Marketing Analytics: phases and stages

When we do marketing seriously, one of the tasks we have to take care of is analyzing the performance of our efforts.

This analysis, of course, will vary in form and substance depending on the resources invested.

It is logical to think that the scope will not be the same if we invest 100€ a month as if we invest one million.

In the first scenario we will probably use only one or two channels, while in the second there will surely be more than ten.

That is why, as our project grows, the analytics we apply to it will be different.

We will go through different phases. We will reach different stages.

And in this article I want to show you the five main ones, which I summarize in this table:

Phase 1: Google Analytics

The easiest way to start measuring “something” in our marketing or digital product.

If we have a digital project with a website or app, it is the first tool we will use, since it is free, very easy to implement at a basic level and gives us plenty of information.

It is the initial phase of marketing analytics.

What you can measure with Google Analytics

When properly configured, it lets you measure things like:

Your users' interactions

• The traffic from your TAGGED campaigns
• The Bounce / Engagement rate of your landing pages.
• Products / services viewed.
• Products / services added to cart.
• Users who enter checkout => Remarketing.
• Other events on the site.

Characteristics of your users

• The devices your users use.
• The countries they connect from.

Conversions

• Conversions by campaign (LCI or “data-driven” model, depending on the report).
• Product purchases and returns.

Advertising on our site

• Promotions on our website (banners)
• Products shown to users.

What Google Analytics does NOT let you measure

Although it gives us a lot of information, there are also things it lacks and that we will have to measure in other ways, for example:

• Campaigns in traditional media (TV, press, outdoor).
• Digital campaigns without clicks (brand banners, email opens, YouTube video views): postview.
• Multi-device customer journey. There are attempts, but they are patches.
• Actions by users who do not accept cookies or use adblockers.
• Complete attribution model (biased data).
• Personal data (CRM).

What it measures only partially

• Clicks on your website (you could measure all of them, but it is not recommended).
• Campaign costs from platforms other than Google (data has to be uploaded).
• SEO keywords (you see very few).
• Sociodemographic characteristics (Signal)
• Data that is not collected directly (JS errors, adblockers, Safari): you will always be missing a % of conversions. The gap gets bigger every day.

Advantages of Google Analytics

Quite a few, actually:

• Getting basic information is quick. We can be collecting data within minutes.
• It is the easiest way to start measuring “something” in our marketing or digital product.
• It is the best free digital analytics tool. And it is better than many paid ones.

Disadvantages of Google Analytics

Of course, it also has some disadvantages:

• It does not reflect the reality of the business. Not even that of the marketing department. Nor that of digital marketing. Because it lacks the data to do so.
• In other words, it only measures a small part, so it is not a good idea to judge an entire business by its Analytics data.
• GA4 today is worse than GA3. It requires a more complex technical implementation in a more complicated environment (GDPR, privacy…)

Summary

For all of the above, I think Google Analytics falls short.

Even to measure only the performance of a digital product (website or app), you need other complementary tools.

Let alone if what you want is to have control of a complete business, however digital it may be…

In any case, it will be the first step in analytics.

Phase 2: Reporting by digital channel

Once we grow and have several digital channels, especially if they are paid (Google Ads, Meta, affiliates…), we will continue analyzing overall performance as in the previous phase, but we will also begin analyzing each channel separately, in greater depth, to try to optimize it.

Frequency

This analysis will vary according to budget:

• Below 2 or 3K/Month it will be monthly.
• Above that it may be weekly.

And the thing is, we need to see a good volume of data to make decisions. And that volume requires more or less time depending on the size of the project and its investment.

Owner

There will always be someone responsible for presenting it and drawing conclusions, who may be:

• Internal team: if marketing is managed inhouse.
• Freelancers: if we use different ones depending on the channel.
• Agency: when we hire one to manage all paid or SEO for us.

What it includes

Depending on the channel, this reporting will include some things or others. To give you an idea, these reports will include:

• The progress of the campaign.
• In paid media, the focus will be sales.
• In SEO, positioning and ORM.
• On social media, reach or contacts and incident resolution.

Objective of channel reporting

In fact, the main one is to see how we are doing against the business objectives set. Because based on this, we will have two scenarios:

• If we are on track, it will just be monitoring.
• If we are below target, we will need proposals for actions, for example:
  • Change in investment.
  • Cambio en la oferta.
  • Cambio en el mix de medios.
  • Cambio en las creatividades.
• …

You should know that direct-response digital campaigns allow very quick changes.

