'Performance Marketing' myths and how to overcome them

John Wanamaker, considered by many a marketing pioneer, used to say that half the money he spent on advertising was wasted; the trouble was, he didn’t know which half. With the whole array of tools available these days though, it’s easy for a savvy marketer to find out the source of 80-90% of a company’s sales, and iterate to slowly but surely grow and conquer the world. Right?

Wrong. Sadly, the process is not quite that straightforward. Two years ago I set out to be a performance marketer - ‘a growth hacker’. Here are some of the challenges I’ve come across over the past couple of years, and what I’ve learned about how you can overcome them.

1. You can track everything

As an MBA graduate, I was challenged to create a marketing plan for a B2B start-up which outlined spends and channels. I’m proud of the attribution model I came up with, which gave a theoretical return on investment when aligned against a costing plan.

What it didn’t take into account, however, was that tracking sign-ups to those specific channels is hard, if not impossible. Yes, you can easily track the number of sign-ups driven via paid search (something I confidently told my clients whilst working at Google), but that doesn’t mean you can track which of those sign-ups were actually valuable unless they purchase something right away. Tying a later purchase back to the lead source that drove someone to your site in the first place requires a complex setup, involving significant data capture to enable all-round marketing attribution.

Even when things have been set up, each channel has its own set of key performance indicators that don’t always make for handy comparisons. If you drive traffic to your site via Twitter, for instance, you get clicks from specific tweets broken down across paid and unpaid efforts – on Linkedin, you can’t split these up.


Realise that guessing is inevitable. Back yourself and your team to pick three or four channels - give them a bit of money and a lot of time, and see what works. This article lays out what we all think but aren’t always keen to admit – mimicking your competitors, is a great way to start.

Don’t expect instant results, nor to be able to optimize from day one – or day 10, for that matter –, as you may end up working on a set of trends that don’t exist in real life. Minas Iartou speaks eloquently about the many variables that social media (and by extension, modern-day marketing) entails and how you can start to tackle them one by one, but it will take time and effort.

Finally, make a point of asking your customers how they found you. Record this in some way – ideally a CRM system – but be careful not to overinterpret what they say and make sure your analysis and decisions are backed up with hard data.

2. CRM + Marketing automation = instant lead generation machine

To clarify the terminologies here, a CRM is a tool to store and track leads as they move through your conversion funnel. To sales and operations people, a CRM is typically used as a means to track interactions with existing clients, and it’s useful for marketers, too – think up-selling and cross-selling. You can even do it in a spreadsheet, but (#protip) I’d advise against that once you have more than a thousand potential contacts.

Marketing automation is software that can assist you in driving your sales. Through a series of emails (also known as lead nurturing) or a content served up on “landing pages”, you can show customers, both potential and current, why they need the products and services you have to offer.

While the combination of the two can work wonders in automating and enhancing your direct marketing efforts, it takes some effort to set up. CRM systems in particular can be hard work – much like the database that powers your product, it can take weeks if not months to build and streamline. It’s worth doing though, as – to borrow a phrase from the world of computer science – ‘garbage in, garbage out’ certainly applies here.

For the marketing automation, you’ll need a bank of content that can be served to your future and present clients, following triggers from your CRM. If you have more than one type of customer, in multiple geographies, this can become even more complicated.


Perhaps the most important thing here is getting a solid time commitment from both the sales and marketing teams in order to collaboratively create emails and content that will resonate with your clients at each stage of the buying process.

Another big challenge is defining the key things for your CRM to capture, how that will be maintained to an appropriate standard (will your CRM automatically capture a lead becoming a customer?) and how the resulting data will trigger the right emails to be sent.

You should also set up a weekly meeting with key stakeholders to discuss live examples of records and ensure that data is being captured and recorded as you expect. If you can eventually get it down to one meeting a month, you’ll be doing well. Engage your sales operations people in understanding the customer issues at they key points in the process, and make sure you document what you’ve learned and what works.

This might sound like it flies in the face of the past-masters – here's Hubspot’s stance on this. If you read through the piece, however, you’ll see they agree that that marketing automation requires care and slow build up, and won’t instantly transform your marketing.

3. A/B test your way to the top

This modern day mantra would have you believe you ought to be running A/B tests every day until you become a FTSE company! But the reality is that with startups it’s often A vs B vs C vs … Z tests. Basically, it’s very hard to attribute conversions to just two significantly contributing factors.

Imagine you’re sending two emails and one (A) yields a click-through rate of 5%, while the other (B) yields 10%. Unless you’ve only tweaked one sentence or key feature between the two, can you really tell exactly which element worked better? At best, all you can be certain about is that the combination of copy and design choices in B, worked better than A.

If you’ve experienced this, it’s likely you’ve been running what are called ‘multivariate tests’. The complexities and challenges involved in such tests are well documented – the second part of this comprehensive article covers them in detail. It’s a real challenge to gather enough data points to align with all the variables and determine which performed best, although tools like this handy calculator can help you see if your volumes and changes were significant enough. It’s less of a problem in B2C, but when each sale is so important it’s difficult to build radically different versions of something you strongly believe needs to be a certain way.


Always record your results and conclusions and build from them. Combinations of ‘inconclusive’ tests can come together to give you a real gem of an insight that you might otherwise have missed. If nothing else, you’ll get better at knowing what sorts of tests work. Certainly don’t give up testing; over the past couple of years, it’s the two or three big insights that have come out the tests I’ve run that have formed the cornerstone of some of our marketing efforts. I’m very proud of that.

Ultimately, what I’m trying to do is, dispel the notion that just having lots of data will help you determine the way forward. As long as you capture data, you might be able to make more educated guesses and hone your marketing channels much faster than even five years ago. but expect set backs, and don’t expect overnight success.

Get as much help as your start-up can afford, keep key stakeholders within your team informed on your progress and predicaments, and keep abreast of new tools and tricks coming out (a few newsletter sign ups should do the job). If any of the above resonated with you, leave a comment and let’s discuss. Best of luck!

Author image
Kartik does online marketing stuff at Onfido. He plays with marketing technology, gets excited by data and comes up with weird logic regarding pool, football and every thing in between.