Marketing is everywhere, but the driving force behind good campaigns is analytics. It transforms clicks, views, purchases and drop-offs into decisions you can stand behind. If you want to understand how brands stop guessing and start adapting on the fly, marketing analytics is the place to look. It sits at the intersection of business, data, psychology and technology, which makes it practical and hard to ignore.
The skills required in this field
To work in marketing analytics, you need more than just an interest in numbers. You translate data into business decisions, so technical expertise and communication skills are equally important. Useful skills include:
- Data analysis and interpretation
- Familiarity with spreadsheets and dashboards
- Understanding attribution, segmentation and A/B testing
- Basic statistics
- Clear business writing and presentations
- Working with tools such as SQL, Tableau or Excel
You don’t need to be a machine learning wizard from day one. Most employers want to see if you can identify useful patterns and explain what action to take next.
For this reason, some professionals look into online master’s programmes in marketing analytics when they want structured training in digital strategy, campaign measurement and data-driven decision-making. It can be a practical move if you’re targeting roles that require both a marketing mindset and analytical depth.
What is marketing analytics?
The process involves collecting, measuring and interpreting data to improve marketing performance. It sounds straightforward on paper, but in practice it means finding out what content works, which audience converts and where your budget is being spent.
You look beyond vanity metrics, such as likes or impressions, and ask tougher questions. Did the advert lead to sales? Did newsletter subscribers stick around? Did a segment of your audience leave faster than a tennis ball bounces on concrete? The aim is to replace assumptions with patterns you can test, and the usual sources of data are:
- Website traffic, from tools such as Google Analytics
- Social media engagement and conversion reports
- CRM data from platforms such as HubSpot
- Paid advertising dashboards from Google, Meta and LinkedIn
- Customer surveys and retention data
- Visits and calls from directory and map listings
Why brands care about analytics
Marketing budgets are rarely approved without questions. Teams are asked to prove their impact, often quickly, and analytics provides a way to do so that relies on more than just instinct. In a campaign for trainers, for example, analytics show whether short videos performed better than banners, whether mobile users converted more often than desktop users, and whether the back-to-school promotion started too early or too late.
It matters because small adjustments can yield big gains. A better headline on the landing page, a smarter audience filter or a smoother checkout process can boost the conversion rate without incurring additional costs. Brands also use analytics to:
- Forecast trends and seasonal demand
- Tailor messages to customer behaviour
- Reduce wastage in the advertising budget
- Quickly identify underperforming channels
- Link marketing activity to revenue
Without analysis, marketing becomes expensive and strangely overconfident.
How analytics changes campaign decisions
Analysis changes what marketers do next, beyond the report in which it appears. Let’s say an online clothes shop notices high traffic from Instagram but poor conversions. At first glance, it looks like a social media issue. A closer look might reveal that users are landing on a slow mobile page, or that the promoted product is selling out before most buyers can get to it.
A streaming service might test two email subject lines and find that one has 18 per cent more opens but fewer completed sign-ups. Attention alone isn’t enough, as the quality of the message and the audience’s intent also matter. Good data sharpens your judgement in decisions such as:
- Which channels receive the largest share of the budget
- Which audience segments to target
- When to start or stop a campaign
- Which content formats are worth expanding
- How to improve the customer journey after the click
What the eBay experiment revealed about brand-term advertising
The difference between a click and a completed sale was measured directly in a series of field tests carried out at eBay. Thomas Blake, Chris Nosko and Steven Tadelis published the results in 2015 in the journal Econometrica. The company paused paid adverts for searches containing the word ‘eBay’ and monitored the outcome. The traffic lost from adverts was almost entirely recouped through organic search results, so branded adverts had not delivered any measurable short-term benefit.
For non-branded search terms, the result was more nuanced: the adverts influenced new users and those who bought infrequently, whilst regular customers – who would have bought anyway – accounted for the lion’s share of the budget, and the average return was negative. Standard reports had suggested the opposite, as people who click on an advert are often those who were already planning to buy.
