Measuring channel performance through web analytics is a key part of optimising any digital marketing strategy – being close to the data ensures we are able to adapt quickly when the need arises.
Sometimes issues that arise will be obvious, born out of a product of extreme change. These issues are easy to identify and can usually be fixed immediately. But what about subtler changes?
How do we identify trends that occur over a long period of time or problems that have always existed but have never been noticed? As Donald Rumsfield was famously quoted as saying:
“There are known knowns. There are things we know we know. We also know there are known unknowns. That is to say, we know there are some things we do not know. But there are also unknown unknowns, the ones we don’t know we don’t know.”
This paradigm translates well into the context of our understanding of marketing performance also:
Problems and inefficiencies will always happen, there’s nothing we can do about that. There are far too many variables that can occur to ever hope for a situation where we can realistically expect everything to run perfectly all the time and for external influences not to play their part. And when issues do occur, for whatever reason, they cost a business money. We cannot stop this, but what we can do is put the infrastructure in place to ensure we are in a position to identify problems or opportunities in the quickest time possible. Through this we minimise risk and maximise opportunities.
An individual problem will manifest itself very differently in the above model depending on the infrastructure of the business in which it occurs. The better the infrastructure, the more likely you are to spot a problem and know the best way to solve it. An example might be a small increase in bad quality leads due to an affiliate campaign generating ‘spammy’ leads. It is perfectly possible that in one company with a medium sized sales force that this increase in bad quality leads goes completely unnoticed as no one is measuring average lead quality on an ongoing basis.
In another company they may observe the variation in lead quality and report this to marketing, however, they have no way of identifying the campaign that generated the leads, so its guesswork from this point as to how to solve the issue.
Finally, you have a company that has an integrated analytics and CRM solution and a daily dashboard that reports on average lead quality on a source medium and campaign level. Here the problem is immediately spotted and the campaign paused as a review is done into why this occurred. Steps are then taken to make sure this particular problem is mitigated from occurring again.
From this relatively straightforward but surprisingly common example, we can see how, by recording the right data and reviewing it in an effective manner, we are able to minimise the risks that we encounter on a daily basis. Companies can sometimes see investment in analytics as a channel as inefficient as it is not a direct revenue generating solution. They then feel justified with their decisions as they look back and see that nothing bad happened. Unfortunately however, in this situation, problems will simply manifest themselves beyond the realms of what you can perceive. Ignorance is bliss, but it is also quite costly.
So what can we do to open our eyes, how can we make it so that when our problems do occur, they appear as clearly to us rather than going unnoticed?
It may sound obvious, but firstly we need to know what we are measuring any of our digital marketing strategies against. This will vary from channel to channel and from campaign to campaign but it is vitally important to define this in order to effectively manage strategies on an ongoing basis. Once we know what we are trying to achieve it becomes a lot easier to improve the strategy or understand if it is underperforming.
For a brand-focused social media campaign, you might look for sentiment analysis and mentions as your measure of success. This would vary greatly from a B2B lead generation focused PPC campaign which is ultimately looking at measures like CPC in the short term and ROAS in the long run.
Understanding exactly what it is you need to measure becomes a lot easier by defining SMART goals.
Once we know what our key metrics are, we need to make sure they are incorporated throughout any reporting done relating to the channel/campaign.
The technology you use serves as your eyes and ears to measure the KPIs that you have defined. Depending on the campaign and goals, the technology that you will need will vary greatly and choosing the right platform can be costly in itself with some products running into the £100k’s per year to use.
Factors that may influence the platforms you use to measure might include:
Once you understand the mediums that drive the most value for your business you’ll be in a much better position to understand which tools you might want to invest in.
Below are some guidelines on what we feel constitute a MVP setup for monitoring different aspects of a campaign.
Google Analytics – it’s an obvious choice, but for a reason. It has a huge capacity for analysis, with the free platform providing more than adequate solutions for companies even up to a relatively large size. A key point here however is that the more you put in, the more you will get out. It’s important to spend a little time configuring the platform to get the most out of it. Aspects like connecting Google Analytics to your CRM can significantly boost your understanding of performance.
The other benefit here is from an attribution modeling perspective. Whilst significantly less comprehensive than other platforms, so long as your traffic has the appropriate tracking parameters, you will be able to compare and contrast how different campaigns are impacting different parts of the user journey. Voucher code campaigns may look good from a last click perspective, but are much less likely to introduce your brand to new customers, for instance. Using and balancing different models gives you a good understanding of the interplay between channels.
Search Console is another free Google platform that gives you insights into your organic rankings. However to properly plan and execute a comprehensive SEO strategy, you will likely need to supplement this with a paid keyword ranking product that will give you a much more granular understanding of what you should be optimising towards and allow you to track organic performance in the niches that are key to your business also.
