Archives for 2019

Conversion Rate Optimization Will Fail Without Analytics.

Once a business is making sales, improving conversion rate is obviously attractive. And Conversion Rate Optimization (CRO) is a great way of tackling this.

But there is preparation required to make CRO successful. Before you have to commit to a CRO program. And before you can decide what order the various projects should be tackled in.

Don’t get me wrong. There are some great tools. From the Google Experiments Framework through to specialist tools like Optimizely and VWO. And marketing agencies to help. But they can’t define the business context for you. They can’t decide what is most important. Or what the criteria for success would be.

Think of it like employing a new staff member.

Before you get anywhere near the interview, and stuff that varies with the particular job..
You need some context, some criteria. You need to prepare

  • what you want them to do.
  • how you’ll judge success or failure
  • and a candidate shortlist.

It’s the same with Conversion Rate Optimization.

Google Analytics is a great tool for preparing, and setting context
It’ll help you identify:

  • the candidates for optimization.
  • what the baseline is.
  • what an acceptable improvement goal would be

And inform the wider business discussion

  • why improving a particular step is important?
  • whether AB split testing (best of two) is appropriate?
  • or we should do more complex multivariate testing?
  • how the improvement goal would impact the broader business context?

I’d be happy to have a chat about putting the foundations in place. So you’d be in a much better position to set the context for a successful CRO program.

Why is the Marketing Reporting Overview missing?

Most Marketing Reporting leads you into a maze. It’s almost impossible to get an effective overview. It’s either missing the rest of the marketing – or sufficient detail to be useful.

How did we get here?

Forty years ago in 1979 Visicalc spreadsheets were invented for the Apple II. During the 1980s software spreadsheets developed rapidly. Soon anyone could generate and print graphs.

This was a hugely important moment. A real democratization was occurring. No longer was costly printing the only way to get neat charting.

However the skills to use this capability well did not increase at the same rate. It became easy to flood the world with badly considered graphs.

Marketing Reporting isn’t immune

Anyone can put a pile of figures into Excel and ask it to produce a chart for them. The charts can look smart. It is far from obvious that the reader will understand the chart or draw the right conclusions.

The problems of user understanding doesn’t only apply to charts in spreadsheets. It applies everywhere.

The weather is rarely constant for those of us living in Britain. In the 1970s our TV screens had weathermen. Back then they were all men. And they showed us maps full of isobars. So we saw high pressure and low pressure areas and cold fronts night after night. As though these meant something?

It took decades for the “presentation” to change. The technology changed from moveable magnetic stickers to huge digital screens. But the underlying communication didn’t improve at anything like the same rate.

It’s only in the last few years that presenters have started to talk about the jetstream. And started to show either this causing

  • warm weather to arrive from somewhere shown as a yellow overlay
  • cold weather to arrive from elsewhere shown as a blue overlay.

Yellow – warm sun. Blue – cold. Much easier to understand. Fantastic.

What about Google Analytics?

Free to anyone who wants to use it. Great. Does this mean everyone gets great value from it? No!

Elsewhere I’ve written how you have to teach Google Analytics what matters. Teach it how to understand your business.

But there’s another problem. There’s a mass of data being collected. It’s easy to assume that everyone wants to see it all. But that’s flat wrong. In fact most people probably don’t want to see any of it. You have to convince them it’s worth their time.

The number of reporting options can be intimidating.

There are 95 standard reports in Google Analytics. Without counting the reports that have multiple sections. For instance like the goal funnel report can have 25 reports under the one heading.

In addition there can be many other custom reports. These can use the hundreds of Google defined dimensions and metrics.

And you can use rules to define segments. To include or exclude particular dimension values. Or to define segments by applying rules in a particular order.

But Google Analytics is just the “front end” of marketing.

If we add reports from CRM and other marketing systems we could have hundreds of reports.

So selecting which reports matter is hard. And marketing will be judged on the decisions made.