For anyone studying marketing analysis, the experiment distinguishes between two things that a dashboard lumps together: the customers a channel has brought in and those it has merely intercepted on their way to making a purchase. The question of whether the sale would have taken place anyway has only one certain answer: the test in which part of the audience does not see the advert.
Attribution models in analytics platforms work differently. They divide the credit for a sale amongst the channels the customer passed through, either according to a rule or a statistical calculation, and assume that every touchpoint counted. None of them tell us what would have happened if one of those touchpoints had been missing, and this is precisely the information on which the budget depends.
A small firm does not have eBay’s traffic, but it can carry out the same test on a smaller scale. It stops running its own-brand advert in one region for two weeks, leaves it running in another, and compares the enquiries and orders. The result will be less precise than that of an experiment involving millions of users, yet it will still tell us more than the report from the platform selling the advert.
The study’s conclusion has a limitation which the authors themselves point out. eBay was already one of the best-known brands on the internet, and the effect of the adverts is small precisely because it is a large, well-known brand. A company that nobody has heard of faces the opposite situation, because for it almost every customer is new, so it needs to find out through which channels people who do not yet know its name can find it.
The tools behind the figures
Marketing analysis runs across a range of platforms, and each tool answers a different kind of question. Some deal with traffic, others with customer behaviour, attribution or visibility.
The key is to choose tools that suit your purpose, without collecting every possible metric. If you want to understand why users are leaving the checkout page, heatmaps and session recordings are more helpful than a general traffic report. You also need to keep an eye on data quality issues:
- Faulty tracking pixels
- Conversions counted twice
- Inconsistent naming
- Missing UTM parameters
- Poor integration with the CRM
Messy data leads to polished nonsense that nobody needs.
Company details published outside the website
On top of the quality issues, there is a problem relating to company data. The name, address, telephone number and opening hours appear in dozens of places, from maps to directories, and every different version breaks an attribution link. A call made to an old number, left in a forgotten listing, doesn’t appear in any report, even though the customer actually existed.
For example, the company changed its telephone number two years ago; the website was updated, yet an old listing still directs calls to the deactivated number. The marketing report shows a drop in calls, and no one attributes this to the listing, as it is missing from all analytics platforms.
An inventory solves most of the problem. Carried out once a quarter, it covers all the places where the company appears, from its profile on maps to the last forgotten directory listing, including the link, telephone number and date of the last check for each one. The inventory may look more modest than a dashboard, but it determines whether the figures on the dashboard accurately reflect the company as customers experience it.
Corrections are made from the source outwards. First come the company’s own website and its map profile, then the major directories, followed by the rest, ensuring the same business name and telephone number appear everywhere. For listings it cannot amend itself, the company requests a correction in writing and records the date of the request in the inventory.
Companies that carry out this sort of work are listed in the ‘online marketing’ category on Jasmine Directory, where an editor reviews each submission before publication. The category shows a small business what kinds of suppliers are available – from surveying to paid advertising – before it knows exactly which service it needs.
Where it goes wrong
Many mistakes in marketing analysis stem from tracking the wrong metrics or treating correlation as proof. If an influencer campaign performed well over a bank holiday weekend, the influencer may have contributed only part of the spike.
Another common problem is over-reporting and under-thinking. Teams can spend hours building fancy dashboards that nobody uses to make better decisions. If your report has 47 charts and no next steps, it’s just window dressing.
Attribution also requires careful consideration. Customers rarely see a single advert and make a purchase straight away. They might discover a brand on TikTok, read reviews later, click on a retargeting advert and eventually convert via email. If all the credit goes to the last click, the bigger picture gets lost. To keep your feet on the ground, ask:
- What decision does this data help me make?
- Which metrics are linked to real business objectives?
- What other explanation could there be?
- Is the sample size large enough to be meaningful?
A little scepticism is healthy. Data only has feelings when people project them onto it.
How do you measure the impact of a listing in a directory?
