A Retrospective of Working in PPC For 20 Years
On December 4, 1998, I hung up the phone with a representative from GoTo.com and was officially a PPC Marketer. Over the course of 20 years, PPC has changed significantly, and this is a reflection of 20 years of PPC.
The Bidding Years
In the early days of PPC, there was initially GoTo.com, founded as an Idea Labs company by Bill Gross. Every single bid could be seen. Ad position was only determined by bid, and there wasn’t a penalty for not getting clicks.
‘PPC Skills’ meant you understood keyword research, a little bit of ad writing, and be pretty good at math. Those were the three primary skills a PPC marketer needed.
This led to two types of advertisers. The first group was those who were using a CPM model payment, as they bought CPM elsewhere. These people realized that you could buy any non-relevant keyword and it would not get clicked; but you would get the ad impressions. They viewed PPC as a way to get low cost exposure; but of course, hurt GoTo’s revenue as they didn’t want clicks. Then there were the performance advertisers. They would watch bids and make adjustments as necessary.
Now, performance advertising was different then – it meant you measured clicks and total sales from your CRM system. There was no conversion tracking. A few years later, companies popped up like Conversion Ruler, which was a 3rd party conversion measurement system. This would change in 3-5 years when conversion tracking became a part of the PPC system’s offerings. However, in the early days – there was very little conversion measurement.
In the early days, the ads were decent; but there wasn’t a good way to test them, so it was really about the bidding. You only paid a penny more than the advertiser below you was bidding (you could not set your bid the same as someone else). This led to bidding types such as gap surfing, bid jamming and so forth, and a lot of manual work day in and day out.
PPC went through its first main change when GoTo.com release its first API. This meant that someone could build a bid management platform. GoToast was the first of many bid management platforms to arise. It could initially change bids 12 times a day, then 24, and then 48 – the number of times set by the API rules. Bids were only changed on the 5 minute mark. This led to ‘bid charting days’. Once a month, we’d pull a 24 hour shift (we were much younger then and working in a brand new industry – there were no rules) and refresh the bids for all our primary keywords and make a notation of when our competitors had setup a bid change. Once this was mapped out, then we’d set rules to change our bids 5 minutes after our competitor’s did, so we could easily claim the most advantageous spots.
This type of dedication led to one of my favorite client interactions. One of our advertisers called me up very confused. He said that his primary competitor had just called him and had asked him to lower his bid as we were causing him to pay to much per click. I explained these were the 3 bids:
- $99 (competitor)
- $98.99 (us)
- $5.83 (next competitor)
That meant his competitor was paying $99 per click and we were paying $5.84. Did he really want us to lower that bid? After he finally stopped laughing, he said to just keep on doing whatever magic it was we were doing and rarely ever called again.
The first 3 years of my PPC life, I was an affiliate. I had learned to build websites a few years back as a hobby, and initially was heavily involved in SEO. As an affiliate, you’re looking for cheap, good quality traffic, from anywhere. PPC was no different. A couple years after GoTo.com launch, there were a flurry of other PPC engines from FindWhat, LookSmart, Excite, and many more. It was not uncommon to be managing PPC across 7-12 different platforms.
With so many platforms, and the way SEO worked those days, I had several contractors who were doing manual work for me. Most of their work was either setting bids for the smaller engines, submitting URLs to search engines, and some pretty boring manual work.
Then Windows 2000 launched with a feature that would significantly change how I worked. Macros. Windows 2000 natively supported macros. That meant that I could program manual actions to be automatically carried out by a computer. Now, I didn’t have any type of a programming background; so I sat down for a week and taught myself macros. Once I was comfortable with them, I realized that just having 2-3 computers on my desk would be more efficient than paying other people to do manual work. I replaced almost all the contractors with macros. Remember, repeated inputs can be replaced by a machine; thought strategy cannot be easily replaced – this is still true today, we just say automation and machine learning instead of macros.