Because every report reader muses brutal questions like:

  • “Why should I spend my time looking at this?”
  • “Is this graph just another way of wasting my time?”

This is the real problem.

Let’s compare Marketing Reporting to Finance Reporting.

Just 3 reports are used to assess the vast majority of companies. These the Profit & Loss, Balance Sheet and Cash Flow statements. People accept them to provide the high level overview.

Shouldn’t marketing have a similar overview? Where are the overview charts for Marketing?

That are equivalent to these financial statements?

As you’ll guess I think the answer should be YES. In fact if marketing is to prosper it absolutely needs to be YES.

So can we pull the rabbit out of the hat?

Can we pull the Rabbit
Out Yet?

Whilst it would be the classic end to a magic trick I’m going to suggest a different approach.
Some authors would leap straight in at this stage and attempt to make some choices and then justify them.
Whilst this is legitimate there are serious problems with that approach.

The big picture gets completely lost. People wind up in a zero sum game. The proponents for each chart type can argue their case with religious zeal.

No. Not Yet.

So it’s much more helpful to set out the criteria that these overview charts should meet. I suggest:

  1. The report should change interactively as the user inspects it. This assumes that the chart will be shown on a screen rather than paper.
  2. The axes (and ranking) must be dimensionless. A single chart with axes must be able to display any quantity. This includes sales, conversion ratio, impressions, number of customers.
  3. It must be easy for readers to see at a glance what is good (acceptable) and what is bad (unacceptable). Using up for good and down for bad feels too simple to be helpful!
  4. Not all information can displayed all the time for every item. The initial report should show a summary of all the items. And allow the user to get more detail when required.
  5. Teaching users online how to use new report styles is acceptable. For comparison the financial reports have always needed explanation. As anything unfamiliar will.
  6. The plans and goals of the business should be shown beside historical performance.

What else would you suggest? Please put your comments in the area below…

Grow Your Agency Effectively With Analytics Engineering

So you want to grow your agency? in this article I want to talk about the way in which Analytics Engineering can help..

You’re looking for..

  • New Clients ?
  • More spend from existing clients?
  • Working with bigger clients?

Well thinking about these in turn..

Getting new clients.

Obviously you’ll want people to talk to. But once you’ve got them…

You’re going to want convincing stories to share..

Stories about your past successes? What you did, and crucially what the client got from it?

If you’re selling to any significant company – you’ve several stakeholders to convince. Some of them may believe marketers are like used car salesman. That it’s going to be hard to tell fact from fiction.

So apart from the brochures and whitepapers what can you show the hard bitten cynic? The CFO ?

Well how about the kind of clear reporting that can’t be argued with? Numbers that relate the marketing activity to business outcomes are good. Like qualified leads, or transactions. Exactly the kind of numbers that I’d expect a good analytics account to produce.

Imagine the difference? You’ve got trusted third party reports. Google will help sell them on how effective your agency is! It won’t outweigh all the other things you’re already using to help you sell.. But it won’t half help?

How about more spend from the existing client, then?

Extending the scope of your involvement involves buy in. Your immediate contacts will have to convince higher management. Lots of people will want to know that your marketing is working for them. And there are going to be cynics there too. Those who want to know the numbers? So again if you’re wandering in without the tools – you’re going to find it much harder than it need be!

To grow your agency you’lllYou need Case Studies – war stories? AND numbers.

Clear reports make it much easier to have a high value conversations with directors. Achieving the same buyin with poor reports is hard.

And the other area you’ll be looking to work on – is client relationships. That’s an area where – as the second part explained – the analytics engineer can help.

Working with bigger clients

Technical employees often don’t appreciate commercial nuances and imperatives. These are more pressing in bigger clients.

So imagine a situation like one I recently had.

Imagine if your agency could introduce your “GA expert”. It the expert could lead a multi way phone conference. That’s you, GA expert, your client, web development partner team to client.

It’s not easy doing web conferences. Particularly if side conversations can happen between participants in the same room

In this case the client wanted to proceed. But the web development “partner” had expressed doubt that it could done.

