Searchlove Conference Live Blog – Day 2
by Pete Wailes on October 30, 2012
Session 1 – David Mihm: The Need to Know of Local Seo
30% of all searches are local on desktop, 50% on mobile.
Mobile SEO used to look much like anything else – 10 blue links. Has been changing since about 09. We’re now seeing author markup, user reviews, local data etc.
Venice blew it open – before about 31% was blended, now it’s way past 60%.
Ranking factors include business name, physical location, customer reviews, references and anchor text, citations etc…
Google must be able to associate you with a real thing, Tie your business to your +local page. Tile should include name, address and phone number.
Upload a KML sitemap to give Google your various locations.
Can use rich snippets from schema.org to mark up your locations.
Doesn’t boost rankings per se, but enhances listings.
Multi-location – have a store page for each store, and submit each one.
…he’s going fast, I’ll add to this once I’ve got the slide deck
Don’t need a lot of links to compete in local, because most small businesses aren’t SEOs, the competition is low.
Branded anchor text seems to be more useful than commercial.
Concept of highest score of a referring document. Means that a site needs at least one really, really good link pointing at you. .gov, high DA sites etc.
Quantity of links from locally relevant domains (things Google can associate as authorities, and local – local blogs, colleges/unis etc)
Name, phone number and address being cited is vital for local rankings – having people actually talking about you.
There’s a local search ecosystem for most countries – the places that send data to local search data providers. For the UK – Companies house, 118, Post Office, Thomson local, Yell, Yelp, Touch Local, Yelp etc…
Niche citations are huge for small business. Mine existing links, job listings etc. Get the name, phone number and address in there too.
Events work well too.
Clean-up – search for your business on GMaps. Look for duplicates, and click report a problem. Take care of your maps listings.
Also you can go to Mapmaker (if you’re in the US, coming to UK soon). Then you can make those edits yourself.
Beware, any user can add a location, so you might get shady competitors spamming you with it.
http://goo.gl/1Dbpg : places where Mapmaker is available.
Look for out of date names, addresses and phone numbers to keep all your data tidy.
Focus in reviews – increasingly important. Reviews on Plus Local, and everywhere else. Not enough to just go after one platform.
Keywords in reviews are important for long and short tail. They’re also looking at sentiment, so bad reviews may hurt.
They’re also tying reviews and users together to create author strength for users and their individual reviews – not all reviews or reviewers are equal.
Really get reviews. Take feedback on board to improve your business.
Google can now understand sub-parts of reviews to evaluate different components of a review. For a restaurant that might be service, food etc.
Anywhere where your business location appears will be considered as a citation.
Citations + sentiment drives the tags that Google grabs for your business.
Send people with gmail addresses to Plus Local, spread everyone else around other places. Qype is good, because anyone can sign in with at least one service. Yahoo Local is good and easy too.
Get feedback incorporated into your business process.
Don’t direct people to Yelp – you’ll get blacklisted. It’s against their ToS.
Use topsy to find power reviews, send them cool stuff, they’ll review it on their site, rinse & repeat. Followerwonk works nicely too, as do Google custom searches.
Find people who live in your area, host meetups etc. Build relationships. Yelp & G+ are more important than other places going forward. Again, make sure they’re active users, who review a number of places a year. Power users are more useful.
Categories are decreasing in value. New interface in the US, probably coming to the rest of the world soon.
Map results can get pulled in to the main search, so pay attention to how you rank there – it’ll be how you rank in maps results in blended search.
To find searches, look at Ubersuggest, Google related searches, Trends etc. Adwords tool may not have enough volume.
Use pure GEO as a proxy. Do non-localised terms for generic volumes, then localise for specific local volume.
Use mobile as a proxy too – you won’t get exact volume, but it’s useful.
Look at form completions, what do people talk about.
Tracking phone numbers are a PITA for local. Messes up your phone number footprint.
Use GA to analyse where people came from, what sources drove them there and so on. Mine that data.
Look at keywords from mobile to see probable local phrases.