My First Agency
After doing quite well at PPC for a few years, I had companies trying to track me down and ask for help. I lived in nowhere Pennsylvania (I had graduated from Penn State a few years back and still lived in the area). I wasn’t advertising any services. I didn’t even have a website for myself outside of a small blog about PPC hosted on a free Angelfire or Excite, I forget which, site. It wasn’t even called a blog yet – they were called online diaries. However, I was doing well enough as an affiliate that both competitors of the people I was affiliates for, and the merchants I was promoting, started asking for help.
In 2002, I formed my first agency, iDjinni Consulting. I didn’t have a business background or an advertising background; so I was just making up my own rules of work, client engagement, reports, and so forth. It was a fun time to just learn a new skill set.
One of my first large clients didn’t have a good sense of how revenue flowed from marketing to their organizations. They made a lot of money, but had no idea how to accurately make good marketing decisions. This led me to make my first foray into software building.
My First Software: GeoTracker
My large client’s budget looked like this (rough numbers as it’s been a long time since I’ve looked at this) on an annual basis:
- Yellow Page Buy: $5 Million
- Radio: $1 Million
- Print media: $2 Million
- PPC: $150k
- Online Yellow Page Buys: $100k of actual spend, and $750k of free spend
- The Yellow Pages didn’t know how to monetize a website, so you just got 10-20% of your print buys as free online spend
- In fact, we spend so much with the Yellow Pages, we created a CMR (old yellow page terminology) just to get the discounts across all the different print editions
They had more than a dozen locations across multiple states. All transactions were handed in store. When you have all these different buys and you are trying to track in-store revenue; it’s difficult to understand how your spend is working.
I created software (I didn’t do the programming, just the business, mapping, and rule set for how it was to be programmed) that mapped the distribution of Yellow Page books, newspapers, radio reach, online, etc in venn and chart diagrams by zip code. Then the in-store revenue was mapped by zip code locations of the actual payments. Finally, you could see how you changed your marketing inputs (i.e. buy a radio ad, put a weekly ad in a newspaper, etc) and compare that to the zip code revenue by location over time. This gave businesses an actual way to track and test marketing by channel across online and offline inputs.
I had no idea how useful this software was. With little exposure to the marketing world this just seemed like a logical thing to create, and was ultimately acquired by one of our clients. A few years later when explaining this software to some other companies, they were amazed at how useful this was and would have bought this for a lot of money.
Another learning lesson in the business world. Everyone has ideas. Not many make them a reality. When you do, make sure you get the proper value out of what you created.
The Rise of AdWords & eWhisper
In 2002, Google launched the first version of AdWords, a CPM buy. We were an early buyer as it was just an insertion order for CPM. Later that year, Google launched AdWords Select, the self-serve PPC business, which was later renamed to AdWords.
When AdWords launched, most people hated it. You couldn’t see other people’s bids. It has a different structure than Overture (GoTo.com was renamed to Overture and then acquired by Yahoo). Ad position was determined by CTR times bid (Quality Score didn’t exist yet). Within a year after launch, if you didn’t maintain a minimum CTR, your keywords would go offline.
With one fell swoop, Google got rid of all the people buying PPC on a CPM basis. If you didn’t get clicks – keywords were disabled. There was suddenly an emphasis on the ads – you had to maintain a good CTR. This was the age of initial Ad Creation wisdom for PPC.
It looks a couple years before everyone really like AdWords, and the main reason I got so good with it is that I would experiment and ask a lot of questions.
I had joined WebmasterWorld, a wonderful marketing forum, in 2000; but was pretty quiet on it, preferring to lurk and just read about SEO tests as I was still heavily involved in SEO. With the rise of a complex system like AdWords, I knew I was going to have to push for some answers. So I made eWhisper as my nickname, which has nothing to do with ‘e’ being electronic like email and so forth. I was a creative writer once upon a time (I helped created one of the Penn State Literary Magazines) and it was based upon a song I had written and later turned into a published poem called Echoing Whispers.
As I started asking questions on WebmasterWorld, initially Matt Cutts (who was still anonymous on the forum using a nickname at that point in time) would find out answers for me. Then AdWordsAdvisor joined the forum, a wonderful person who likes to remain anonymous, but worked at Google. He would find out amazing details for me. I was experimenting so much, and pushing the boundaries of AdWords, that Google would often reach out to me for feedback and product questions.