My game plan was to talk through the data gathering needed. I had analysed the website. I had an informed view about how implementation might work.

I created a situation where I could quiz the partner on the details of their current solution. I could listen for the situation where the client hasn’t grasped the detail of what the partner has done. Despite the client believing they understood.

We wanted to leave the partner “having” to play ball to secure their relationship with the client.

Now you could get an employee to do this – but it wouldn’t be a trainee – it would be a senior (expensive) employee.

That kind of meeting enhances your relationship with the client. You’re helping them deal with difficult but important partners. You’re ensuring that the stakeholders see improvements from your involvement. That intransigent third parties can’t delay things.

As you grow your agency it will be increasingly involved in multi party projects. There will be peers – and even potential competitors round the virtual table.

A tame expert can exert rare skills that enhance all the other relationships. The client will value this. It makes an important contribution to building your business.

Wrap Up

I hope these articles have given you a new perspective on Analytics Engineering.

In the first article I looked at the contribution Analytics Engineering could make.

In the second article I looked as how the freelance model would help many agencies work with bigger clients.

And finally that this could help you grow your agency deal with complex customers. To follow this up with a chat – just fill in your details in the box below.

How Freelance Help enables Better Analytics Engineering

Analytics Engineering opens new kind of conversation. In these 3 articles I’ll explain how. In this article we’ll explore how freelance help can be a good model for many.

Some will immediately object that external help is expensive. It costs more than internal resource. I want to explore how the wider benefits of expert help should be part of a proper cost benefit analysis.

Let’s tackle costs of external help up front.

Freelancers vs longer term commitment

Freelance help is a very effective way of getting someone who is expert in the area you want. Your agency doesn’t have to find work to fill the expert’s week. Nor pay the expert rate for that “lesser” work. You pay as you go.

So the investment depends on the size and scope of the project.

Ongoing Relationship

Some will be concerned about commercial confidentiality.

However commercial confidentiality and reputational risk make external experts risk averse. The junior employee may be much more cavalier.

Independent small businesses and agencies both understand that business development requires investment. So both may invest time to win an opportunity.

Greater Skill Levels

The other issue is the skill level. Freelance help gives you access to the bits of expert help you want.

Developing clear reports is the aim. But manipulating the more arcane areas of Analytics is often required to clean up the mess. Regular expressions, Google Analytics filters and views aren’t designed for enthusiastic novices.

Particularly when there are subtle consequences of these.

Expert Analytics Engineering isn’t about the buttons. It’s about how the website interacts with other web systems to achieve a result.

Using Google TagManager is very similar. It’s a bit like keyhole surgery. You’ve got a small keyhole through which you want access to everything going on the website. The task is to implement reporting. Using Google TagManager itself isn’t hard. The challenge is understanding what systems you’re accessing. And deducing which triggers will be effective.

Expert Freelance Help shouldn’t be cheap. Experts Aren’t.

Recruiting an expert Analytics Engineer as an employee would require a considerable salary.

Chances are you’re comparing apples with pears. A qualified Google Analytics Individual isn’t similar to the freelance Analytics Engineer. For some tasks – the GA skills alone will be fine. But as the complexity increases it stops being about pure Google Analytics.

For instance I’ve seen comprehensive reporting that failed. Because the developer’s point of view was treated as paramount. And understanding the Google Analytics Reports relied on coding concepts!

At that point one needs the experience to stand back. One must consider what clear reporting to fit with marketing goals should look like. But the analytics engineer needs to understand the internals. Together these can enable a workable compromise.

Wider value

Expert analytics engineering is much wider than pure Analytics. Three of the other ‘non marketing’ skills that many agencies need are:

  • Technical facilitation in multi faceted projects
  • Liaising with a range of experts from multiple companies involved.
  • Reading code and instructing programmers.