Depending on the size of your website, you may need to consider a site crawler. For large sites it’s incredibly difficult to understand how Google perceives your site structure. By investing in one of these tools you are able to ensure that no issues arise from an architecture perspective.
Your best bet for performance tracking in Social Media is using the platform itself through which you are driving campaigns for both paid and organic. It’s worth noting however that there is often significant overlap with social attribution and that of other third party platforms making it difficult to define the value of the channel exactly without investing in expensive attribution solutions. You may find that PPC and Social claim the same lead, however it is tricky to define where the merit truly lies.
Tracking wider trends in your audience is also important so that you can understand how your brand fits into the wider ecology of your sphere of influence. Platforms like Pulsar allow you to compare yourself against competitors and see sentiment shifts within the industry. You can, for instance, quickly isolate if negative discussions are taking place about your brand and take steps to mitigate this.
Similar to Social, there can be issues with Paid Media campaigns in terms of attribution depending on which platforms you are using to track performance. Generally however, using the ad serving platform is the best way to measure performance over time by looking at trends within the channel itself.
The only way to accurately measure the true benefit of display is by using a multi channel attribution solution that uses a data driven model. These solutions are great if you can get them, however they are costly and without them you have to take the benefits with a pinch of salt.
Billboards and print media have long accepted the value of visual ads and, whilst the comparison is not completely like for like, they still play on a similar psyche, reinforcing the brand in the mind of your customers on both a short and long term basis . To this end therefore, for most, the KPIs of display will often be limited towards and CPC or CPM. Direct clicks and leads may occur but this is not the main point of this channel. You can try to define the exact benefit of display but be warned, it is not for the faint of heart!
While often an undervalued area, well visualised and communicated data is worth its weight in gold. There is a subtle balance to this: too much and you become overwhelmed and unengaged reviewing and keeping an eye on performance becomes a laborious task – too little and you may miss an important change.
The problem is, what this balance looks like may change from one person to another. Creating a good dashboard therefore should be thought of in a similar way to getting fitted for a suit. Generally there will be a person collecting the data (the tailor) with expertise and knowledge of how all the threads fit together and a person responsible for reviewing the performance of that data (the customer).
If a tailor does not know the size and preferences of his customer, it’s likely that the end product will be ill fitting and will never be worn. Therefore, it’s key for any stakeholder to work closely with whoever holds the data to ensure that any dashboards or reports fit well with their preferences. This may take some rounds of feedback but once you have something that works for you, you will know its worth.
Your ideal scenario here is that, by having a comprehensive suite of dashboards and reports, any problems that do occur will manifest themselves as unknown knowns – you will see the signs even if you do not know the cause. However, with a comprehensive mertech stack it should be straightforward to turn this into a known known through a process of further analysis.
A common problem faced in many businesses is the siloing of data. Either due to lack of process or missing technology, information exists but ends up decentralised and lacking context as it sits alone.
If we are selling a product online then trying to upsell a subscription model offline further down the line, how can we accurately calculate the true LTV of a customer associated with a particular channel? Without some kind of integration, we won’t be able to accurately measure the return and know how much we can invest in future campaigns.
It stands as a universal truth that any data that can be combined will be more powerful when it is in that state. In combining data you not only gain more insights, it also becomes easier to process. Sometimes it is hard to know what is even possible to integrate and one’s ability to integrate also depends on budget and software. However some degree of integration is always possible and where possible, strongly advised.
The ultimate dream is creating a single customer view, however, in reality, the first steps for many are often smaller but still very effective.
A recommended solution is to begin by mapping out data sources where you contain information on a customer. This might consist of areas like email platforms, CRMs, social media etc. Once you have mapped out your data, you can then investigate all potential connections and assess what the best way to combine the most data with the least connections.
To understand what is unusual, we must first define what is usual. With that in mind, it is key to have an understanding of what normal trends look like. Key metrics fluctuate all the time so we must be able to pinpoint what lies inside or outside the boundaries of expected behaviour.
Historic data sits at the core here and there are many options available to us for predicting potential trends. Certain software can forecast future behaviour based on historical trends. One advantage here comes from having your data stored in databases like BigQuery. This makes it a lot easier to undertake predictive modelling.
There will always be unexpected behaviours that occur that can never be predicted, however having a base level at least allows us to spot when we deviate from normal as opposed to natural fluctuations in behaviour like seasonality. This allows us to prioritise our investigation towards trends that we are most likely to have some control over.
At its very essence, the point we are trying to make here is that, the better the reporting infrastructure, the less money your business will lose when things go wrong. Faults and issues are constant in life and we should always look to minimise them occurring. But at the same time, we have to accept the problems will occur that could never be realistically prevented.
A robust reporting infrastructure allows us to catch these issues and remedy them quickly, thereby significantly minimising costs to the business over time.
At Semetrical, we’re well placed to help you with measuring the performance of your digital channels and our analytics team thrive on delivering insights to maximise growth.