A good lesson learned: If you are asking intelligent questions, and experimenting yourself (i.e. not being lazy), people are more than happy to answer questions or help out. As an introvert, having the anonymity of a computer screen let me interact with a lot of marketers and make a lot of connections I would not have normally made.
I had been a marketer for 6 years, and was still just making up my own rules. I really wanted to work for an agency just to learn the ropes of how an agency works. I had moved to Chicago and knew a couple people who worked there from WebmasterWorld. I went in for an interview with LocalLaunch just to see what it was like, and ended up chatting with Werty (George) and Chicago (Justin Sanger) from WMW for 6 hours. That might have been one of the longest interviews in my life.
I didn’t want a job yet. I had just moved to a huge city and really wanted to explore it. However, I got a job offer the next day. I told them I wanted a few weeks to enjoy the city and then I’d probably accept it if they could wait. The next week I got a call from them asking if I could just sit in a meeting with a prospective client.
It turns out that client was Interland (now Web.com). A huge massive company and I caught the jist of the pitch (which was software – not something I was prepared for) fairly quickly, and ended up being a much larger part of that conversation than I would have ever guessed.
Being someone who wasn’t even an employee yet, and sitting in a meeting with high level executives, started me on another path that I would continue and ultimately adopt: Ignore titles. If a company has a strict hierarchy they adhere to (you can only talk to your boss, don’t jump the ladder), I wanted no part of it. Great ideas are great ideas no matter the position of the person. I would go on to often be the lowest level person in the room and not care about the positions and just say what I had to say. As long as you are moving a conversation forward, contributing idea (not just wasting everyone’s time); then I wanted to hear from anyone and I would talk to anyone regardless of their position. I can’t count the absolutely amazing ideas I’ve heard from low level to high level employees. Positions do not matter to me, nor do titles.
In fact, later I would adopt a title ‘Director of Search’ as it didn’t mean anything. The entire company was search, so being a director of it meant I could be in just about any meeting where I could contribute regardless of the level of employees in the meeting.
I did join the company; they were doing innovative things. These days we hear about NAP (name, address, phone number) consistency for local businesses and have data submission services. You can thank LocalLaunch for inventing that world.
What I really learned at the company was scale. We (and we being a huge collective of very smart people) built a platform that would build and manage data at scale from local business data, websites, SEO, and PPC for companies. Before I left we were managing 660,000 business records, 110,000+ websites, 42,000+ PPC accounts. I got heavily involved in the product side and for a few years, didn’t manage PPC (I consulted on it internally), and learned a ton from Aaron (a wonderful boss) about how to build and manage huge amounts of data and companies at once.
This is another lesson. Managing a single account can be done however you want it to be done. Managing thousands of accounts means good processes and consistency. Processes and consistency come from being able to diagram out workflows and decision trees. To this day, if I want to think through a problem and how to manage something – I create a diagram or flowchart. A flow chart forces you to think about every decision that can be made and how you would handle it. That runs from data flow to decisions. Flow charts are amazing, they make your thinking very clear and shareable with others.
When you are managing this much inventory, your day becomes hectic filled with emails and meetings and it’s easy to become overwhelmed by the sheer amount of things going on. I went on vacation for a couple weeks and came back to more than 3,000 new emails. That’s when I went on my productivity crazy binge and created meeting rules, email rules, and put many things in place to guard time while ensuring everyone can get their jobs done. I still stick to all those rules today.
However, I got a good piece of advice from one of our SVPs during this time: Try to be lazy at work. You’ll never accomplish it, but by trying to be lazy, you learn how to delegate work, be efficient with your time, and ultimately accomplish more by trying to be lazy.
