Loose definitions.	“Contact page view” counted as a lead.	Problem.
Broken tracking.	Form tracking stopped after a website update.	Problem.
Cross-domain gaps.	Payment gateway or booking tool breaks the session. So sales look like “referral” from the gateway.	Problem.

The aim isn’t to make every number match. The aim is to know why they differ, remove the gaps caused by mistakes, and agree which source answers which question. For revenue, your store’s order system or CRM gets the last word. For comparing channels, one consistent analytics setup beats adding up what each ad platform claims.

One habit beats any tool: for each key number, write down which system is the source of truth. Revenue from the store’s order system. Qualified leads from the CRM. Traffic and behaviour from analytics. When a number is questioned, everyone knows where to look. And the conversation moves from “whose dashboard is right” to “what should we do about it”.

What we fix first

Most measurement problems sit in a few predictable places. We check these before building anything new.

Are the right conversions tracked: purchases, qualified leads, calls, bookings, not page views?
Does each conversion fire exactly once, on every device and browser?
Are values passed for purchases, ideally with tax and shipping handled every time?
Is cross-domain tracking set up for payment gateways and booking tools?
Is internal and developer traffic excluded?
Are UTM parameters used every time on campaigns?
Is consent collected and respected where the law requires it?
Do Google Ads, GA4 and other platforms use the same conversion definitions?
Can website data be matched to CRM or sales records?

The most common single issue is counting the wrong thing as a conversion. A clinic counting every visit to its “Contact” page as a lead will report far more “leads” than it gets calls. An online store counting “begin checkout” as a purchase will think its ads are wildly profitable. Ad platforms then optimise towards those false signals. Fixing definitions first protects every later decision. Our analytics audit and rebuild service starts exactly here, and the UTM parameters and conversion tracking entries cover the basics.

Getting GA4 right

GA4 replaced Universal Analytics, which stopped collecting data in July 2023. GA4 thinks differently. Everything is an event, reports are flexible but less obvious. And a default install misses most of what a business cares about.

Key events

Mark only real business outcomes as key events (GA4’s term for conversions). Keep engagement events for analysis, not as goals.

Custom events

Track form submissions, calls, WhatsApp clicks, bookings and downloads with clear, consistent names.

Ecommerce

Implement the recommended ecommerce events (view item, add to cart, begin checkout, purchase) with item details and values.

Settings

Extend data retention, exclude internal traffic, link Google Ads and Search Console, set the right time zone and currency.

GA4 also samples and hides some data in reports, especially with Google signals switched on and small numbers. For deeper digging, export to BigQuery. It’s free for standard properties up to generous limits. Some businesses prefer privacy-focused alternatives like Matomo. Our comparison GA4 vs Matomo explains the trade-offs, and the analytics implementation entry covers setup.

Privacy law decides what you can track and how. India’s Digital Personal Data Protection Act, 2023 sets rules for processing personal data, including consent and notice needs. Visitors from Europe and the UK bring GDPR into play. Many other countries have their own rules. Browsers also restrict third-party cookies and tracking more each year.

For measurement, this means a few practical things. A clear consent banner where required, with real choices. Tags that respect those choices. Google’s Consent Mode, which adjusts how Google tags behave based on consent and allows modelled conversions for users who decline; Google made Consent Mode v2 a requirement for advertisers using EEA user data from March 2024. Privacy policies that match what’s actually collected. And collecting only what’s needed.

A proper consent setup protects you legally and keeps data cleaner, since Google’s consent mode can model what it can’t observe. Our consent and privacy service handles this. See also consent mode, cookie consent and DPDP Act. (This is general details, not legal advice.)

A simple UTM convention

UTM parameters are tags added to links (utm_source, utm_medium, utm_campaign) so analytics knows where a visit came from. Without them, clicks from emails, WhatsApp broadcasts, QR codes and social posts show up as “direct” or vanish. With inconsistent ones (“Facebook”, “facebook”, “fb”, “FB-Ads”), reports split the same source into several rows.

Parameter.	Rule.	Example.
utm_source.	Where the link lives, lowercase.	whatsapp, newsletter, instagram, qr-flyer.
utm_medium.	Type of channel, from a fixed list.	email, social, paid-social, offline, referral.
utm_campaign.	Campaign name with date.	diwali-sale-2026, admissions-jan-2027.
utm_content.	Optional: which version or placement.	banner-top, story-2.

