The relational database is dying as the sole abstraction fror how we think about storage – giving way to friendlier more appropriate models.
Hank Williams, Founder CEO Kloudco
#rdbdeath
We’ve spent the last 3 years thinking about data – how people react to it and what they want to do. This has informed our work.
Why are databases important?
Mainly because we want to store our stuff – the real answer is that the structure of the database and how it holds your data and how you can ask questions about your data these all impact your application. We’ve been using relationsal databases since the 70s – and people presume that that’s the way to store data.
If you look at programmes you can tell why they did things the way they did – because the database was structured that way.
How do you define database death?
For 30+ years there was only one way to store data. But in the last 3 years people have begun to question this – are there benefits of other ways of storing data? Relational databases are NOT the way to store relationships between objects in the database in real time. The relationships are stored in the applications. This is a huge issue/distinction.
The benefits of relational databases – they are massively powerful and fine-tuned. It is now really good. The stuff inside Oracle, MySQL is really efficient.
The problems of relational databases – [photo of book the design of everyday things - teapot with handle and spout on the same side]. How we interact with our d/b is important the language is set from the beginning. The problem is it doesn’t reflect how you think about information problems. Software is complex typically because of the problem space. To store a tree I have to write code to create the tree abstraction – elemental. There’s a mis match between how you store stuff in memory and how you store it on disc.
Second issue relates to scale – the abstraction model goes away. When you do things like sharding (taking data and splitting it across multiple machines) the abstraction does not work at scale.
Third problem is precog skills are needed – you’ve got to be a psychic – when you start creating the application everything you’re going to use it for in the future! What happens is there’s a priesthood of database people who tell you how poorly you structured your data. In some areas – like accounting – flexibility is a problem. But in most high-volume data applications it’s impossible to know when you start where you are going to want to go.
Why are things changing now?
- Tools are now free. You don’t need to buy servers, software and data centre hardware now. Online web services – it’s not free but it’s cheap to do what you want to do and you’re not locked in.
- Our processors are now fast. They rotate at 54k or 72k per second. You only get one chance to read/write from that database every time the spindle turns. If your processor is really fast you can do really sophisticated things and the speed is no longer the negating factor. The hard drive limits you. With fast hardware we can create more sophisticated abstrctions when accessing our data.
- We are overloaded with data – individually and as businesses. We capture data and never again look at it. We need tools that will more effectively help us to make sense of the information we’re capturing.
What are people exploring?
MongoDB, CouchDB, Neo4J, Riak, Redis and many more. They simplify the abstraction, they are faster, flexible and more scalable. There is not yet a winning direction but people are willing to explore.
What would the best database interface look like? the StarTrek / Space2010 spoken instructions is our fantasy. But heterogeneous co-location of data is needed to achieve this. You do need relationships as objects. Treat the relationships as First Class Citizens in the database.
We need a graph database – like the Facebook “social graph” based around nodes (people) and the relationships (friendships) it’s now more sophisticated because there are now different types of object and you can specify different types of relationship. Everything you want to represent in the database are now your objects and the relationships between them mean you can represent everything and you cna query the system in any way. There are no more limitations if the data is there. It’s easy to add more object types and more relationship types.
What does it solve?
The graph storage model solves the largest problem in data management – “siloization”. The relational database model means every time you add a new dataset you create a new silo. This works both for B2B and B2C data.
So what about the semantic web? Time Berners-Lee’s statement about reinventing the web to do this means he created something that nobody understood. For most of us it’s a terminology we don’t understand in depth. Mainly because the tools are hard to use, they are designed to make the web better not to make it easier as a programmer to do your thing. It was not designed to represent ownership of information easily.
Fun stuff that happened along
Building apps using the technology we realised that when all your ‘stuff’ is in one place you need to identify who owns it and who has access, i.e. users and permissions. We re-do this over and over. It should be part of your database. This data is owned by X but can be used by Y and Z. You want your interface to do the stuff that’s obvious, but current relational databases are like an assembly line – it will do everything but only in a set way.
The second concept is “conferrence” [Deliberatly has two Rs. It means the same as conference but pronounced confer-ance]. It works like this – I know this person and connected to them by two other people who are friends. Conferrence means if you have access to the first person all the permissions to do with access to the first person confer onto the other subsequent objects that have a conferring relationship.
Notification - once you modify something or add to it or connect something to it you’d like other pieces of code to know you’ve made the modification.
Agents - once you have permissions in the database code can operate on behalf of a specific user. this is like a stored procedure. In a relational database only the ‘high priests’ can modify. But an agent has the right to operate on behalf of a specific person which they have the right to do on their own behalf.
What are some of the opportunities?