From a PPC perspective the years of 2004-2008 were about ads and scale. The rise of resellers (Yodle, Reach Local, LocalLaunch, etc) all happened during this time frame. Google’s changing of the ad rank formula (and introducing quality score) forced people to think about the ads. The money that flowed into PPC was huge (hence parties by the search engines would consist of things like renting private yachts or even all of Great America Theme Park); and with all of this money, you had to know how to scale keywords, ad groups, campaigns, etc to these new companies who were suddenly spending over a million a month on PPC and really cared about the brand message.
We sold LocalLaunch in 2008; and within a couple years, I had moved on to doing a lot of PPC training – and Larry Page had a lot to say about that job tenure.
Larry Page’s Influence
I had been lucky enough to meet Larry Page, Sergey Brin, Marissa Mayer, Cheryl Sandberg, and more on multiple occasions. That’s what happens when you spend a ton of money with Google and get invited to private events. I managed to have lunches with Al Gore, Queen Noor, and others at Google events. This was when I learned how Larry believes in technology more than relationships.
Back in 2004 through 2010, Google’s support system was amazing. Even middle sized advertisers could have excellent reps. Google would send advertisers swag (from scooters to fooz ball tables to refrigerators), and they put on a lot of educational events. It was in 2006 that we signed a contract with Google to help them do AdWords Seminars for Success and start doing in-person trainings. Talking to many advertisers from all walks of life was quite an interesting time and that you can learn something from anyone.
Never dismiss someone based upon their spend or business size. There would be someone in the audience spending only $1,000 a month doing interesting attribution and testing and someone else spending over a million who didn’t know match types existed. You often learn from those who have learned how to stretch their dollar over those who don’t care about any individual dollar but the entire collective.
That was Eric Schmidt’s tenure as CEO (technically, Larry didn’t become CEO until 2011). When Larry Page became CEO, Google’s support suffered. They cancelled most of their in-person events. They restructured their support system yet again. As a result, it became in-vogue to ‘hate the borg’. I don’t know if Larry was directly responsible for this shift or if it was someone else, but under his leadership, Google’s profits grew while advertiser’s frustrations grew as well.
When Facebook, LinkedIN, Microsoft, etc all launched platforms over the next few years, advertisers were quick to try them out. I think if they were happier with Google, we’d have seen slower adoption of some of these other platforms.
Form a company perspective, there are some interesting takeaways:
- Scaling with software over people is how Google grew
- If you have eyeballs, people will spend money with you no matter how you treat them
- If you cut support, your brand might grow with consumers, but your direct client relationships will suffer
- This support leaves you open to money being moved to areas that might not be as productive (other platforms) because people like to work with those that will help them
I once had a client in the insurance industry who gave me this advice:
- They must like me
- They must trust me
- They must think I know what I’m doing
If I can get 2 out of 3; I’ll get the sale. It’s good advice to think about in your positioning. You can’t be everything to everyone – so what do you want to be known for?
Nothing lasts forever, and in 2015, Larry Page stepped down as CEO. That’s about the time when Google’s support got better, we saw more education coming out of Google, and Google started really thinking about how they were viewed. This could be Larry Page stepping down or FaceBook’s revenue exploding and Google knowing it has to do something.
The Animals Attack
While I’ve been known for PPC more than SEO; I was still highly involved in SEO over the years and worked with many well known brands on their SEO strategy and execution. Then the animals attacked. Panda. Penguin Pigeon. These SEO updates were annoying. I felt that Google SEO was about penalties as much as it was good content.
I think Google’s organic results got much worse as their entity matching system cannot figure out a good long tail search result. Google is good at the basic queries, but when you are trying to find something specific – their system is just wrong on many occasions and I often find myself wanting to just go back to boolean search operations.
This is when I decided that PPC and SEO had become so complicated that you had to just pick one and focus on one; so I gave up SEO as I like PPC a lot more since you can really be creative in how you execute strategies.
I also think that during this time; Google started messing with machine learning in a bad way to try and get users to adopt their new technologies; and that it made a lot of Google products worse. For instance, before there was Google Assistant, there was Google Now. It would show you your commute times, if there was more traffic than normal and you had to leave early, your flight schedule, your next meeting time, interesting articles, and so forth on a single easy to use screen. It was so good, it was the only widget many of us had on our phones. For 4 years, this was my digital assistant and it was amazing.