A shared sheet that builds tagged links from agreed rules keeps the whole team consistent. Never tag your own internal links. It wipes the original source and wrecks your reports.

Events worth tracking on most sites

A few extra events help spot problems. Track form starts to see drop-offs. Track taps on phone and WhatsApp buttons. Track downloads like brochures. And site search terms. These shouldn’t be treated as goals, but they explain why goals rise or fall. A form with many starts and few submissions has a usability problem. A brochure downloaded often but rarely followed by an enquiry may need a clearer next step.

Server-side tagging

Traditionally, tracking tags run in the visitor’s browser: every analytics tool, ad pixel and chat script loads its own code on the page. That slows pages, exposes data to many third parties and is more and more blocked by browsers and ad blockers. Server-side tagging moves much of that work to a server you control. The browser sends one stream of data to your server, which then forwards what’s needed (and only what’s allowed) to analytics and ad platforms.

Benefits

Fewer scripts in the browser, so faster pages.
More control over what data each vendor receives.
More resilient measurement as browsers tighten rules.
Easier to apply consent choices every time.

Considerations

Hosting costs for the server container.
Technical setup and ongoing maintenance.
Still must respect consent. It isn’t a way around privacy law.
Worth it mainly for businesses with meaningful ad spend or traffic.

Server-side tagging isn’t a magic fix for missing data, and it should never be used to track people who’ve declined consent. Used properly, it gives cleaner data and faster pages. Our server-side tagging service sets it up and maintains it.

Attribution, honestly

Attribution is the question of which marketing deserves credit for a sale. A buyer might see an Instagram ad and Google the brand a week later. Then they read a blog post, get an email, and finally buy through a Google ad. Which one “caused” the sale? The honest answer is that they all played a part, and no model captures it perfectly.

Model.	How it assigns credit.	Tends to favour.
Last click.	All credit to the final click before conversion.	Brand search, retargeting, email.
First click.	All credit to the first touch.	Awareness channels like social and display.
Data-driven.	Credit spread based on statistical patterns in your data.	A more balanced view, if there’s enough data.
Platform self-reporting.	Each ad platform claims any sale it touched.	Whichever platform you’re looking at.

GA4 uses data-driven attribution by default for many reports. It’s more balanced than last click, but it’s still limited to what GA4 can see. Channels like word of mouth, offline ads, podcasts and AI assistant recommendations often leave no trace.

That’s why the strongest measurement setups combine methods. Attribution reports for day-to-day decisions. Incrementality tests (switching a channel off in one region or for a period and measuring the effect on total sales) for big budget questions. And simple questions at checkout or on calls (“How did you hear about us?”) to capture what tracking misses. Our attribution modelling service and the comparison last click vs data-driven go deeper.

Measuring what leaves no trace

Some of the most important marketing leaves no clean digital trail. A podcast mention, a hoarding on a busy road, a recommendation from a friend, a WhatsApp forward of your guide, an answer from ChatGPT that names you. Analytics will never capture these fully. That doesn’t mean they can’t be measured at all.

Ask “How did you hear about us?” on forms, at checkout and on calls, with useful options.
Watch branded search volume in Search Console and Google Trends after campaigns.
Use unique offer codes or short links for offline campaigns.
Compare regions with and without a campaign over the same period.
Track direct traffic to deep pages, which often signals word of mouth or AI referrals.

Combining these softer signals with tracked data gives a far more honest picture than tracked data alone, which tends to over-credit the last digital touch.

Ecommerce details that skew numbers

Online stores have a few measurement traps of their own. Refunds and cancellations often aren’t sent back to analytics. So revenue looks higher than it really is. Cash on delivery orders that are refused at the door still count as purchases. Discounts and coupons may or may not be reflected in reported values. Repeat customers buying through branded search or email can make acquisition campaigns look better than they are. Handling refunds, recording COD outcomes and separating new from returning customers turns ecommerce reports from flattering into useful.

Connecting website data to real sales

For many businesses, the website isn’t where the sale happens. A coaching institute gets an enquiry online, but enrolment happens after a counselling call. A real estate developer gets a form fill. But the sale happens after site visits weeks later. A B2B firm gets a demo request, and the deal closes three months on. If measurement stops at the form, you only know which marketing produces enquiries, not which produces customers.