Rank tracking is basically pointless for local – way too focused on searcher location.
Overcoming centroid bias – distance from the center of a place isn’t as important. Distance from the searcher location however is vital. As such where you’re located is going to influence your local results.
Volume of reviews that’s far higher than your competitors may well effect a good ranking in local.
Backlink analysis to find local-based links.
Local SEO, traditional SEO, social – the trifecta of local SEO.
The more data Google has to play with, the further they’ll go with showing more limited data from a smaller group of sites. The smaller the number of results, the more certain Google are the those are the ones that a searcher really wants to find – becomes harder to displace those.
Look at piggybacking on really high DA sites to get those listed, if your site is new. You can displace it with your company site later.
Local search may well blend with product search in the future.
Session 2 – Richard Baxter: How We Build Links at SeoGadget
SEO is constantly evolving – it’s exciting.
SEO is becoming more and more like real marketing.
Guest posting is not marketing. List posts are not marketing. Infographics are not marketing. Blog comment spam is not marketing.
We’re all guilty of this – we’ve all done rubbish stuff in the past.
You can optimise everything for links.
This feeds in to everything – site design, email, guides, blogging, infographics, careers, video, cheat sheets etc… Every asset.
This is link earning, not link building
Engaging with your community and brand advocates – that’s link earning. Talking to your customers and educating them – that’s link earning.
Do RCS. Grow the email list, work with charities, help people out, do stuff properly and push for it to be amazing. Work with amazing people, hire amazing people. Make everything amazing.
Ask people what you do, then find where they get it wrong, and address those things for new people, so they immediately get what you do and how you do it.
Shout about your patterns and how great they are – they’ll be inclined to link back to you.
Open source the technology that you build. Builds relationships and credibility and means that people who are important talk about you.
Target the audience that you want to talk to. Don’t think about link building. Think about audience targeting and trust and mention earning.
Stop marketing to people like yourself. Don’t create personas based on ephemeral concepts – create them based on actual people. Find someone who’s like the persona you want to model, and work from that. Real data > fake data.
Once you’ve worked out who you want to target, use TweetArchivist to find what people have talked about.
Mine social data to find out what your link targets share. Once you know what they’re all looking at, you can get in those places.
Drill into the people that the people you want to target listen to. You’re really targeting those people. Because they can put you in front of your final end point targets.
Use followerwonk to find people who are in the right place, who are socially active, who are relevant to the interest that you’re trying to target.
When you know what the people who influence your targets are doing, you can go to the places where those targets find stuff that they share, and get on those places. It then gets picked up by the influencers, and they push it out to your final targets.
This gives you places to outreach to, which means that you’re getting a link (which is nice), but also getting in front of your outreach targets. Means that you get your brand in those places, and you then get in front of the people you want.
Targeted outreach: who to target, who influences, what do they share, be there.
Rinse and repeat.
You can geo-target this, by taking data published by local places or local news stories (think broad with what constitutes news for this).
You can then push out that content to high influencing people in particular locations, and then it’ll be pulled out by people sharing that story, all linking back to the site with your location being featured, because the story is about the location.
After the storm has died down, then do link reclamation – find the people who mentioned your story/news/article/infographic etc. Then contact those people, and ask if they’d mind linking back. Publish under CC, so that you have a reason to ask for it.
Brand links for the next 1,000 years.
Great content gets mentions and links forever. Poor content is one hit, great content is always useful.
Work smarter not harder. Learn how to use APIs so you can automate this sort of stuff.
Get contact details of the people who shared your stuff so you can outreach out of them later.
Just make stuff that’s remarkable. Make stuff that’s actually worth sharing. If the page sucks and the site sucks and your site isn’t amazing, then there’s a clue there as to what you need to fix to get links.
Choose a target phrase, deconstruct whether you can do better than what currently ranks, wireframe, design, create a story, create an outreach list, get outreach support and buy-in, then launch when it’s amazing and you love it.
