Monday, April 8, 2013
But as someone whose degree is in engineering who wrote code for 7 years before becoming an internet product guy, I must confess that I have never liked to look upon myself as a hacker of any sort. I know there are hackathons, and maybe I am just a little too old school and taking it personally. But I remember staying up late nights and busting to get a change done right instead of hacking something together. And I took great pride in considering myself NOT a hacker. So I am reluctant to accept this label today even though I have not written real code for a long time.
Personally I always always spoke of distribution. I always felt like Marketing as a title was frowned upon by some folks deep in tech. I remember a time when I met one of the iconic tech founders for the first time, introduced by another former hardcore tech CEO now turned venture capitalist. And both wrote plenty of code for sure. And at the time I had a Marketing title as opposed to some points in my career where I held a Product title. As soon as I was introduced, there were qualifications thrown out. "But he is not a Marketer, he worked for Boeing as an engineer." It confirmed what I had already felt previously. So I always tried to speak of distribution as I felt it did not have a stigma attached.
But I do like the growth label, as that is definitely what it is all about. So I wish the label were Growth Engineer. Perhaps it is not because folks without engineering degrees feel less comfortable calling themselves engineers? Not sure, would love to hear some feedback. I think there is some real process focus that is engineering-esque, but there is definitely some art as well. Growth Maestro or Growth Conductor might have been fun too, but perhaps it does not respect the tech enough? I stole the word Maestro applied to this game from Geoff Clapp, who I heard use it once for this.
Anyway, the boat left a while ago, and I love the direction it is sailing, but I had to go on the record just to get it off my chest. And for the small unrealistic hope that lives somewhere within me that hopes others might prefer to be Growth Engineers as well. hehehe
Tuesday, March 19, 2013
Books and other more permanent quintessential works
Predictably Irrational - Dan Ariely's book on decision making. Dan makes the case that buying decisions are not as rational on the micro level as economics on the macro level would have us believe, but it is at least predictable and can therefore be engineered to be effective on that basis. Dan is a very sharp guy.
Designed Addiction - this is a talk by Nir Eyal about his desire engine structure of looking at the world. It is rock solid.
UXD, About Face 3 - from our own local Alan Cooper, the freakin' man
UX testing, Gorilla Style - Steve Krug is classic and spot on
Designer Portfolio Sites - I like looking at the well reviewed portfolios for inspiration : )
Dave McClure's - Startup Metrics for Pirates - is a classic and solid growth structure
Sean Ellis' blog - He invented the term Growth Hacker and also has a company called Qualaroo (formely KISS insights) that is interesting. My favorite quote form him is Conversion = Desire - Friction
Andrew Chen's favorite articles - This guys articles are what led to the popularization of the growth hacker label. He is a solid product and distribution guy in his own right, legit.
Tom Eisenmann's best blog posts of 2012 - solid reading list, perhaps a bit verbose
11/25/2012 How to be a better manager, spouse, and parent, all in one11/23/2012 Peter Theil's startup class - Pretty solid stuff all the way around wrt to startup
11/09/2012 Kalzumeus.com article on AB Testing10/30/2012 http://blog.getvero.com/ - optimizing the email channel
10/29/2012 Quantifying Event Impact - ideas from around the web
05/25/2012 Why startups fail - data from Noam Wasserman based on thousands of startups over a decade
Monday, November 17, 2008
These guys are all about sales. They are especially good with copy. They sometimes distract onto esoteric things, but they are smart and know what they are doing selling online.
http://www.grokdotcom.com/ and particularly this article about copy http://www.grokdotcom.com/2007/10/29/copywriting-101/ or this one with links to free web resources http://www.grokdotcom.com/2008/11/13/33-free-tools-to-make-your-website-better/ or go to http://www.grokdotcom.com/archives/ and search for "top 10 online retailers by conversion rate" to get articles with monthly listings as well as some benchmarks.
Ralph Wilson (the owner, author and very good guy) gives very practical tips. Not overly sophisticated, but good advice and very practical because of that.
Jared Spool gives pretty good stuff out though mostly usability focused and not online Marketing per se
Jakob Nielson is the father of web usability. He has a lot of studies and things that are useful. He's a geek and his info is pretty good.