For instance a marketing agency trying to sort out the shopping cart for the client may have to liaise with:

  • Website Developers
  • Shopping cart providers
  • Client Marketeers

Freelance Help with Technical Facilitation

The agency has the marketing skills. But the need to deal with CTOs, and other agencies is often a real challenge. The latter may feel they own the site. They may feel confident enough to ‘worry out loud’ that something won’t work. And lapse into jargon to confuse both client and other stakeholders.

Analytics Engineering can find a way round problems that vex many marketing agencies. Part consultant, part technical project manager, part geek. The marketing agency and client often want to know if the advice given is reliable.

Technical facilitation helps the client resolve the logjam when multiple interests conflict. Or when the common language that would resolve matters is lacking.

There’s a long list of skills and plenty of jargon – including:

  • HTML
  • AJAX
  • Network Protocols
  • TagManager
  • Analytics
  • Web Server configuration
  • Domains, subdomains
  • cookies

Analytics Engineering can bring along a range of skills. Skills which are extremely useful in unblocking projects.

Ask yourself – what if you had not only analytics skills but a wider range of technical skills. Wouldn’t your agency find easier to unblock important projects. Wouldn’t it be more confident about expanding into new areas?

Solving Integration Challenges

Modern developments often rely on complex system interconnections. The analytics engineer will understand how clearly the business requirements have been defined.

Some challenges will exist for good reasons. Some will exist for poor reasons. Clarifying the reporting is often challenging. It can be one of the more intricate parts of the solution. Particularly if one wants to avoid dramatic change to the website or infrastructure.

Diagnosing deep seated problems

As an example.

  • I had to step through the customer journey.
  • I inspected the raw HTML, and javascript to work out what was happening.
  • I used network monitoring software to confirm what data was being sent. And to which server.
  • I then had detailed conversation with the software system provider technical support.

We were clarifying and resolving problems such as:

  • Complex Page Naming
  • Many subdomains
  • Caching
  • Multiple devices
  • Timing & Performance

Conclusion – Freelance help

So using freelancers enables you to get the help you need. You can dip your toe in the water and find out how the relationship works. All the agencies I’ve worked with come back for more.

Now smart viewers will use this article as a checklist. Many may claim to be expert in Analytics. But many of these spend much more time working with some other aspect of web marketing (e.g. SEO) than with Analytics. And they haven’t the wider technical background to back up the analytics. In short they aren’t analytics engineers.

In the next article I’ll explain how you Analytics Engineering can help you grow your agency.

How Analytics Engineering Can Transform Agency Client Relationships

In this series of 3 articles I want to explain how Analytics Engineering opens up a new kind of conversation. A conversation between agencies and the wider corporate. Not just the immediate marketing contacts that you sell to.

Distrust of agencies is real

There’s often a fundamental distrust of marketing in the corporate. And people aren’t magically more comfortable with spend on marketing agencies. Buyers will pick up on this culture.

It doesn’t make it easy to argue that your agency should become more important to the client.

Clients use marketing agencies. And I’d emphasise the word use. Certainly clients will want to use agency skills to help market their business. But sometimes use agencies as if they are just a commodity.

Larger corporates use a number of marketing agencies at the same time. Becoming the sole marketing agency for a larger corporate may be impossible. Your agency can’t necessarily gain all the client’s business.

But you can be first in line when new opportunities come along. Particularly if you deliver great value. And this makes other procurement criteria less important.

It’s hard to measure the value of a marketing agency. So it’s logical for buyers to want to:

  • run a continuous competition among multiple agencies.
  • use the insights of the best agencies to drive up the general standard.
  • keep them lean and hungry.

So how can you respond?

Distrust & Poor Relationships

Using Measurement to move beyond Distrust

Analytics is now the currency in process management. Six Sigma has spread across most sectors. The core tool for Six Sigma projects for business processes is DMAIC. This stands for Define, Measure, Analyze, Improve and Control.

This doesn’t leave any room for doubt. Improvement & Control come after define, measure and analyze.