Target, SWOT, Strategise, Build, Launch
When featuring people in work, track those link clicks, and then you can tell them the traffic you sent them.
To sum up – linkbuilding sucks. Instead, identify target segments, find out influence, be there.
Can blend this with PR to take it offline and spread reach.
Session 3 – Lauren Vaccarello: Conversion Tracking – Online/Offline Attribution
Over the last 10 years, we’ve evolved from being about rankings and keywords to doing RCS. If you’ve got a billion dollar company, rankings isn’t where it’s at. You really need to think bigger than that.
Increase returns through data mining and analysis, rather than pushing up a keyword three places.
Lead gen is not ecommerce – you’re going to have long lead times and delayed purchase, with multiple touches from multiple marketing formats.
When you’re dealing with large purchase decisions, you’ve got to influence more than one person. If it’s a house extension, that’s both partners. If it’s a million pound software purchase, it’s a bunch of people in the company.
Getting the form complete is only one part. It’s then following up with every other point that needs addressing. Unless you sell leads, the form completion is only one tiny part of a sales cycle.
Imagine two keywords, red widgets and blue widgets. Red has cost per lead of £10, the other has cost per lead of £65. You need to cut one. Which do you cut?
If you go deeper and look at final sales, end ROI, you might find that the £65 CpL version works way better. Look at the right metrics.
You need to use a CRM to tie online into offline so you can actually see what the heck happens with all that traffic that gets driven by your marketing teams.
Side note: you can get GA to contain the data from the final ROI bit, but you’ll need to send that upstream, and it’s a pain. GA isn’t great at tying offline data in. As a result, you need a system designed for this kind of stuff. GA isn’t where it’s really at for this.
Multi-touch is absolutely needed for this, so you can tracking marketing efforts over time, so that you can get really tight with the end point results of what’s going on.
Build dashboards so you can get high level, at a glance views on what is actually happening to your business via different marketing channels so you can respond and evolve your marketing efforts by channel.
Work with your sales team to educate them on the data that you generate, so they’re able to do their jobs more effectively.
Attribution modelling is where it’s all at.
First touch, last touch, all touch, weighted touch.
SEM > EMail > Event > SEO
Total sale is £1 million.
First touch: SEM – £1 million. Obviously ignores everything after, so not hugely useful.
Last touch: SEO – £1 million. The opposite; ignores everything before, and so again not everything is useful.
All touches: £250k each. Useful, but weights towards things that get touched more. Better than first and last touch.
Weighted touches/influence model: Allows for weighting individual components.
As a result, each different part gets a different amount, based on how you model your data. Can fail if you mis-attribute weight to the wrong channel(s).
Remember that you can change which you’re looking at, and all have issues, so you’re going to want to use more than one to be able to model this sort of data.
When constructing weighting models for weighted touch attribution, hire really really brilliant people to work it out. Alternatively, look at hiring an external specialist agency to do it. Don’t try and figure it out yourself; you’re probably not a stats PHD and aren’t going to be able to work that kind of stuff out.
Also, as an aside to that, the more data you have, the better your modelling can become.
Not perfect, because it’s going to rely on cookies. However, you can augment this so that even if someone deletes their cookies, you can probably still re-make that cookie if it’s deleted, providing your technology is clever enough.
Session 4 – Dave Peiris: The Unexpected Value of Creating Things
Side projects can blow up huge. When they do, mine the contacts that you get and keep them live. They’re awesome.
Tangentally useful stuff tends to drive sales, because it’s going to get in front of the people you want to target for your main projects.
Awesome for branding.
Start with an idea. Work out what the simplest version of it is; that’s what you’re going to build.
Marketing that idea before you build it. Check there’s a market, and if there is, you can build an email list for it.
Build a prototype version (MVP). Make sure that this isn’t done by a committee – if it is, that’s a certainty for making your thing rubbish. Make sure you do one thing, and do it amazingly well. Don’t spend ages on this. If it’s going to fail, fail fast.