To link them, save a click ID or campaign source with each lead in the CRM. Update the lead’s status as it moves: qualified, visited, won or lost. And sending those outcomes back to analytics and ad platforms. Google Ads offline conversion import and Meta’s conversions API both support this. Once connected, reports can show cost per enrolment or cost per closed deal by channel. And ad platforms can optimise towards real customers instead of cheap form fills.

ExampleA real estate developer in Gurugram gets 900 enquiries a month from Google and Meta. Meta delivers leads at less than half the cost per lead. After connecting the CRM, site-visit and booking data shows Google leads book site visits at three times the rate and convert to sales far more often. The cost per actual booking is lower on Google. Budget shifts accordingly, and Meta campaigns are rebuilt around higher-quality audiences. (Figures in this example are illustrative.)

See our glossary entries on offline conversion import, CRM and first-party data.

Measuring AI search traffic

Visitors more and more arrive from AI tools: ChatGPT, Perplexity, Gemini, Copilot and others. Some of this traffic is easy to see. Some isn’t. Visits that come with a referrer from chatgpt.com, perplexity.ai, gemini.google.com or copilot.microsoft.com can be grouped into their own channel in GA4. Some AI apps strip the referrer, so visits appear as direct. Clicks from Google’s AI Overviews are counted as normal Google organic traffic. So they can’t be separated in standard reports.

Create a custom channel group for known AI referrers.
Watch landing pages that AI tools tend to send visitors to.
Track conversions from AI referrals separately.
Monitor unexplained growth in direct traffic to deep pages.
Ask leads how they found you, with “AI assistant” as an option.
Check server logs for AI crawler activity on key pages.

AI traffic is usually small in volume but often well informed. Because visitors arrive after reading a summary. Measuring it separately helps judge whether GEO work is paying off. Our AI traffic measurement service sets this up.

Lead quality scoring

Not every lead is equal, and treating them as if they were distorts decisions. A simple scoring approach helps. Sales or front-desk staff mark each lead with a quick status: junk or spam, not a fit, interested, qualified, won. Over a few months, patterns appear. Some campaigns bring mostly junk. Some keywords bring mostly qualified prospects. Some landing pages attract people who never answer the phone. Feeding the “qualified” and “won” statuses back to ad platforms lets their bidding favour the better sources. Even without automation, a monthly look at lead quality by source often changes where budget goes. See lead scoring.

Meta and the conversions API

Meta (Facebook and Instagram) relies heavily on its pixel, which browser restrictions and ad blockers more and more limit. Meta’s Conversions API sends sales data from your server straight to Meta. It works with the browser pixel and recovers some lost signals. Set up properly, with event deduplication so the same purchase isn’t counted twice from both sources, it usually gives Meta a more accurate picture. As with all tracking, it must respect consent and avoid sending sensitive personal data. Our paid social team works with measurement on this.

Reporting people actually read

Most marketing reports run too long, cram in too many metrics, and say too little. A 40-page PDF of sessions, bounce rates and impressions rarely changes a single decision. Good reports start from the questions the owner actually asks.

For the owner or leadership

Are we getting more customers and revenue from marketing? What did it cost? What’s changing, and why?

For marketing

Which channels, campaigns and pages produce qualified leads and sales? Where should budget move?

For sales

Which leads are best, where do they come from, and how quickly are they followed up?

For the website team

Which pages lose people? What broke? What improved after changes?

A useful monthly report fits on a page or two: a few key numbers against target, what changed, why. And what we’ll do next. A Looker Studio dashboard can give live access to anyone who likes digging. Our dashboards and reporting service builds these around your actual decisions.

Metrics that mislead

Some numbers look important and rarely are on their own. Pageviews without context. Bounce rate (GA4 now reports engagement rate instead, which is more useful). Impressions without clicks. Follower counts. “Conversions” that aren’t real outcomes. These can still help diagnose problems, but they shouldn’t headline a report.

Measuring SEO properly

SEO results often get reported as a list of rankings or a sessions chart. On their own, neither tells you much. Good SEO tracking starts with Search Console: impressions and clicks for the topics that matter. Then look at organic landing pages, what visitors do there, and the sales they bring. Compare with the same month last year, not last month, and most seasonal noise disappears. Split brand searches from the rest and you’ll see whether SEO is finding new customers or just catching people who already know you.