Another usability character involved with early Apple days has this list of bugs I like.
Random, some articles are good and some are less important, but there are a broad list of topics
This blog gives as good perspective of affiliate programs from a publishers perspective and I try to use it as my proxy to keep in tune with what publishers REALLY are thinking (I already know and accept that they are just using me, but it is mutual so I don't mind).
The classic Search site started by Danny Sullivan (but later sold I think)
Classic association of online marketers. The DMA for online.
Another search site from Jill Whalen who has been in the search space for a long time. http://www.highrankings.com/
Coupon Codes is the new site
Coupons.com has been very fortunate to get a lot of free press about clipping coupons and printable coupons. Here are some of my favorite links to various articles. Few people know that we are also behind Coupon Suzy as well as Safeway coupons and Kroger coupons. Regardless, here are some of my favorite stories about Coupons and coupons.com.
This is my favorite so far. Some poor frustrated newspaper guy from Pennsylvania (right near where I used to live actually), rips on Steven after his Today Show appearance, saying that Steven "doesn't look much older than my golf shoes." Hahahahahaha. Oh bummer, I don't think this is a good link any more.
I like this simple article running through different coupon websites. Nothing earth shattering, but original work from a local, which is cool to see.
There are a ton more, but many derive from these in some form. These are my faves for now I guess :-)
It's funny that a prevalent theme is that the online coupon traffic increase is a sign the economy is sucking. But if the economy was broadly sucking, wouldn't coupon usage be up 83% in other channels as well? Hitwise gets the closest to seeing what's up by at least noticing that Coupons.com growth is a LOT of the growth for this sector online, but they don't go so far as to notice that many of the other sites experiencing growth are partners "powered by" coupons.com like Coupon Suzy, the television face of coupons.com and the others in the space are perhaps in the coupons.com wake ???
Saturday, October 25, 2008
Internet Marketing Channels
- Display: CPM, CPC, CPA/Revenue/Profit Share
- Search: SEO, SEM
- Email: Paid, List Rental, Newsletters
- Affiliate: Internal and External programs
- Ad Networks
Display: Display mostly refers to purchasing banner ad space on third party internet sites. The site paying for the ad and resulting traffic is usually called the advertiser, and the site showing the ad to their visitors is the publisher. My rules of thumb are that well designed banners with compelling offers to reasonably qualified traffic should get about a 0.1% click-through rate (CTR). And well designed eCommerce sites with solid traffic (not junk) and a well-designed and optimized web site can see about a 5-10% conversion rate, with anything over 10% representing a pretty solid site and pretty solid traffic targeting. To me, there are three primary types of contract terms talked about for display deals. My diatribe on each follows with some commentary from both a publisher and advertiser perspective.
1) CPM - Cost per mil. This means an advertiser is paying a fixed amount for every 1,000 impressions of the ad on the publisher's site. Typical price ranges are from $1 to $10 but it varies widely depending on the demographics and intent of traffic. A wealthy demographic looking for information about helicopter tours in Hawaii might demand a very high CPM, but social media traffic that is not paying much attention to banners and is instead looking for ways to interact with their friends right now might demand a very low CPM. A $10 CPM or $10 effective CPM eCPM is considered pretty good, and anything higher than $10 is solid monetization. Realize that a $10 CPM is 1,000 cents / 1,000 impressions, so effectively that is a penny per impression.
Publishers on CPM deals: Normally, the ad impressions are recorded by both sides, on the publisher side they redirect the ad request so they can count requests and the advertisers count requests for the ad and track the referring url for the requests. Unfortuantely, unscrupulous advertisers take advantage of the ability to track visitors to publisher websites upon serving ads to them and they use the visit itself as behavioral targeting data which they then resell to other advertisers (acting as a publisher or simply a traffic data provider). If a deal sounds to good to be true, it probably is. Aside from behavioral retargeting, Publishers typically love CPM terms because it gives them guaranteed income regardless of the advertiser's capability to generate good banner ads or convert traffic. Also with CPM deals, publishers often have the flexibility to mix and match up their inventory as they can best sell it to reduce the amount of remnant or unused inventory they have and maximize the monetization of their traffic.