Data source

All this information will be provided by different tools. So we can use:

• Platform reports (Google Ads, Meta…).
• Google Analytics.
• An AdServer, if we use one (recommended for high budgets, since it has a cost)
• Search Console.
• Metricool
• …

Examples

Let's look at some images of this type of report so you can get a better idea:

Weekly paid media report

Monthly SEO report

Monthly Social Media report

To sum up

In this second phase we begin to understand our digital marketing: how the channels relate to one another, acquisition channels and conversion channels.

That said, they leave aside everything that happens in the digital product: it has to be combined with Google Analytics (and some others) to give us a better view of what is happening on our website.

In addition, they require time to prepare and to read.

Because of the above, sometimes they are limited to producing numbers rather than providing information and recommendations, which is what matters.

Phase 3: Econometric models

What an econometric model is

An econometric model is nothing more than a mathematical simulation of reality, which helps us understand why something happened in a certain way based on certain variables.

And the good thing is to use this knowledge for future decision-making.

The idea is to explain a dependent variable (generally revenue or sales) based on other variables (explanatory variables), such as website traffic, price, public interest, advertising investment or our brand's NPS.

When to use econometric models in marketing

As you grow, you start considering actions outside digital channels. That is when you begin using econometric models.

It is not that this is newer or more advanced than monthly reports by tool; rather, it serves to provide another type of information.

The frequency will be at least monthly. It may increase depending on the project or seasonality (it is not normal to be doing television continuously).

Who is responsible

The person responsible for preparing it may be:

• Internal team: if you have a strong analytics team.
• Marketing or media agency: when we hire one to manage all paid activity or media for us.
• Attribution agency / tool: independent of the agency. Better, but more expensive (another fee to pay).

What is included here

Basically two things:

• How our investment works in the long term, not just in the short and medium term.
• It includes traditional media.

What the objective is

• See which media generate sales.
• See how external variables affect things (competition, weather).
• Adjust the investment as well as possible.

There are multiple data sources

Even more so than in the previous phase:

• Google Analytics.
• AdServer
• Attribution tool.
• Search Console.
• Pixel from advertising platforms.
• CRM.

Advantages of econometric models

It has many, some very obvious ones that the previous phases lacked:

• Includes non-digital channels.
• And external variables (seasonality or competition).
• Provides strategic, actionable information.
• Explains the past and helps make projections for the future.
• It helps a lot to create the optimal setup for the marketing plan.
• Despite its complexity, one can even be made with Excel.

Disadvantages of econometric models

Of course, it also has a number of disadvantages, which make it suitable only for projects of a certain scale:

• It is not easy to create a robust model:
  • The variables must be selected carefully.
  • Tener el histórico de datos.
• Being able to produce it and present it in the right format.
• This involves a laborious task that is difficult to automate (many data sources).
• You need A LOT of data to make it reliable. It is not valid for small projects.
• If we outsource it, it is not cheap.
• Agencies are judge and jury.

In summary

We now understand quite well how our marketing behaves, not just the digital side anymore, and how our investment performs.

Again, the part about the digital product (conversion rate and events) is largely left aside, although we can include them as variables.

You have to believe in the model. It is an act of faith (until it proves its validity). If we don't believe in it, that's a bad sign.

It is costly. In time, money or both.

Phase 4: Attribution

Econometric analysis is usually carried out when we make the leap to non-digital channels.

But we may never make that leap.

Then, to advance the analytics of our business, we will focus on Attribution, which is nothing more than assigning a conversion to the channel or channels that made it possible.

Let's look at an example journey to make it clearer:

This user has entered our website through different channels, with different devices and at different times, until they finally converted.

So, which channel do we attribute that conversion to?

• To the first one?
• To the last one?
• To several?

Attribution can be single-channel or multi-channel, depending on how many user touchpoints with our website are taken into account.

In addition, using an attribution model helps us understand that there are conversion channels “goal poachers”, such as remarketing or brand SEM, and assist channels , such as email. There are also others that open the funnel.

Ultimately, you have to understand what each one contributes.

The most common attribution models

Popularized by Google Analytics, we could say the main ones are the following:

• Last-click (single-channel) => advertising tools.
• Last-click indirect (single-channel) => Google Analytics, in most reports.
• Time decay (multi-channel) => Google Analytics.
• Data-driven (multi-channel) => Google Analytics and other advanced analytics tools.