Unit economics: the numbers that decide budgets

Marketing choices get much easier once you know a few business numbers. Customer acquisition cost (CAC): total marketing spend for a channel divided by new customers it brought. Customer lifetime value (LTV): the profit a typical customer brings over their relationship with you. Contribution margin: what’s left from each sale after product and delivery costs, before fixed overheads. If a clinic’s patients bring ₹12,000 of profit over two years and Google Ads acquires them for ₹2,500, there’s room to spend more. If an online store’s average order leaves ₹300 of margin and acquisition costs ₹450, every new customer loses money unless they come back. Measurement that connects these numbers turns marketing from guesswork into arithmetic. (Figures here are illustrative.)

Helping the team use the data

The best tracking in the world is wasted if nobody looks at it, or nobody trusts it. A 30-minute session showing each team where their numbers live, what they mean and what to do about them changes a lot. So does agreeing one place for each number: revenue from the store system, leads from the CRM, traffic from GA4. Once everyone quotes the same source, meetings stop being fights about whose dashboard is right.

Reporting cadence

Different questions need different rhythms. Daily: spend and any sudden breakage, ideally via automatic alerts rather than someone staring at dashboards. Weekly: campaign performance and lead quality for teams running ads and sales. Monthly: channel results, costs and revenue against targets, with a short written explanation. Quarterly: bigger questions such as budget allocation, attribution checks and whether the measurement plan still matches the business. Match each report to the decision it serves and teams stop drowning in numbers they never use.

Three worked examples

These are composite scenarios based on common patterns, not specific client stories. The businesses are invented.

Example 1: an online store whose “profitable” ads weren’t

The situation. A kitchenware store’s Google Ads account reports a healthy return on ad spend. Yet the business isn’t growing, and the owner suspects something’s off.

What we’d find. The purchase tag fires on the order confirmation page, which reloads if customers refresh or return to it, counting some orders two or three times. The payment gateway breaks the session. So many orders in GA4 appear as referrals from the gateway rather than from ads. Tax and shipping are included in conversion values for ads but not in GA4.

The fix. The purchase event now fires once per order ID. Cross-domain tracking and referral exclusions sort out the payment gateway. Values are standardised. Google Ads conversions switch to the GA4 purchase event plus enhanced conversions.

What changes. Reported ROAS drops to a realistic level, some campaigns turn out to be losing money. And budget shifts to the ones that actually work. Profit goes up, simply because decisions now rest on real numbers. See ecommerce and retail.

Example 2: a B2B firm that couldn’t connect marketing to revenue

The situation. A software company spends on LinkedIn, Google Ads and content. Marketing reports leads. Sales reports revenue. Nobody can connect the two.

What we’d do. Forms capture UTM source, campaign and click IDs into hidden fields, passed into the CRM with each lead. Sales stages are standardised. A dashboard joins CRM outcomes with marketing data, showing pipeline and closed revenue by original source and campaign. Won deals flow back to Google Ads and LinkedIn as offline conversions.

What changes. Turns out a modest content programme and branded search bring in most closed revenue, not the expensive campaigns. Budget is reallocated with confidence. See SaaS and technology and measuring SEO for a long sales cycle.

Example 3: a clinic chain with no idea which branch got which call

The situation. A chain of dental clinics runs ads and SEO for five branches. All calls go to one central number, and nobody knows which pages or campaigns drive bookings at which branch.

What we’d do. Each branch page gets its own tracked number and booking link. Call tracking records the source of each call. The front desk marks whether calls became booked appointments, and that data feeds a weekly branch-level report.

What changes. Two branches were getting far too little marketing, and one was getting far too much. They rebalance budget and SEO work, and bookings rise without spending more. See healthcare and clinics.

Patterns behind these examples

All three examples share something. The businesses weren’t short of data. They had dashboards full of numbers. What they lacked was trust that the numbers were real, and a link between marketing and actual revenue. Every time, the fix began with definitions and testing. Not new tools. That’s true of most measurement problems: get the basics accurate, connect them to real outcomes. And the right decisions usually become obvious.

Signs your measurement needs attention

Ad platforms report far more conversions than you have customers.
Nobody can say which channel brought last month’s best clients.
A large share of revenue shows up as “direct” or “referral” from your payment gateway.
Leads in reports don’t match leads in the CRM.
Conversions dropped to zero for a few days and nobody noticed.
Different team members quote different numbers for the same thing.

Any two of these usually mean the setup needs a proper audit before the next budget decision. Because money spent on bad numbers is hard to get back.