Advertisers on CPM deals: Advertisers typically dislike these terms because they feel like they can get burned if they are not paying attention. For instance, maybe other advertisers that paid higher CPM's get all of the good impressions from people likely to convert and you get stuck with the ones that are unlikely to ever convert like mobile phones, international traffic, traffic brought to the publisher site under false pretenses, unix or other useragents (browser and operating system types) which can not easily facilitate monetary transactions, or spiders, crawlers and bots which are not going to follow an ad tag link. Accordingly, advertisers will want to closely monitor the conversion of these ads, and have the ability to shut them off quickly before they have spent too much money if they are not performing well from an effective Cost per Action (eCPA) perspective. They may also want contracts that specify what times of day, exact site locations, browser types, etc the publishers can run impressions on. Brand advertisers whose products can not be easily purchased on the web (like frozen foods or pizza that one typically buys at the grocery store) may be less concerned about click through and conversion and happier with CPM deals than other types of advertisers.
2) CPC - cost per click. Each time a site visitor clicks on an ad, the advertiser must pay the publisher for the click. Becoming a de-facto standard as a compromise position for publishers and advertisers, largely due to google using CPC as their only method for search engine marketing and others following suit. Typical CPC costs are in the range of $0.25-$1.00, depending on the expected conversion rate and value of the conversion. Publishers willmontior effective CPM (eCPM) and advertisers will monitor effective CPA (eCPA) for CPC deals.
Publishers on CPC deals: Normally, the clicks are recorded by both sides using a redirect on the publisher side and referring url on the advertiser side. This is a compromise between CPM and CPA deals for all sides involved. The publisher needs to provide quality traffic for the ad impressions to get click-throughs, but it is incumbent on the advertiser to convert traffic directed to them. Publishers will monitor their eCPM (CPC price * average clicks per thousand impressions) for CPC deals because they want to ensure that they are monetizing their traffic as effectively as possible.
Advertisers on CPC deals: Advertisers are very concerned with conversion of CPC traffic. When the traffic arrives at their site, they have already paid for it, so any lost conversion is lost opportunity and turns the lead generation cost into sunk cost. Conversion should be carefully monitored from every source so that sources which convert well receive greater advertising expenditure while sources which do not convert at a profitable level are shut off. Unscrupulous publishers will try to find advertisers that do not carefully segment and measure conversion and marginal profitability so they can pile junk traffic to those sites. Advertisers also worry about click-fraud initiated both by the publisher and their agents and by the advertiser's competitors. There are many many well documented cases of people in low labor cost countries paid to erase all cookies then click on ads so that a publisher can bill for the clicks. Also, bots/http servers have been used plenty of times, even masking useragent values to make it difficult to tell it is a bot and increase CPC charges to advertisers. Finally, and especially for high cost high margin deals like helicopter tours in Hawaii and mortgages (historically) there are many cases documented where competitors click on your banner ads so that you incur the CPC charges but don't get any sales. There is further motivation to do this in SEM, but more on that later.
3) CPA/Revenue/Profit Share - Cost per Action. A publisher gets paid when they send traffic to an advertiser's site and that traffic performs some action on the advertiser's site, ie converts in some way. Variants include full revenue share where the publisher receives 10-90% of all the revenue received by the advertiser or similarly profit share. The publisher versus advertiser profit share rates depend on the margin of the product being sold by the advertiser and the leverage in the market of both the publisher and the advertiser. For eCommerce, CPA typically means purchasing something, and the action is recorded on the checkout page, often with the value of the transaction noted for revenue sharing calculations. But the 'A' can also be registering for a newsletter, installing software, and in the malware industry it can even mean installing a virus successfully on a machine and verifying that by having the virus phone home and demostrate control of the host machine. The CPA may be a fixed amount or it is frequently a percentage of the purchase amount on the other site. Typical CPA's are in the $2-$20 range, and vary widely depending on the value of the action to the advertiser and the conversion of traffic sent by the publisher. Viruses which turn host computers into zombies as parrt of bot networks garner about a $0.50 per machine payout.