Launch. Look for the little guys and take care of them. See Rich’s presentation above for how to go about getting mentions and press. Make sure you know the story, and make it really easy for people to find that and write about it. Give away media materials, videos, copy. Give it in zip files etc. Really simple.
Don’t get down if something doesn’t get loads of press coverage. It’s nice if it does, but it’s more important that you have users that find your service useful, than that you get lots of press attention.
Normal people don’t link, so don’t appeal to them. Instead, create content designed to be really interesting and useful to people who will link (designers, coders, journalists etc). Normal people don’t have websites, but these kinds of people do.
Talk to people and show them your idea to make sure that it’s great. Then show them the executed version to make sure that that’s great. Then sort out your outreach to make sure that’s great. That way you’ll minimise the risk of failure.
Don’t worry too much about stuff going bad – most of the time, people just don’t care about a bad idea; they’re not going to turn rabid and attack.
Session 5 – Lisa Myers: International Seo – One Size Doesn’t Fit All
8.26 million international sites. Each engine runs different indexes for different countries, so you need to get your site in the right bucket. Not as easy as it should be.
TLDs – good, but potentially expensive and a pain legally in some jurisdictions.
Subdomains – not great
Subfolders – really quite good
URL parameters – dumb. Don’t do it.
ccTLDs are the best way to send a signal saying where you want to be geolocated. Pain generally worth it, where it exists, as long as you actually have a location in that place.
Subfolders inherit the link weight of the entire domain, which is nice. Also cheaper, no legal issue, particularly useful for informational sites.
Content management for ecommerce sites for international sites with subfolders for localising can be a nightmare of duplicate content.
Subdomains are possible, but you lose a chunk of link weight, and there’s no benefit really that you can’t get out of subfolders instead. Go for subfolders or ccTLDs instead.
25% of sites stuff up their geotargetting settings.
Have seperate sitemaps in seperate subfolders for different languages, if you’re going the subfolder route. Then submit those to WMT and geotarget.
Hreflang is awesome. Think of it as rel canonical for languages. It’s for duplicate versions of pages, in different langauges. Don’t use it with canonical though, unless you’re migrating from subfolders/subdomains over to ccTLDs. Very useful for that, so you can transfer weight, before 301ing later.
Hreflang can be done either in code on the page, or as a sitemap. Go the sitemap route. There’s a great tool developed by theMediaflow. Use it, it’s awesome.
IP detection is great, but remember that Google crawls from the US, so you need to be able to get the spiders through to the right content.
Get content translated by a natural speaker. Google translate is nice, but it’s not perfect. Don’t use it for content translation.
You need to get links built to your site from sites in the target country (ccTLDs).
To built those links, use native speakers who can write brilliantly, project manage closely, good reporting and great SEO tools (SEOmoz and Linkdex are a good combination).
Do Richard Baxter’s stuff for each area. Understand the market you’re marketing to.
You can use the same content in different countries. The outreach will need to be different, but the content can be the same. Re-use stuff that you’ve already developed that will work.
Reporting, recruiting, organisation and quality control are hard.
Session 6 – Paul Madden: Building an Outsourced Automated Infrastructure from Scratch
Everything can be automated to scale better. Needs a robust manual process first, before you can scale it.
For the processes, start by defining the process in minute detail, which can be deskilled. That way a computer can do it.
Work out for each thing whether it’s best to have those things being done by humans or hardware.
Hire people on odesk. Hire and fire fast, promote excellent workers to team leader, then let them take care of hiring, firing and training.
Spot check and review all data. Hand over more responsibility over time, build better reporting, and only interfere where needed.
Survey your people to ensure they’re happy working for you. Happy workers are better.
Mine the data you’re creating to refine your process to make things better.
Now hire a team to automate the data automation and mining, by automating using code.
Script everything that’s currently done to make sure that your end product works as well as possible.
Re-look at the process, and slowly try and remove where you can replace people with more software. This will be clearer now that you have a level of automation in place.