How a measurement project runs

Week 1
Audit
We go through GA4, Tag Manager, ad platform conversions, consent, forms and ecommerce tracking, and check what’s actually firing.
Week 1 to 2
Measurement plan
A written plan: your goals, the actions that represent them, how each is tracked, and which system owns which number.
Week 2 to 4
Implementation
Tags, events, consent, cross-domain tracking, server-side setup where useful, and CRM connections built and tested on every key device.
Week 4 to 5
Validation
We place test orders and leads, compare them with real orders and CRM records, and fix whatever doesn’t match.
Week 5 onward
Reporting
Dashboards and monthly reports built around the decisions you make, plus training so your team can use them.

At handover you get the measurement plan, a list of every tag, naming rules and access to every account in your name. A future analyst can pick it up easily.

What to fix first

Do first: high impact, low effort. Remove duplicate conversion tags. Redefine conversions as real outcomes. Exclude internal traffic. Fix payment gateway referrals.

Plan next: high impact, high effort. CRM connection and offline conversions. Server-side tagging. Consent Mode setup. Incrementality testing.

Batch together: low impact, low effort. Consistent UTM naming. Link Search Console and Google Ads. Extend data retention.

Usually skip: low impact, high effort. Tracking every scroll and hover. Building dashboards nobody asked for. Perfecting attribution models before basic tracking works.

← Lower effortHigher effort →

Count conversions wrong and every report and every automated bid is wrong too. Fix that first. Then hook up the CRM. Attribution comes last. That order saves the most money soonest.

What a useful dashboard looks like

Here’s a simple dashboard we’d build for a multi-branch physiotherapy clinic, to show what “useful” means in practice. The top row shows four monthly numbers: new patient bookings, cost per booking, revenue from new patients, and bookings from Google search. Each sits next to last year’s figure. Row two splits bookings by branch and by source: Google Ads, organic, Meta, referrals. Row three lists the landing pages behind most bookings. So the team knows which services and pages are pulling their weight. A small panel lists anything broken or unusual that month, such as a form error or a sudden drop in calls at one branch. That’s it. Anyone in the clinic can read it in two minutes, and every number links back to one agreed source.

Testing and experiments need clean data too

Every A/B test, landing page experiment or budget test depends on measurement. If conversions are double-counted or missing on some devices, test results become unreliable. And businesses end up rolling out “winners” that were never real. Before any testing programme, the conversions it relies on should be validated: firing once, on every device, matching real outcomes. Our CRO team always starts with a measurement check for this reason.

Budget experiments follow the same logic. Want to know if a channel truly drives sales? Pause it in one region for a few weeks. Then compare total sales with a similar region where it kept running. These geo or holdout tests are one of the most honest ways to measure incrementality. But only if baseline tracking is solid and the test is planned in advance, with enough time and volume to see a real difference.

A free measurement checklist

Definitions

Business goals written down.
Each goal mapped to a trackable action.
Only real outcomes marked as key events.
One trusted source named for revenue.
Consistent naming for events and campaigns.

Implementation

Every key conversion fires exactly once.
Tested on mobile and desktop, several browsers.
Purchase values and currency correct.
Cross-domain tracking for gateways and booking tools.
Internal and test traffic excluded.
Tag Manager tidy, with unused tags removed.
Google Ads, Search Console and GA4 linked.

Privacy

Consent banner where required, with real choices.
Tags respect consent choices.
Consent Mode configured for Google tags.
Privacy policy matches what’s collected.
Personal data minimised and secured.

Outcomes and reporting

Leads stored with source and campaign in the CRM.
Lead status updated through the sales process.
Offline conversions sent back to ad platforms.
AI referral traffic grouped separately.
Monthly report answers business questions.
Data checked against real sales every month.

Measurement myths worth dropping

Myth. All the numbers should match.

Reality. Different tools count differently. The goal is to understand gaps and remove the ones caused by errors.

Myth. More tracking is better.

Reality. Tracking everything creates noise and privacy risk. Track what supports decisions, and review the rest every quarter.

Myth. The ad platform’s numbers are the truth.

Reality. Every platform credits itself generously. Check against your own sales data.

Myth. Last click shows what works.

Reality. It over-credits the final step and hides channels that create demand.

Myth. Server-side tracking avoids consent rules.

Reality. Privacy law still applies. Server-side setups must respect consent.

Myth. Set it up once and forget it.

Reality. Website changes break tracking regularly. It needs monitoring.