Publishers on CPA deals: Publishers typically measure CPA by placing a cookie on a visitor's machine while they are on the publisher's site, then on the page on the advertiser's site after the action is complete, the Advertiser will place a call for a gif from the publisher's domain in the page so that the Publisher can read their domain cookie again and match it up to the referral. The gif is typically a 1x1 transparent pixel so the typical visitor does not even know that the publisher is monitoring their use of the site they were referred to. Typical CPA deals say that as long as the action is consummated within 7-30 days of the referral, the publisher will get their commission, so publishers get paid even if the action happens shortly after the initial click-through to the advertiser site. There are issues because this only works when cookies are accepted and able to persist. Historically this has been 95% of the time, but this is dropping as Firefox use is becoming more widespread, as Firefox has some nifty privacy plugins that allow cookies while you are on a site but erase them once you leave. Also, publishers will monitor their eCPM (average CPA payout * click-through rate * CPA conversion) for CPA deals because they want to ensure that they are monetizing their traffic as effectively as possible.
Advertisers on CPA deals: Advertisers generally like CPA deals as they can ensure that their advertising deals are marginally profitable and cash flow positive. The downsides are usually that publishers want to put a pixel on the conversion page of your site, and advertisers do not neccesarily want to share their conversion volume with every publisher whose site they run ads with. Also, since publishers may be paid up to as many as 30 days after the inital click-through to an advertiser site, it is possible that multiple publishers may try to claim a CPA fee for the same single conversion. Finally, the fact that someone converted on your site is behavioral data which third parties may use to try to target other offers at your site's traffic. Many of these problems can be avoided by building a 3rd party tracking system as I describe elsewhere on this blog, but most advertisers are either unaware or unwilling to pursue this option.
Which of the above methods are used predominantly seems to swing with economic conditions. When times are good and advertising budgets are flowing, publishers can hold out for CPM deals, but when times are tight and advertising budgets are scarce and highly accounted for, the few advertisers out there with budgets still flowing can demand CPA deals. There are exceptions, of course, for highly valuable traffic and sites that convert exceptionally well, but generally the ebb and flow goes with the economic times.
At some point, I need to do a post on ad servers, which is the tool typically used for all of these display advertising methods and more. These high volume servers are typically built for speed and very high availability, and include all of the tracking that one desires for any of the methods listed above and more.
Monday, September 22, 2008
Over the past decade, I have tested hundreds of different forms while working for US banks, and rule number 2 in the Grok post, "look at every question," is huge.
One of my colleagues, Rick Starbuck, once said, "the war is lost one little soldier at a time" as we argued with legal staff at a top ten bank to try to eliminate yet another low value input field on a new account opening form (not personally antagonistic, just legal arguing, I think legal peeps are great). We used to keep a scorecard of how many input fields we had, and how many each of our competitors had on their online application, with the goal always being to have the most streamlined application possible. A scorecard for arguments with legal might have also been useful too come to think of it. As a result of attention to these details, the sites I have been working with have almost always had the best conversion of any banking sites.
There is a LOT of upside in everything in forms. My checklist would include the following at least:
- Basic layout: How easy to follow can you make it? How clean? I have found that it is better to have one medium length page with vertical scrolling (only) rather than 2 short pages with a submit button between. But there is a limit. If legal is forcing inline terms accept and your one page is 10 pages of scrolling, don't mix that in with form fields.
- Call to action button position and design: This one is big. The best designs I saw after testing hundreds on a credit card app incorporated high contrast call to action color (red against white and maybe black edging), diagonal edges (think arrowhead), and bullseye patterns. Sounds obnoxious? You bet, but it works too. The challenge for me was getting a design that was not too personally embarrassing for me but still converted well.
- Which form fields are present, absent, required and optional: win the war one soldier at a time. Benchmark your competition. Be analytical about this. Get rid or Mr./Mrs./Ms., it is usually not legal and rarely adds enough value to justify it's existence. You want gender id, use a database match on first name or behavioral profiling.
- Terms acceptance methods: In page text vs pop-up, vs radio button, vs scroll boxes in page, default checked or blank, etc all have different conversion in different situations. My favorite terms accept is to put the terms in a pop-up box, and actually track to ensure that the box was opened with an image or some other html tag at the end of the terms text that can not fire unless the text was really presented. You need to do something with this solution to catch pop-up blockers and provide tips or a workaround as appropriate too. Sound overly sophisticated. Well, maybe, but if you have decent volume it is well worth it for the conversion gains.