GA4 is based on the data it collects from the website and, if we configure it that way, what it knows about the user, but it lacks information.

For example, it does not include advertising impressions and the effect of post-view, nor information about offline channels or a user's different devices if they are not logged in.

That is why Google Analytics is not the best tool for doing attribution correctly.

Other tools (not free) solve these problems and incorporate more variables, not just media (CRM, calls, store visits, competition, sign-ups…).

With all these variables, a statistical mathematical model similar to econometric ones is applied.

What this phase includes

• As in econometrics, the contribution of each channel to sales.
• Although it often includes information only from digital channels (simpler), in reality you can add whatever is required to the model (including investment in traditional channels), depending on the tool.

Objectives

• See which media generate sales.
• See how other variables affect things.
• Adjust the investment as well as possible.

Data source

Again, there are multiple sources:

• Google Analytics and AdServer.
• Surveys.
• Business data.
• Store visits.
• BTL.
• CRM / calls / SMS.
• …

Advantages of attribution

Very similar to those of econometric models:

• Includes all the information we need (at least with advanced tools).
• Provides very actionable and strategic information for investment decisions.
• Explains the past and helps make projections for the future.
• It helps a lot to adjust the optimal setup of the marketing plan if we think the approach can be improved.

Disadvantages of attribution

Again, similar to those of the previous phase:

• It is not easy to create a robust model:
  • The variables must be selected carefully.
  • Tener el histórico de datos.
• Being able to produce it and present it in the right format.
• This involves a laborious task that is difficult to automate (many data sources).
• You need A LOT of data to make it reliable. It is not valid for small projects.
• It is usually outsourced and good tools are not cheap.

Summary

This stage of analytics is great for a business's digital marketing department. Especially if no actions are carried out in offline channels.

It is a very reliable analysis of the channels that most influence sales, which helps distribute investment appropriately.

Variables from events that occur in the digital product (microconversions, relevant page views…) can be added gradually.

In addition, they require a team and external tools and time for internal preparation to feed the model.

Phase 5: Incrementality

Attribution distributes conversions among the different marketing channels.

But the truth is that not every website conversion, nor every sale a business makes, is the result of marketing.

Remember the model Product Led Growth, for example.

Or think about supermarket private-label brands, which have never run a campaign and still sell.

That is why, when understanding sales, there are other factors at play besides promotion, such as distribution or the baseline earned.

And, working in marketing, the ideal is to understand the contribution we actually make. That is incrementality:

Chart taken from this very interesting article:

https://www.marketingweek.com/ehrenberg-bass-advertising-growth/

It is definitely worth reading.

How to measure incrementality

“Half the money I spend on advertising is wasted. The trouble is I don't know which half.”

John Wanamaker (1.838-1.922)

And the thing is, measuring it is not easy at all. Not at all.

It is so complex that in no project I have participated in has it really been measured.

 Here is an example of how it can be done, from the great Avinash Kaushik: basically, it means using different strategies with different groups of similar users and measuring their results.

Summary

It should be our goal, but the truth is that only large projects have the staff and financial resources to implement incrementality measurement systems.

Taking the last three rungs of marketing analytics, we could say that:

• If econometrics are ideal for communications departments and offline media…
• Attribution models for digital marketing…
• Measuring incrementality is the ultimate objective of the director of any business in relation to marketing.

Final conclusions

If we are talking about a small or medium-sized project, the normal thing will be to move up gradually, one stage at a time, starting with the first.

However, if we are talking about projects with greater investment, the normal thing is to start directly and at least:

• In scenario 2, if we only do digital.
• In 3, if we run TV campaigns.

I would tell you that you should really be obsessed with reaching phase 5 in your project, but I would be lying. I have worked with very successful businesses of different sizes and the only thing we had –including mine- was a desire to reach it, but without taking action in that direction.

Because a good marketing plan and proper analytics in phase 2, 3 or 4 are more than enough on many occasions.

That said, phase 1 almost always falls short, so if your project is at that stage and you are dedicating resources to marketing, do yourself a favor and make the leap at least to the second. Because even if you do not work with paid channels, SEO and email (to name the two most common) also need to be optimized.

And as your project and your marketing grow, do not forget that analytics has to go hand in hand and advance too. It is the only way to optimize that investment.

If you only take one thing away from this article, let it be this:

Marketing, and especially digital marketing, always, always, always has to go hand in hand with good analysis.

If you need help with yours, take a look at this business analytics course for ecommerce or contact me and we'll review it.