What goes into a measurement plan

A measurement plan is a short document marketing, developers and sales all agree on. It stops the slow drift where every new tool or campaign invents its own definitions. A useful one covers:

Business goals, in plain words (more booked appointments, more online revenue).
The key results that show progress towards each goal.
The exact actions tracked for each, with event names and parameters.
Where each action is tracked and which system is the source of truth.
How consent affects each type of tracking.
UTM and naming conventions.
Which reports and dashboards exist, and who uses them.
Who owns changes, and how they’re tested and documented.

It doesn’t need to be long. Four or five pages is usually enough. It’s only useful if agreed upfront and updated whenever the business or site changes.

Who owns measurement

Measurement often falls between teams. Marketing thinks developers handle tags. Developers think marketing does. Sales keeps its own spreadsheet. Name one owner who approves tracking changes and the plan stays intact. In a small business that might be the founder or marketing lead with outside help. In a larger one, a dedicated analyst. What counts is that one person clearly owns whether the numbers are right.

Tracking calls and WhatsApp

For many Indian businesses, the most valuable conversions don’t happen in a form. They happen on the phone or on WhatsApp. A clinic might get three times as many calls as form fills. A home services business might get most enquiries on WhatsApp. If those aren’t measured, marketing reports undercount the channels that drive them.

Call tracking

Dynamic numbers on the website show which source, campaign or page led to each call. Calls above a set length can be counted as leads, and outcomes (booked, not booked) recorded.

Click-to-call

Tracking taps on the phone number shows interest, not whether the call went through. Useful, but not the same as a real call.

WhatsApp clicks

Tracking taps on WhatsApp buttons, ideally with a pre-filled message that identifies the page or campaign. So conversations can be traced.

Outcomes

When staff mark which calls and chats became customers. And that flows back to reports and ad platforms, the picture is finally complete.

One caution: call tracking numbers must be used carefully on local listings. Because inconsistent phone numbers across Google Business Profile and directories can confuse local search. The usual approach is to keep the main number as the primary listing and use tracking numbers on the website and in campaigns. Our local SEO team coordinates this.

Keeping tracking healthy

Tracking setups decay. A developer changes a form and the event stops firing. A new marketing tool adds another pixel. A redesign changes the thank-you page URL. Someone duplicates a tag “just to test” and forgets it. Within a year, a clean setup can become a mess.

Keep a tag inventory: what each tag does, who added it, whether it’s still needed.
Use Google Tag Manager with clear naming and folders.
Limit who can publish changes, and document each publish.
Test key conversions after every website release.
Set alerts for sudden drops or spikes in conversions.
Review and remove unused tags every quarter.

A simple alert, such as “tell me if leads today are 50% below the usual level”, catches broken forms within a day instead of at month-end. It’s one of the most valuable pieces of measurement work a business can have. And it takes less than an hour to set up.

A monthly data quality routine

Even a perfect setup drifts. A quick monthly check keeps the numbers honest, no full audit needed.

Compare with reality
Compare GA4 purchases or leads with your order system or CRM for the month. If they’re off by more than a little, dig in.
Look for spikes and drops
Scan key events day by day. A sudden zero or a sudden doubling usually means something broke or duplicated.
Check new campaigns
Check that new ads, emails and posts followed the agreed UTM rules.
Review referrals
Payment gateways, your own domain or spam sites appearing as referrers point to configuration problems.
Test the main forms
Submit each key form and check the conversion fires once and the lead reaches the CRM.

Thirty minutes a month is usually enough. And it saves far more time than fixing a quarter of bad data after the fact.

After a website redesign

Site launches break tracking more often than anything else. Thank-you page URLs change, form plugins get swapped, the data layer vanishes, and nobody tells the analytics person. Before any launch, the measurement plan should be checked against the new site on staging. And every key conversion tested again on the live site within hours of launch. Comparing the first week’s numbers with the previous weeks catches anything missed.

Search Console: the free data most businesses ignore

Google Search Console shows which searches your site appears for, how often it’s shown, how often people click and which pages they land on. It also flags indexing problems, mobile issues and manual actions. For many businesses it’s the most useful free source of marketing insight they have, and it’s often never opened. Linking it to GA4, checking it monthly for new queries and falling pages. And exporting data before the retention window passes are simple habits that pay off. Bing Webmaster Tools offers similar data for Bing, which also feeds parts of Microsoft’s AI products.