A customer who discovers the business in one place and makes a purchase in another undermines any report based on the last click. Discovery channels are the most disadvantaged: social media, recommendations and listings in directories or on maps appear at the start of the journey and disappear from the report by the end of it. A local business also receives phone calls and in-person visits, which leave no trace in the analytics platform.
Measurement can be carried out using modest means: the link in each listing is assigned its own UTM parameters, for example utm_source with the directory name and utm_medium=directory, so that visits appear separately in the traffic report. A dedicated telephone number for listings shows how many calls come from there. Asking every new client where they first heard about the firm covers what the other two methods miss. Parameters are only applied to links the firm controls, as some directories do not allow the destination address to be changed, in which case the dedicated telephone number remains.
Analysing the figures requires patience. A customer who saw the listing in March might ring in May, and a monthly report would classify them under a different channel or none at all. A three-month observation period, compared with the same quarter of the previous year, provides a more accurate picture than a weekly report.
According to the logic of the eBay experiment, an advert under your own name captures people who were already looking for you, whilst a listing under a category reaches people who were looking for a particular type of service and didn’t know you existed. The Jasmine Directory blog features a comparison between directory listings and paid adverts in 2026, which is useful for anyone splitting a small budget between the two.
A directory listing usually generates few visits, and with such low volumes, it is impossible to distinguish between actual effect and chance. UTM parameters show that someone came via the listing, but this leaves unanswered the question of whether they would have found the business by other means. A verified listing confirms that the business exists, that it operates in the category in which it appears, and that it can be found again. Sales remain outside the scope of this confirmation, and a sales claim made on the basis of a listing is one that no honest analysis can substantiate.
A starter project: auditing the listings of a local firm
Anyone wishing to enter the field has a small project at their fingertips, with real data and a result they can demonstrate. A local firm, belonging to a relative or a friend, appears in several dozen places on the internet. The candidate makes an inventory of these, notes where the name, address, telephone number or opening hours differ, adds UTM parameters to the links they can edit, and tracks where visits and calls are coming from for a month.
The project requires almost everything an employer looks for in a beginner: data cleansing, attribution, a brief report and a recommendation. It also has the advantage of being verifiable, as the listings are public, and the business owner can confirm whether the phone rang more frequently.
An honest report on such a project also states what has not been measured. If, in a month, there were eleven visits from the listings and two calls, the sample size is too small for a budget recommendation. It is sufficient for a data-cleaning recommendation, however, as incorrect data is corrected anyway, and an employer reading the report sees a candidate who knows where the evidence ends.
Why the field is growing
Marketing analytics is expanding because digital marketing is generating more and more trackable behaviour. Every scroll, search, click, add-to-basket and unsubscribe leaves clues, and companies want people who can interpret them without getting bogged down in the details.
Changes in the area of privacy are also reshaping the field. With stricter data regulations, restrictions on cookies and more vigilant consumers, marketers need better measurement strategies. First-party data, consent practices and cleaner attribution models are receiving more attention than they did a few years ago.
In April 2025, Google announced that it would retain the current approach to managing third-party cookies in Chrome and that it was scrapping the separate pop-up window through which users would have been asked for consent. Safari and Firefox have been blocking them by default for years, so measurement remains fragmented: some of the audience can be tracked from one site to another, whilst others cannot. The company’s own data, collected with the customer’s consent, is the only source that behaves the same way across all browsers. For a small business, this means an up-to-date customer list and a contact form that asks where the visitor is coming from.
With less tracking data, the places where customers search for themselves, with a clear intention, are becoming increasingly important: search engines, maps and category directories. In social media feeds, which companies appear is determined by an algorithm over which the company has no control – a topic covered in an article about how social media algorithms choose the companies we see.
For you, this means that the industry rewards adaptability. Tools will change, platforms will update their rules, and consumer behaviour will shift again – probably as soon as a team finishes its quarterly presentation. The core principle remains the same: link data to real decisions, explain the results clearly, and keep an eye on business outcomes.