- Form field labels: Do you call it Address 1/Address 2, Or Street Address/Line 2, SSN, Social Security Number, mobile phone/cell phone, etc. Test to get the right answer with your audience.
- Order that you present form fields in: Generally the questions should get more sensitive as you go lower on the page. This matches Marketing commitment principal. The only time you break this is when logical grouping require, but even then, put the group with the most sensitive stuff at the end (rank groups by their most sensitive field) in my experience.
- Instructional copy: What does it say at the top of the page? DO THIS! Why? You need one headline that reinforces the navigation or how someone got to this page, then some very short instructional text motivating form completion. You can tell them it is easy, you can tell them how great the payoff is, you can tell them there is no risk, you can tell them everyone is doing it, just test whatever you tell them and optimize it. Text is a great spot for segmentation too, by the way, because it is so easy to implement and so effective (great discrimination based on traffic source).
- Offer reinforcement: good practice to remember what brought them to your site and reinforce it all the way through. Did they come on a $10 cash back deal? Can you include a small reminder letting them know that you remember? Don't worry or stress, you will get your $10?
- Testimonials: test including testimonials so folks feel like others have trusted you and been happy about that decision. Trust in general is a broad topic with a lot of potential for you to play around with on any form. One rule of thumb that an 800 number on any shopping site prominently displayed at check-out will increase conversion even if no one ever calls the number (even if the number does not work, or is really for some unrealted entity like the US post office customer service ;-)
Tuesday, September 16, 2008
Internet Marketing Channels
- Display: CPM, CPC, CPA/Revenue/Profit Share
- Search: SEO, SEM
- Email: Paid, List Rental, Newsletters
- Affiliate: Internal and External programs
- Ad Networks
Search traffic goes to a search engine or uses a search box on some website syndicating search, types what they seek into a search box, and back comes a response from the search engine. In classic Marketing style, Google would be the largest search engine with the 60% market share, Yahoo second with the 30% piece, and Microsoft third with the 8%. These numbers are not exact and they change year to year, but they follow the model. Search traffic is special because search traffic represents surfers with an explicitly stated intent, which is very valuable to an advertiser selling a product or service online. What comes back in the search results are generally two types of links, which google created as a standard and termed results and sponsored links. The internet Marketing world refers to these two areas respectively as Search Engine Optimization (SEO) or Natural Search, and Search Engine Marketing (SEM) or paid search. Both are dynamic and vibrant fields. This article is about SEO, I will have to do another about SEM because each area is pretty big and pretty distinct except for quality score.
SEO is my number one question area. SEO is distinct from SEM (Search Engine Marketing) in that you don't have to pay for SEO. SEO is the the links that show up when you type words into google, yahoo, msn, or any other search engine. Not the paid links or "sponsored results", but the regular old search results. The spiders from various search engines will crawl your site, index it, and determine keywords for which you will show search results. There are a lot of details. I will give a quick high level, then you are on your own! I don't know everything, and everything I know is not here, but I will expand this post as I get time in the future.
For the rest of this article I will focus primarily on Google as my subject matter. Most folks in the industry will try to focus on Google search results due to the greater traffic volume, and usually if you are the number one search result in Google, you will be number one or at least very high up in the other search results for the same keyword(s).
First some definitions:
- Keywords - the word(s) you type into the search engine. Presumably the sites that come back have subject matter related to those keywords.
- Spider or crawler - the server and software at google that reads every page on the web that it can get to and determines what the important content of that page is as well as what that page links to and what links to that page. The spider feeds the program that builds the index. Normally the spider will go to a web page, digest the content of that page, then follows links on that page that it has not visited previously to discover new pages on the internet. Spiders also have scheduled checks they do to the same web sites to see if the content of the site has been updated. For sites that it finds frequently updated like blogs and newspapers, the spiders will return more frequently, while sites that are not as frequently updated will be visited less frequently.
- Index - like the index of a book, except for it points to addresses on the entire world wide web instead of just a page number. Google has built an index that knows all the content that it has seen and cross references keywords requested by a searcher to web addresses it has seen that contain those words.