Measurement by industry

Industry.	What matters most.	Common gap.
Ecommerce.	Purchases, revenue, margin, new vs returning customers.	Duplicate purchases, gateway referrals.
Healthcare.	Booked appointments by branch and treatment.	Calls not tracked. Privacy of patient data.
Education.	Enquiries, counselling calls, enrolments.	No link between enquiry and enrolment.
Real estate.	Site visits, bookings by project.	Lead quality invisible to ad platforms.
B2B and SaaS.	Qualified pipeline, closed revenue, trial activation.	CRM disconnected from marketing data.
Local services.	Calls, WhatsApp chats, booked jobs.	Calls and chats not counted at all.
Publishers.	Engaged readers, subscriptions, ad revenue per visit.	Pageview obsession.

Clinics, insurers and lenders need extra care with personal data. Measurement should never send sensitive details (health conditions, financial details) to analytics or ad platforms. See our industry pages for sector specifics.

When to go beyond GA4

For many small and medium businesses, a well-configured GA4 plus a CRM and a simple dashboard is enough. As you grow, a few signs say it’s time for more. You need to combine data from ads, CRM, store and calls. GA4’s screens can’t answer your questions. Or sampling hides detail.

The usual next move is sending GA4 data to BigQuery and joining it with CRM, ad and store data. This gives raw, unsampled data and full control. You’ll also need someone who knows SQL and data modelling. So only do it when the questions justify it. Start with a list of decisions the extra data will help with, or you’ll build a warehouse nobody visits.

Cost and timelines

Measurement work is usually a one-off project, then light ongoing support. Cost depends on the number of platforms, the complexity of the site and checkout, consent needs, CRM integration and how much reporting is needed. Our pricing page explains how we scope.

Audit1 week

Measurement plan1 to 2 weeks

Implementation2 to 4 weeks

Validation1 to 2 weeks

Reporting live5 to 8 weeks

Quick fixes like duplicate tags and wrong conversion definitions can be done in days and right away improve ad platform optimisation. CRM links and server-side setups take longer. The real reward: every marketing decision after that rests on numbers you can trust.

Measuring email, social and offline channels

Each channel has its own blind spots. Email tools report opens, but Apple’s privacy features now inflate or hide them. So clicks and sales matter more. Social platforms are generous with reach and likes. What matters is visits, sign-ups and sales. Events, print and hoardings need offer codes, their own landing pages or QR codes with UTM tags to be measured at all. Treat each platform’s own dashboard as a starting point, and judge channels in your own analytics and sales data.

Measurement for small businesses

A small clinic, shop or consultancy doesn’t need server-side tagging, BigQuery or attribution models to make good decisions. It needs a few things done properly. GA4 installed correctly, with form submissions, calls and WhatsApp clicks tracked as key events. Google Search Console set up and checked monthly. Google Business Profile insights reviewed for calls and direction requests. A simple way of recording where each new customer came from, even if it’s a column in a spreadsheet filled in by the front desk. And a monthly half-hour to look at what worked.

That costs almost nothing and answers most small-business questions: where customers come from, and what each one costs. Only add complexity when a specific question can’t be answered without it.

Tools we commonly use

Need.	Tools.	Note.
Web analytics.	Google Analytics 4, Matomo.	Choose based on privacy needs and integrations.
Tag management.	Google Tag Manager (web and server).	Keeps tags organised and testable.
Consent.	Certified consent management platforms.	Must integrate with Consent Mode.
Search data.	Google Search Console, Bing Webmaster Tools.	Free and essential.
Call tracking.	Dedicated call tracking services.	Keep main listing numbers consistent.
Data warehouse.	BigQuery.	For larger, multi-source needs.
Dashboards.	Looker Studio and similar.	Build around decisions, not metrics.
Behaviour.	Microsoft Clarity, Hotjar.	Mask personal data.

Tools matter less than how they’re set up and used. A well-configured free stack beats an expensive one nobody trusts.

From reporting to forecasting

Once the data is reliable, it can do more than describe last month. With a year or two of clean numbers, businesses can estimate how many enquiries a given budget is likely to produce, which months need more spend. And what a realistic target looks like for the next quarter. A coaching institute can plan counsellor staffing around expected enquiry peaks. An online store can plan stock and ad budgets for festival season based on last year’s patterns plus current trends.