Then Google decided to let their people mess with it and it became worse. These days Google Now would rather show me an article about a topic I’ve said I’m not interested in 100 times than remind me that I have a flight in 2 hours. Google Assistant shows me scores for my football teams about 25% of the time. It shows me my flights about 5% of the time. It’s so inconsistent that it can’t be trusted. When a company has a great product and it gets worse, that is not a good trend. Google has done that several times; and it’s a trend to really watch with their machine learning today. Can Google be consistent with their machine learning; or will they muck up something that is working?
Rise of the Machines & People Marketing
Over the years, many automation platforms have been built. They started with bidding, then moved to cross platform reporting, then to multiple inventory management, and finally to recommendation engines.
It turns out that as much as machines can replace humans – they also can’t replace them. Machines can replace humans doing redundant work. They can’t replace a human’s ability to think, empathize, provide data context, and create marketing strategy.
As machines were getting more powerful and more and more people talked about automation, audience marketing also grew in popularity – how to talk to individual people at scale. These two concepts are in contradiction with each other. One is letting the machines take over marketing and while machines can do a lot – they don’t understand people. Audience marketing is all about diagnosing behaviors and characteristics (which a machine can do) and then talking to each group individually (which is the marketer’s job).
Since these two things have risen in popularity at the same time, it’s made people question the machines more than if the machine learning had gone through its revolution in the 2000s before the rise of audience targeting.
This is a good thing. Machines make decisions based upon data. Machines don’t know the context of that data. For instance, there’s a direct correlation between beef consumption in the US and how often lighting strikes the ground. The inference being that eating beef causes lighting or if lighting hits the ground more; we decide to eat more beef. Both of those correlations are ridiculous; but the data is very well correlated regardless of it being a coincidence or a direct cause and effect. This is where the human ‘common sense’ approach often trumps the machine – we have context for data.
While machines can automate a lot of things; there are things that should be suggested and not automated. For instance, I looked at 4 advertiser accounts and said what if they would have automated negative keywords being added to the account based upon their target CPA and ROAS targets. In every case, over the course of 18 months, all 4 would have had roughly 70% of their top keywords, and all their brand terms, turned into negative keywords. This was due to a delay in conversion tracking, broken conversion tracking, a landing page test gone wrong, a strange seasonal period, and so forth.
This is why humans and machines are a better pairing together rather than just one or the other leading the way. The machine can suggest things like negative keywords and then the human provides the context if that is a good suggestion and apply it or a bad one and ignore it. That’s the basis of how we built Adalysis, my current software company that automates recommendations and then gives the humans tools to easily make changes at scale.
While machines will continue to rise in popularity; the human context is incredibly important to your marketing.
There are a few reasons I love PPC.
It is always changing. What you are doing today, you won’t be doing the same way next year.
The various ways in which people make money. I’ve been amazed at how many different business models I’ve seen over the years; and new ones I still continue to see by interacting with different companies.
It’s at the intersection of creativity and technology. I’m a geek. I love my xbox, computer, and most things tech. I enjoy being creative. I can’t count the number of fictional stories and other works I’ve written. Being a creative storyteller is in my blood.
PPC is where math, technology, and creativity all meet and intertwine with each other. It’s one of the few careers where you can get dirty with Excel and new math formulas and then engage the other side of your brain with creative short copy ads or long copy landing pages.
Then you engage the math to see which creative efforts were better – you can test your creativity; and then think about why users engaged with your content that way.
My college major was Personality Theory Psychology. Why people think and act in certain ways; and advertising might be one of the largest social experiments in the world on looking at how individuals think and act in highly different manners and the causes of those reactions.
In the next few years, we’re going to see humans and machines learn how to work together. It won’t be humans vs the machine. It will be humans working with the machine. If you love technology and creativity, then you’re in for a fun ride over the next 20 years as were go through yet another evolution in PPC driven by brining humans and machines together to create wonderful advertising strategies.