It makes perfect sense. Everyone’s lost if you don’t know what’s working, what’s failing. There’s little opportunity to improve. No ability to control.

So other disciplines have the same focus on analytics as marketing.

So Chief Financial Officers and Operations Officers will recognise the approach. But only if you state it. It’ll help build trust.

Building Trust With Analytics Engineering

With Analytics Engineering your agency can show it is better than the others. You don’t have to rely on simple assertions that you’re more innovative or more creative. Marketing can show other disciplines a commitment to improvement and control.

You can sidestep the corporate procurement impulses. Move beyond distrust.

Analytics Engineering helps your agency provide the best advice. You have the information to recommend an increased spend. Or suggest the waste be used better elsewhere. You can delight your client either way. You become a trusted advisor.

High quality measurement and analysis mean clients ask better questions. People can see how complex the market is. They can discuss how to conduct better marketing trials and campaigns.

This will change the relationship. Clients won’t be able to adopt a binary approach. They won’t be able to simply ask “did they bring us results”. You can ask for better quality direction. More involvement.

So the client and your agency have to work together. High quality measurement makes partnership a natural model. Optimising the shopping cart becomes a series of experiments. These build the partnership.

Everyone has a greater emotional investment in the process and the results.

Reporting Clarity inspite of Customer Journey Fragmentation

This is a real challenge for marketers. It’s one reason that one size fits all advertising is less popular. The proportion spent on Direct and Digital Advertising has been rising steadily. It’s expected that it will account for over 50% of global spend in 2019.

All sorts of factors contribute:

  • Growing Buyer Sophistication.
  • Enhanced information that enables more precise targeting.
  • More use of Multi channel campaigns to establish one to one relationships.
  • More frequent buyer/marketing interactions.

But none of this makes impenetrable reports acceptable. The reports should be even clearer. And avoid burying management in unnecessary detail. Clear reports will mean wider questions get asked. Detailed investigations will be spawned.

And the evidence I have is end customers react well to clear reporting. Clients invite the agency to contribute in other areas.

Much of my work relies on understanding the customer journey. And ensuring that Google Analytics provides clear numbers that suit that business.

Using Analytics Engineering to Outpace Competitors

Your competitors measure and collect data to track all their campaigns.

To compete you’ve got to measure outcomes and offer differentiated marketing.

But your competitors are often using bog standard measures. Measures that don’t reflect the core concerns of a sophisticated corporate client. Producing higher quality reports is an opportunity to stand out.

And if you don’t…your client is either going to look for that from someone else OR

Remain severely handicapped at the very least.

Clients remain less financially efficient. Earning lower profits than they could be.

Without analytics engineering you and the client are accepting this chronic handicap. Accounts are at risk from any other agency that walks in offering them smart analytics that pays for itself.

So to conclude. Developments in marketing have pushed analytics engineering to the forefront. Analytics and data are key. Engineering can achieve the right blend for your clients. Analytics Engineering is a smart way to set your offering apart.

Wrap Up

In the next article I’ll explain how you don’t have to invest in expensive recruits & skills training. How you can achieve similar results with cost effective external help.

7 Reasons Why Your Marketing Dashboard Can Fail

The typical marketing dashboard enables data to be thrown onto a screen. I chose “thrown” deliberately. It may look appealing, but the viewer can be easily misled. They can feel as though they are wading through a mess of data.

Your Data isn’t clean Sir..

The dirt can be well hidden. But destructive. This makes it hard for the reader to understand the story you want to tell.

To explain…

Many dashboards emphasize their ability to include data from online sources. This source data is simply fed through to the dashboard.

Sadly this simple flow creates the problem. The data in the typical online system has many problems. These include:

  • There are bits of data that should be filtered from your dashboard. Possibly from development or other test scenarios.
  • Encoded names are like jargon. The names can include countries and other identifiers. These have meaning to those that manage the online system. But the marketing dashboard viewer may be a business executive. This executive doesn’t know or care about the internals of some online system. So the jargon can make viewers reject the dashboard.
  • Splintered items due to case mismatch. These “duplicated items make reports useless Aggregation would make the reports readable.