- Spider trap - Some folks would like the spiders to think that they have millions of pages with content about everything in the world so their site always comes up first no matter what the search term. So they build things like dynamic urls that always link back only within their site, and dynamic content to make the spiders think they are seeing links to pages which they have never seen before and crawl them. This prevents the spiders from finding things on other people's web sites, so they are commonly called spider traps. The search engines try to avoid spider traps.
- PageRank - a metric named after Larry Page, one of the Google founders. THe higher the PageRank, the more authoritative the site is supposed to be. All else being equal, a site with a larger PageRank (like 9) will appear above another site with a lower PageRank (like 4). A PageRank of 0-2 is not uncommon for sites without much presence, many links, much traffic, etc. A PageRank of 5-6 is respectable and typical for medium to large companies with modest web programs. And anything aboove 7 is typically a larger company or someone with a very solid online Marketing and SEO program.
Keyword and Competitor Research: If you really want to work SEO, do some homework before you build out your homepage and site. To find the best keywords, use Google Adwords Keywords. This will give you an idea of both the search volume and the amount of competition out there for each of the search terms. Then go do a search for each of the terms you are interested in optimizing. Visit the top results for each term set and look at the Page rank for those top results. Look at the number of words they have on their page, the keyword density, the titles, urls and url structure, etc to see if you can do better than them on page design. Go to compete.com and use their search analysis to see what the main keywords are that are sending traffic to your competitors. See how concentrated they are on a few keywords vs a larger number of keywords sending traffic to get an idea of how sophisticated thir search program is. Perhaps install a google toolbar or Firefox seo add-on with the Page rank options turned on to see how sophisticated they are as well. If you are seeing Page Ranks that are 6 or better, and you are a small business trying to do this on your own, you are going to have a tough road ahead. Then monitor your site with the free Google Webmaster Tool Kit. Look for errors and flags, and make sure that google is indexing all that you hope they will.
Links to your site: For the keywords you are interested in, do a search for each high ranking results' url and see how many links (from an indexed page) there are to each of them, or use the seoBook Backlink Organizer. Can you get more links to you than the top results? Remember the first result gets the most traffic, positions 2-5 on the search restuls do ok, 6-10 get a little bit of traffic and by 11, which is on page 2 for most searchers, there is not much traffic left. Given that, do you want to target the most popular keywords which might have the most competition, or would you rather be the big fish in a smaller pond. This is an active decision to make and pursue. You can change it later but not quickly. You need to win on both page design and quantity of links to your site if you can.Page coding: Now that you have done your keyword research and developed a lot of content to target the keywords you think you have a shot at, it is time to code your pages so that spiders can easily digest your superior keyword focused content.
- Search engines look for bold letters near the top of the page and assume those keywords are titles, so make your page titles h1 headers big and bold, and use your search terms you are trying to optimize.
- Search engines like H1 headers near the beginning of the html load on your page.
- Search engines like good keyword density without spamming. Rule of thumb for me is roughly 300-400 words on the page with about a 20:1 keyword density.
- Search engines can not read text in images, so don't use it, put clear text on top of a background image.
- Tags: always put keyword dense alt tags on every image that you use, and of course use your search terms in the page titles, bookmark titles, and requisite meta tags, etc.
- Make sure your page titles are distinct throughout your site as well, so the search engine can discriminate between your pages.
- Realize that your home page will get the highest google Page rank, so you should optimize it especially if possible. If not possible, then optimize something that is linked to from your home page with no query strings before you optimize something that is two links or more away from your home page. Proximity counts for the algorithms. I have been told that each level away from the homepage drops your PageRank by 1.
URL's count big: As far as I can tell, if you want to rank well and you can swing it, have a domain with your search term in it. If you can, separate words with an underscore to make it easy as pie for a spider to digest your search terms. Even if you cant make this work in a domain, perhaps a subdomain, and at very least a directory and page name following this convention.
There are link share programs available which can get you a lot of links quickly. One of my friends owns http://www.linkhelpers.net/ and they have a pretty decent priced program which is pretty effective too.
If you want more search engine tips, check out real experts on the subject;
http://www.wordtracker.com/ (take the free trial here to do some basic keyword research)