Machine learning or “AI” applied to everything. Artificial Intelligence is looking at the data and doing intelligent things with it – like recommendations engines. If your data is all in one place this is possible but it’s not in siloed data. The co-location is really important. ”Sense data”. The more data that’s co-located the more possible it is to make sense of it. As the volume of data grows it’s critical we have tools to make sense of it.
A modular web. The idea if databases can define a type and make it accessible you no longer have to write monolitich programs that do everything. Mash-ups are a type of this but are a limited way of accessing the data – it’s usually more about reading than writing. Create apps that don’t know each other but can inter-operate with other databases.
What does this all mean?
It’s about smart data storage that lets you extract real value from accumulated information. Create tools everyone can use and interfacing apps that use some of this stuff.
Questions -
What are the barriers to the vision?
I think it’s grinding it out. Plus a lack of money. There is not a lot of money for doing these things from the venture world.
how does the graph model improve the schema situation?
The biggest burden in relational database management is schema migration. In the graph model the idea is you don’t have to do that – it’s still possible to mess it up! I think it’ll be much easier.
Is this the second coming of gemstone?
New ideas sometimes take time to get taken up. There are no new ideas.
This is a guest blog by Larry Kim, Founder & VP of Product Development.
Over the past seven weeks I’ve covered six steps that can transform your search marketing campaigns into lead- and revenue-generating machines. We’ve come to the final step:
Step 7: Do it all over (and over) again!
SEO and PPC are ongoing endeavors. The more you put in, the more you get back. To see continuous improvement, you need to commit to continuous effort in the areas of:
- Keyword Discovery and Negative Keyword Discovery
- Keyword Grouping and Organization
- Keyword and Keyword Group Analysis
- Content Production, Content Optimization and Prioritization
- Search Campaign Refinement
It sounds overwhelming, but it doesn’t have to be. With one robust platform that handles all these keyword management tasks, you can create large-scale, high-performance search marketing campaigns without sacrificing productivity or blowing away your advertising budget.
If you want to start taking steps to build a better search campaign, contact us today to set up a live demonstration with one of our customer service representatives, or start a free, 14-day trial to see what WordStream Keyword Management can do for your business.
]]>This is a guest blog by Larry Kim, Founder & VP of Product Development.
He is outlining seven highly effective keyword research tactics that will help you generate more leads and sales from your search marketing campaigns.
Step 6: Clean up your Research with negative Keyword Discovery
Last week’s email covered the importance of prioritizing workflow based on your top-performing keywords. Now we need to talk about the keywords that aren’t working in your campaigns.
Focusing on top keywords is important, but it’s just as crucial to your long-term success to identify keywords that are hampering performance and filter them from your research.
This will save you time and money on both the paid and organic fronts. For PPC, negative keywords reduce wasteful spending on irrelevant impressions that drag down CTR and irrelevant clicks that don’t convert, ultimately boosting your Quality Scores. Eliminating those same keywords from your SEO research prevents time and money being wasted on optimizing content for the wrong audience.
So how do you go about finding negative keyword candidates?
- You can start by brainstorming. For example, if your business is cabinetry, your negative keywords should include “cabinet members” and other queries that reference the U.S. Cabinet. In addition, you can find negative keyword lists online for your industry.
- You can look at your search query reports and identify negatives on a case-by-case basis. This will help you catch negative keywords that wouldn’t have otherwise occurred to you.
- You can set up rules, or filters, that actually learn from your behavior and eliminate irrelevant PPC or SEO keywords from your research before you use them in your campaigns.
WordStream’s smart negative keyword tools do just that—it’s faster and easier than identifying negatives keyword by keyword, so you can start saving money faster. To try out our keyword tools with your own data, sign up for a free trial of WordStream Keyword Management for PPC or SEO.
Next week you’ll receive the last email in our “7 Steps to a Better Search Campaign” series. There’s just one more element that will bring everything we’ve talked about together for high-performance search marketing.
]]>This is a guest blog by Larry Kim, Founder & VP of Product Development.
He is outlining seven highly effective keyword research tactics that will help you generate more leads and sales from your search marketing campaigns.
Step 5: Focus on your top Keyword Performers
Web analytics are a window into your search campaigns’ performance, and integrating analytics with your keyword workbench enables you to act on that valuable data, as we discussed last week. The next step, of course, is to actually start taking action!
There’s no end to the number of activities you can carry out in search engine marketing. The trick to finding continued success lies in being able to determine where best to spend your time (and your money). The most powerful way to prioritize workflow? Use analytics to ensure you can closely follow your best performing keywords and recognize opportunities you’re not currently capitalizing on.