Your dashboard charts don’t help readers with their real problem

Data gathering normally captures as much easy data as possible. The marketing dashboard can then show this presented at will. For example Google Analytics can easily capture every page on a website.
The data gathering may be deep as well. It may capture lots of detail. This allows the reader to segment the data.

For instance.
Google Analytics events are often labelled with the page. This means the reader knows on which page a particular event type occurred. There can be thousands of different pages on which even a single event type occurs.
Let’s suppose each time a web visitor downloads a brochure for a kettle an event is counted.
And suppose our dashboard user looks after household electrical products.

It still requires much work to make the numbers useful to our dashboard user.
1) We should be able to add kettles to all the other household electrical brochure downloads.
2) But we should also be able to see the kettle downloads on their own.
2) We should be able to exclude countries that the dashboard user isn’t responsible for.

But segmentation by page doesn’t aggregate products. And aggregating all products could include non electrical product downloads.
Geographical segmentation of viewers doesn’t precisely match responsibility for marketing in Southern Europe. The former is about the location of the potential consumer. The latter is about the location of the vendor. With easy global distribution of products we cannot assume these two are the same.
So our dashboard user has several nasty choices. None really address the precise problem the reader has.

The more we leave the user to do, the less likely it is that the dashboard will help answer their real questions. We relegate the dashboard to “interesting.. but not crucial”.

Your Data hasn’t been polished prior to display on the dashboard.

The example above shows how much we have to manipulate data to make it crucial information for the user.

Bad compromises result when the data doesn’t match what the user needs.

A dashboard can allow the user to segment and manipulate data. Yet this forces the senior manager to become an analyst.

Polishing takes work

I believe this is an abdication of responsibility on the part of the dashboard provider. Is the dashboard vendor just hoping that the consumer won’t notice?

Your dashboard doesn’t give adequate context

Dashboards often show data on a continuing trend. The dashboard user may compare this month with last month visually.
But the data has other relevant contexts. There may, we hope, have been a plan. So the dashboard should allow the user to compare the actual result against the plan.
But can every user see if it’s too far away from plan?
Think about the plane flying from London to New York. It will be off course for most of it’s journey. But the auto pilot will make sure the error is minimal. It would be a disaster if the plane ran out of fuel off the Brazilian coast.

Your dashboard doesn’t show comparators or tolerances

Any competent engineer understands the concept of tolerances. For instance diameter of each piston within a combustion engine has a tolerance. Each piston will be slightly different. But no piston can be so different it doesn’t fit into the cylinder.

It’s the same for marketing dashboards. There ought to be a tolerable deviation from plan. This enfranchises the non expert dashboard user. As a side note, the discussion of deviation is tolerable helps improve the plan. The qualitative what is good vs bad. What is better or poorer may be hard to assess.

Your dashboard assumes the user can accurately assess trend

Whilst most users have good eyesight this isn’t a good assumption. Changes in longer term trends can be difficult to see. Short term fluctuations may hide any acceleration or deceleration.

Struggling to see the trend is downwards

It’s like assuming that people know whether an approaching car is speeding up or slowing down. In this case the viewing angle makes a huge difference.

So the data often needs to be smoothed for the dashboard user. And manipulated. These can reveal subtle but important changes.
These changes may be important indicators of environmental changes. The effectiveness of marketing locations may be changing. Early warning of these is really helpful.

Your dashboard for marketing doesn’t relate to external objectives.

The client business has objectives. The marketing campaigns are subordinate to these. If the business executive can see a “line of sight” from a chart to a current business objective it’s much more relevant.

And if the link isn’t obvious the dashboard should state the link.

The client business has objectives. The marketing campaigns are subordinate to these. Can the business executive see how a chart relates to a current business goal? If it’s obvious the chart is much more relevant.
And if the link isn’t obvious the dashboard should explain the link.