Sometimes it takes a WTH! moment to realise how little you really know

So the team at McCormick AMS have been beavering away on our new marketing middleware platform – Chrysalis which we’re now playing with after receiving our prototype a fortnight ago. There’s some info on what it does over at

I’ve found in all of the marketing jobs that I’ve had, and teams that I’ve been involved with that people are different. There are many paths into a digital marketing career, and various different qualifications but once you’re in the job, the language of your day revolves around budgeting, negotiation with partners to run your ads, integrating them in some way, monitoring them, and generating reports which can let you plan out your days work.

Back in the “bunker” that is the second floor of the Digital Exchange building in Dublin, I joined an ambitious multilanguage, multicultural Japanese/Korean games publisher, GALA Networks Europe, which was a massive step up in career pace. Up until then, I’d worked in a variety of marketing, business and communications roles in Ireland across lots of types of companies.

But shifting from locally focused marketing, which revolved around dealing with ad and pr agencies, clients, local media, and direct buying media was one thing. Being thrown into a situation where there was quarter of a million euro a month being spent across a portfolio of products, in this case free to play Massively Multiplayer Online Role Playing Games (MMORPGs) which required you to register with the portal, download or play browser games, and then lodge credit on your account to spend on items in-game. This was before smartphones.

The company was very technical, but the marketing department was a purely functional one. It got budget, blew it, traffic came into the portal, people went on to register, activate, log in, play, like the product or drop out, come back and play more, and sometimes commit real money to their characters. But for the marketing specific links, we at least had basic click numbers by partner, registration and activation figures, our primary metrics at the time.

Everything was about conversion rates, and the company was very data driven. If you wanted to put forward a case, and had the data to back you, as well as a clear explanation of what specifically the problem was, and how to fix it, it got done. Maybe not quickly, but in the end.

Fast forward a year, and the new financial year had come along, and we had a big expansion planned for the following summer, so from April we stepped back our marketing to build up a battlechest. The product slipped from June to August, and we kept things ticking over, negotiating all the specialist media, and doing all the paperwork with our other campaigns, so all we had to give was the go dates. We’d announce to our members, do PR and specialist media, let that wave pass and then switch on the marketing budget across dozens of networks one by one once that started to die off to sustain a good first week.

We’d been over at Gamescom showing off the game to the media, and the marketing team were meeting a lot of the people we dealt with day to day over phone, skype and email. These generally were international advertising networks and local affiliate networks in markets where we had products.

Many were just to put a face to a name, because the relationship was going well as we could see clicks and registrations coming in, and our overall numbers. The other side was meeting with new advertising networks who wanted us to commit “test budget” of $10,000 minimum for double opt in CPA. On the first time we met them. As I’m sure a lot of people do, you go yeah yeah, and then at the next week you go back to the people you already know, and proceed as planned.

As we wanted a “big bang” first few days to fill up our servers with new players, and give existing players loads of new people to play with, trade, and raid, we decided that we’d break up our campaign budget into the different countries. France and Germany, the lead markets would go first, and we’d follow with the UK and rest of Europe after that the following week. We had limited manpower, and it worked for us.

So we picked one of the networks who’d delivered us plenty of volume in the past, when we had multiple ad networks running and google in the background. Seemed like a safe bet, when we logged into their platforms we could see loads of clicks and we had conversion pixels on the thank you for activating page so report back so their network could optimise correctly.

So we said right the expansion launches on Wednesday, there’ll be a ton of players patching up which will hammer our download servers for the first three or four days, but by Friday afternoon it’ll have calmed down enough that we can switch on the marketing. Existing players would spend straight away, and we had several million users, so they were always our priority.

We planned the campaign, a sizeable $20,000 campaign with the network in question, set a CPA of about $1.66 for an activated account and they went hell for leather promoting it. They delivered 12,000 activated registrations over the course of a weekend, and we were very happy with ourselves in the marketing department when we came in on the Monday morning.

Until the Producer walks across and says something along the lines of “You know the way you just dumped in loads of new registrations? We’ve been waiting here for a few days now, and haven’t seen any bumps in our new user numbers. We’ve had GM’s sitting at the spawn points and there isn’t any more usual people than normal going through. We’d expect to see three or four a minute at the pace of registrations we’re getting”. cue wtf moment number 1

But the screens from the ad network, and my own click tracker were telling a different story. Cue investigation that realised the marketing click tracker wasn’t tracking the same metric as the one the producers used, and had completely wrong data. So we fixed that from that point onwards, and luckily earlier in the year I’d asked the developers to record what game people went into after registering. You signed up for a universal account and logged into a particular game, and we’d wanted to see what “bleed” there was with people coming into different products,We hadn’t made a report to look at it yet, so we got the registrations from the campaign in question in a report and went through them. Took a few hours to crunch, so we had to wait overnight.

We could see that 12,000 registrations came in, about 11,000 of these activated, but only about 300 had actually triggered our “play” (going through the authentication server) which happened after they’d downloaded, installed, patched and logged into the game, usually about 2-3 hours later and noone had really gone past level three or four. cue wtf moment number 2

So we’d been fed a complete dud of a campaign. I spoke by phone to the network, and eventually found out that what they’d done was put our campaign on some offers on facebook, told the people there to click to our page, fill in their name, email address and activate their account, and come back fifteen minutes later and the game would give them 100 bongo points or whatever.