There are a couple of ways to approach this:
- Focus on the groups of keywords that drive the most traffic and spend. Remember, your top-performing keyword group may not be the same as the single top traffic-driving keyword. Target your ad campaigns and organic content at these groups first.
- Identify gaps in your PPC and SEO campaigns by analyzing them side by side and against potential traffic gains. With keyword source analysis, you can easily compare the performance of keywords from paid and organic search to each other and to traffic estimates from a keyword suggestion tool.
A great way to start identifying your most important keywords is to take advantage of a free trial of WordStream’s PPC Management Software or WordStream’s SEO Tools depending on your search focus.
All this positivity is good, but for fully optimized search campaigns, we need to spend some time on negatives too. Next time I’ll talk about how identifying negative keywords can reduce wasteful PPC spending and even save you time and money on the SEO side.
]]>
- Image via Wikipedia
I was reading Marketing magazine and one of the columnists was Will McInnes writing about Cadbury’s Wispa.
Itself a brand that’s been well used to great PR (thanks to inspiration from Mark Borkowski in the past. What interested me in what Will wrote was his criticism of Wispa asking visitors to register in order to get into the deeper engagement features of the site.
The registration was described as
a direct marketer’s fantastical wishlist
And that started me thinking.
The Early CRM years
When I was working with PRG doing CRM strategy – we’d set out a database specification that had hundreds of data points for each individual customer. We WANTED to know all about you.
The difficulty was getting that information from the customer in an appropriate way.
The way it is now
McInnes suggests that Cadbury use Facebook Connect to handle log-in and registration for its site.
Not a bad idea – the fan page relationship stays in FB and the marketing platform remains on the Wispa site.
As a marketer, I’d be nervous initially about allowing a third party to handle customer data particularly a closed one that didn’t allow data extraction. But then I had second thoughts.
How to assess your brand’s data needs
Wispa is a B2C brand – it has no expectation of selling direct off its site to customers and the ‘engagement’ is just about getting people to live the brand values of playfulness. And so why not outsource the ‘grunt’ registration to a third party. Particularly one where security and data handling is a key expertise?
Who knows, FB might do a combined advertising plus deal with Wispa if they use its platform to drive new memberships into FB. I see a nicely balanced relationship with each brand delivering part of the site and playing to their own strengths.
Let’s face it, what will Wispa do with your customer data anyway?
P.S. Afterthought – when the brave new world of VRM starts and customers hold their own data and brands have to ‘subscribe’ to the customer’s data rules and regulations – the boot will be firmly on the other foot and all those multi-field-integrated-user-databases will be worthless. And I’ve heard on the grapevine that Mydex has a launch date for its first version. Grin.
Let’s get the tables turning.
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_c.png?x-id=58d467b8-0165-422e-a48f-14da459821ea)
Help is needed:
Rebecca has been asked to speak at a Cambridge University Careers service recruitment evening about our industry.
I’d like to give them a decent informational talk with a bit of humour thrown in.
Here’s the brief.
You may remember we met back in the Spring and amongst other things I mentioned that I may run a Careers Evening of talks around the Advertising and Marketing sector. Unfortunately, this hasn’t come to fruition. However, the Advertising and Marketing Communications Careers Fair is happening on Thursday 29 October and I am putting on some talks during that event.
We talked about the possibility of your doing 20 minutes on understanding the family tree of the sector, what job titles really mean with some jargon busting. Would you still be willing to come along and do this?
Questions:
- List all the different types of Advertising and Marketing Communications agency specialisms
- Create a ‘family tree’ of the sector
- List all the main job titles found in agencies
- List all the main job titles found in client-side marketing departments
- What does each title REALLY mean?
Sounds like an A-level exam….. but can you help me out?
I’ll get going with some of the lists. More tomorrow.
Please post your suggestions below.
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_c.png?x-id=b4e50485-7da8-4f3e-85cc-c66d65f754c1)
My new evangelism is data. Long-time readers know that I bang on about databases quite a lot. And of course, the information stored in the database is where the value lies.
Gaining customer insight from research and activity observations is one of the finesses that I enjoy in direct marketing. As a marketer I often complete questionnaires sent by other companies. I do this partly to see if there's stuff I can learn but also to get insight into what level of sophistication their marketing is running.
I bought a couple of bags from online manufacturer, Timbuk2 over the past year. Then they sent me a survey.
Much of it is bog-obvious (no less useful because of that) with questions about what you've bought from them, what you'll use it for, other retailers you buy from, and demographic information. [aside - note that they should / might know about me from direct selling and should have known what I'd bought from them before but they also sell via a distribution network and so some recipients may not have bought direct. If this had been me, I'd have done two different surveys one for direct customers demonstrating that we know what you bought from us and one more plain-vanilla.]