Forecasts are never exact, and they should always come with a range rather than a single number. But even a rough forecast based on trustworthy data beats planning on gut feeling. And it makes conversations about budgets far more grounded. The precondition is the same as everything else on this page: tracking that measures the right things, accurately, every time over time.

Common questions about analytics and measurement

Why does Google Ads show more conversions than GA4?

Google Ads credits conversions to the date of the ad click, uses its own attribution and includes modelled data, while GA4 counts differently and shares credit across channels. Some difference is normal. A large gap usually means duplicate tags, different conversion definitions or tracking problems.

Is GA4 good enough for my business?

For most small and medium businesses, yes, if it’s configured properly with the right events, conversions and links to other tools. Larger businesses with complex data needs often add BigQuery and dashboards.

What is Consent Mode?

A Google feature that adjusts how Google tags behave based on whether visitors consent to cookies. It allows some modelled measurement when consent is declined. Google requires Consent Mode v2 for advertisers using EEA user data.

Do I need server-side tagging?

Not always. It helps businesses with big ad spend or traffic that want faster pages, more control over data sharing and more resilient measurement. Small sites often don’t need it yet.

What attribution model should I use?

Data-driven attribution in GA4 is a sensible default if you have enough data. Combine it with incrementality tests for big budget decisions and simple “how did you hear about us” questions to capture what tracking misses.

How do I track phone calls from my website?

Call tracking services show dynamic numbers on your site and record which source led to each call. Combine that with outcomes recorded by staff to see which calls became customers.

Can I track traffic from ChatGPT?

Partly. Visits with a chatgpt.com referrer can be grouped in GA4. Some AI traffic arrives without a referrer and appears as direct. So the true number is higher than what you see.

How often should tracking be checked?

Test key conversions after every website release, review data against real sales monthly and audit tags quarterly. Alerts for sudden drops catch problems in between.

Is it legal to use heatmaps and session recordings?

Generally yes, if set up with privacy in mind: masking personal data, getting consent where required and following laws such as India’s DPDP Act and GDPR.

What’s the most important thing to track?

The action that most directly represents revenue for your business, such as purchases, qualified leads or booked appointments, tracked accurately and connected to real outcomes.

Why does a lot of my traffic show as “direct”?

Direct includes people typing your address, bookmarks, some apps that strip referrers, untagged email or WhatsApp links, and tracking gaps. Consistent UTM tagging reduces unexplained direct traffic, especially for email, WhatsApp and QR code campaigns.

Who should own our analytics accounts?

The business, always, with admin access for at least two people. Agencies and freelancers should have user access you can remove. So a change of partner never means losing years of data.

We’re a small business. Do we really need all this?

No. Start small: GA4 tracking your real enquiries or sales, Search Console linked, UTM tags on emails and WhatsApp broadcasts, call tracking if phones matter. And a simple monthly sheet comparing leads with actual customers. That covers most of what a small business needs to make good decisions. Add more only when a real question can’t be answered.

How do we know our tracking is broken?

Watch for these signs. Conversions suddenly double or fall to zero. Your payment gateway appears as a top referrer. GA4 sales are far below your order system. Any of these is worth a check. A quick test order or test enquiry usually shows where the leak is.

Who should own analytics in our company?

One named person who approves tracking changes and checks the monthly numbers, even if outside help does the technical work. Without an owner, tags pile up and definitions drift.

What is the first number we should trust?

Your real sales or real leads from your own system. Start there. Then check how close GA4 and your ad tools get to it. Close is good. Far off means something is broken.

How long does a tracking fix take?

Most fixes take one to three weeks. Big CRM links take longer. We test each step before we call it done.

Inside Analytics and Measurement

The specific pieces of work this service is made of, each scoped separately.

Analytics audit and rebuild

Fix the tracking before arguing about the numbers it produces.

Analytics audit and rebuild

Consent and privacy-safe measurement

Collect what you are entitled to collect, and be honest about the rest.

Consent and privacy-safe measurement

Server-side tagging

Recover the measurement that browser restrictions take away, where the loss is worth the cost.

Server-side tagging

Attribution and incrementality

Work out what actually caused the sale, rather than which touchpoint was last.

Attribution and incrementality

Dashboards and reporting

Build the report that changes a decision, and delete the one that does not.

Dashboards and reporting

AI and agent traffic measurement

See the part of your traffic that arrives with no referrer and no obvious cause.

AI and agent traffic measurement