So what they’d promised to my face in Cologne was they’re one of the best games ad nets who have such massive volume and drive players (from which a percentage monetise) , but when it came down to it they cut to the chase and said something along the lines of “tough, the insertion order specifies that you are paying $1.66 for a double opt in registration. And you didn’t specify that we couldn’t use incentivised traffic”. wtf moment number 3

We counter argued that in our all affiliate campaign t&c’s which were the same for ad nets, it was strictly prohibited, and we’d discussed it in person a few weeks previously. But again, it came back to it not being in the insertion order. We had a stalemate situation. And we never spent another penny with them again.

While this was unfolding over the course of a week, I asked our developers if they could run the same report for everyone we had stats for, as far back as the data went – about 6 months. It was a big job, so they left it overnight to crunch away.

The next day I came in, and chopped up the data into excel, and started to generate some pivot tables. Once I had them configured, something jumped out. We were working with about 30 different ad nets at the time. Some of them, about 6, which we’d been spending five figures a month with up until them were shocking. I think the worst had only 3% of their registrations playing and only up to about level five, no matter what product we looked at. And we were promoting six products in five markets, across dozens of affiliates. wtf moment 4

Immediately, we rang them to stop the campaign, went into the reports, cancelled every single outstanding lead in the chargeback period by importing a cancellation report (we pre-paid and topped up our account with this network), and had another very angry phone call. Unlike the other situation, this clearly was in breach of the terms of the affiliate programs, and this particular network was managing which affiliates were on the campaign, as we didn’t have the manpower. The CPA triggered though, but the behaviour after was wrong.

So this time, we were able to go to them with a very clear problem, a reason for the cancellation, and asking them what they were going to do about it. In the end here, they agreed to a $50,000 refund, and we stopped working with them.

We repeated the conversation with another half a dozen networks or affiliate programs on smaller but still considerable scales, which in our marketing budget meant that liabilities we thought we had weren’t there any more, and we could redistribute that budget.

There were some clear problems, and we dealt with them first. Then we went through the other ones, where things were mixed. In many cases they were grand overall, but cases where say all the behaviour looked fine in two market programs, but there were problems from individual affiliates. So we could cancel these leads, and boot them off the program, and more pro-active networks promised to blacklist them from the network. If they were doing it to us, they were probably doing it to others. And they had well deserved, and hard won reputations which they stood over.

This helped reduce the burn rates, and actually dropped our conversion rates one by one, as the artificially high conversion rates were masking the actual performance of the campaign.

We then decided on tackling the behemoth that was a €100,000 monthly adwords camapign. We’d had to run this report separately to the rest, and I remember this took three days to generate. We had to buy a souped up gamer PC to get Excel to run to open it.

Side by side, I sat down with our Search Engine Manager, who we’d recruited from Google. He was all about the data, and we took any emotion out of the situation. We’d always managed Google ourselves, so we didn’t think there was any fraud or anything like that., but there were multiple accounts with products for their markets, and each of them had loads of adgroups, and in them loads of keywords, each which had to be priced towards a CPC or CPA target.

Luckily, and again having been trained in-house at Google, was incredibly data driven and had been asking for months for the same sort of insight to do his job. He’d immediately put in adwords value tracking tags into our campaigns, and because we’d been storing affiliate information since earlier in the year, we got this too going back seven or eight months. We chopped the data, using text to column, and had a nice hierarchical structure for the data.

We went through each campaign, switching off any keywords where we’d had enough clicks (and associated costs) to make a call, and either zero or few registrations. This culled €40,000 of the €100,000 budget overnight, which immediately flowed into the ads left online. But he now had the data he’d been craving for since day one, and he could see cause and effect himself, and work away.

This was back in 2010.

We got this done by the end of the year, so going into 2011 we had everyone working properly. I think we spent about €2.500,000 that year on marketing, and directly got back €900k from our marketed leads. But by December, we had everything cleaned up, and were able to react based on real agreed internal metrics. The next year, we spent €2m and made €2m, and the year after we spent €1.5m and made €3m from direct marketing activities, turning a cost of business into a profit centre that a significant section of the business.

So on founding McCormick AMS last summer, and closing down my video games publisher Scraggly Dog Games, and wrapping up my research into the Irish games industry, I set out to design a piece of middleware for marketing managers like me the day I joined every company I’ve worked at, who are facing all the problems I had, or know they have a problem somewhere, and want a way to see exactly how what’s causing what, and how it can be fixed.

I got my prototype a fortnight ago, and am going to be working for the next few months to build. I learned from the lads in Popcap Dublin that when they were making Plants vs Zombies mobile, they had the game ready in six months, and spent a year polishing it until it’s perfect. Pretty much everyone in my company is from the games industry, but we’re not making a game, we’re making a sophisticated piece of software for marketing managers controlling large budgets, who want to know what came from where, and where to put their money. Plus a load of other things, but I’ll talk about them later.

But for the moment, we are looking to identify some companies who’ve found themselves in a similar case to those outlined above. If you’re looking for a way out of the problem, and would like to conduct a 4-6 week audit of some of your traffic, and are technical enough to generate reports for us, then we want to hear from you. If CPC and CPA are in your budgets, than we think we can help!