Session 7 – Phil Nottingham: The Building Blocks of Great Video
Video is not great in and of itself; it’s just a form. Video can suck, like anything else.
Video > What can we do video about? > Make video
This process sucks. Do not do this.
Does it need to be a video? Could it be good as a blog post or images or infographic? If yes, it should be those things, not a video. Use it right.
Video is about more than viral. Viral is nice, but not needed.
Host using VimeoPro, Wistia, Brightcove or vzaar. Don’t use YouTube because if it does popular, you can’t get rich snippets. Use something else instead. Make sure you shut down the indexation there though, so that people find and link to you, not the hosting site.
Don’t destroy your host – h.264, 1280×720 etc.
HTML5 video player so content can be indexed. JWplayer is great. Don’t use iframes.
Video XML sitemaps are vital. You can also (and should) use schema.org markup too.
Appliances Online do great online video for products. 100% higher conversion rate for users who watch a video. Adds 9% price to average transaction.
Why do they work? They aren’t ads, they don’t “sell”, they just explain the facts and detail, showing the story and use cases, and give the details.
Doesn’t have to be expensive. With two lights, a great mic and an OK camera, you can get great results for about £1,000.
Graphics work well too. Cheaper and quick to produce. Need to be great though.
Make sure you get the video transcribed, and get that on the page too.
Don’t put video for conversion generation on YouTube, just stops sales.
Instead, put that on your site through some other way.
For YouTube, put up stuff to entertain, drive brand, stuff that’s cool. Make sure it sells the story of your brand, that’s what you’re going to be using it for.
Use the YouTube keyword tool. Then include keywords in the ile name, include a CC transcript file, optimise title, description, tags etc…
Get links, views and shares – these feature in how well you’ll rank.
Look at retention, not views. Below the line is bad, above is good.
For YouTube, use full HD. Audio is key – focus on it.
You can create curated playlists on YouTube and embed them on your site using the API, which won’t appear on YouTube.
Session 8 – Patrick Mckenzie: Eating Cro: Real World Case Studies for 20-100% Increases in Revenue
Test theories about people, not ideas about pages.
Fix what’s broken. Where usability fails, make it suck less.
Someone needs to own the website – it needs to be someone’s job to make stuff happen on the website. If no-one is willing the be the guy where the buck stops, you’re going to fight for ages.
Guide people’s purchasing decisions. Help them to find what’s good about what you’re offering, what they’re supposed to do next, what’s right for them etc.
Ask for referrals explicitly, like you ask for reviews.
This is just hard work, but the gains will pay for it.
CRO improvements compound on themselves to make vastly greater changes in bottom line.
GA is terrible for funnels, KissMetrics is good, homegrown is best.
Track core use stuff and main revenue pathways. Understand what stops people from doing stuff that makes them give you money.
Get an idea for what numbers should be like first, then look at the outcome and see where there’s huge shortfalls.
Big changes get the biggest improvements.
Your customers don’t live on your website. They live outside of it, so get them into some form of contactable relationship, where you can make sure that you can build trust, credibility and brand awareness, so when they’re ready to buy, they buy from you and not someone else.
Build credibility, sell, build more credibility, sell, if they’re just not buying, build credibility and engage slowly over time. Eventually, when they’re ready, they will.
Over time, get people on to annual or lifetime billing. Means they’re far less likely to leave, and you get a huge spike in revenue in a tiny amount of time. Makes budgeting and spend far easier to work with.
Test price points as part of design changes. Also look at weeding out really poor customers.
Pictures of faces help increase trust. Make headlines actually sell value propositions. Make sure you use testimonials that counter objections, not just random quotes.
About 15% of your tests will make stuff better. Fortunately, the benefits of those will pay for the other 85%.
Go after UX/UI fixes first, then go after the big redesign. Big costs a lot, but the rest will pay for it.
Session 9 – General Thoughts
Disavow tool – probably a good thing as long as you’re not doing bad things.
Disavow is probably a data play.
Social web will become increasingly important and harder to spam.