But my eye was caught by two questions that seem out of the ordinary.
18) If Timbuk2 was a person, how would you describe him/her? What's their personality? What do they look like? What do they do in their free time?
24) Please choose the statement that best describes you:
I live in a city because I love being around diversity in people and culture.
I measure success by personal fulfillment as opposed to cash.
When I am passionate about something, I spend a lot of time and money to be an expert.
I am curious about the world and motivated to make a difference.
I am part of a large friendship network.
I am an idealist when it comes to government and believe progressiveness can delivers results.
I am environmentally responsible and take active steps to reduce my carbon footprint (ex. Recycle, etc.)
None
Now the first one is a pretty standard ask in research (sometimes you compare to a car marque). And it gives nice insights – but the second question is REALLY USABLE.
Imagine how you'd treat different customeers differently if you knew they valued cultural diversity in a city versus someone who spends money on their passions.
I think the company must have created a range of sub-groupings of theie customers and then will use this information to communicate about the brand relating to the desires and attitudes of customers in each group. The friendship network question is very astutue as it may reveal 'connectors' who can promote the brand through their network. I am not certain that all the statements are mutually exclusive but I was really impressed by this company's customer insight gathering.
If you want to see more on differentiation by customer need read the Intuit / Quickbooks rant -cum-debate part 1 and part 2 where my readers share their expertise on how to look for needs-based differences between customer groupings.
PS if you want a copy of the whole survey, get in touch.
]]>I often do surveys because I'm a marketer and I appreciate it when people fill out my surveys so I pay back by doing theirs. I also look closely at the questions they ask and see if there is insight and improved questioning methods that I can learn from.
I did one this morning for Responsica
Now the structure of the questions and the insight they are seeking is pretty obvious – looking for 'pain points' in current marketing practice. They are clearly about to launch a new service that aims to assist SMEs suffering in this area.
Question 4. Which of the following have you USED in your business IN THE PAST 12 MONTHS? (followed by a list of marketing tools/methods)
Question 5. Which THREE marketing tools generate MOST SALES revenue for your business:
I like the list which has a good spread of older and newer techniques and when answering question 5 I really had to think hard to separate my perceptions from the hard measures that we actually use.
Then they move onto
Question 8. Rate how EASY or DIFFICULT do you find the following when PREPARING your business MARKETING: [finding a copywriter,collecting customer data, a list, lead follow-up, lead conversion, materials]
and then they go into price sensitivity questions for their un-specified service offering that 'overcomes the issues' you list….
But it's the last page that I liked the best. Here's the screen grab.
- The thank you is prominent
- The URL is back off the survey tool site and onto their home domain
- The sharing button is prominent
- Large number of sharing tools suggested
- It works
Best Practice in ending surveys. Well done, guys.
]]>Out of my usual curiosity, I checked the Firefox SEO tabs for a client's news feed recently.
here's what came up
#1 | PR: 4 | Google Cache Date: Jan 20 2009 | Traffic Value: 3 | Age: 12-2002 | del.icio.us: – | del.icio.us Page Bookmarks: – | Diggs: 0 | Digg's Popular Stories: 0 | Stumbleupon: – | Twitter: 5 | Y! Links: 1,220 | Y! .edu Links: 0 | Y! .gov Links: 0 | Y! Page Links: 720 | Y! .edu Page Links: 0 | Technorati: – | Alexa: 2,540,535 | Compete.com Rank: 0 | Compete.com Uniques: 0 | Trends | Cached: 356 | dmoz: 0 | Bloglines: 0 | Page blog links: 8 | dir.yahoo.com: – | Botw: 0 | Whois | Sktool | Yahoo position: 1 | Majestic SEO linkdomain: 247
Now there are some new terms in there I'm not familiar with.
i need help understanding the following. Any advice?
- Traffic Value
- complete.com
- Trends (why no number next to it)
- Botw
- Sktool
- Majestic SEO linkdomain [sounds marvellous an SEO that's majestic!]
Image by Daniel F. Pigatto via Flickr
I got an early invite to play on Viadeo.com when it launched and I duly plugging into the browser window and had a look around.
Today I got a connection request from a friend in France. Neat – when I clicked the link in the email it took me back to the site and had written a standard reply message to him in impeccable French! That made me LOL. So I added a P.S. in bad schoolboy French so he knew it was really me writing.
Then I thought I'd play more and clicked the 'Outlook import' link thinking I should really try and make the functionality perform and got this error message
Sorry, the Outlook import works only with Internet Explorer on a PC.
/Use File Import to import your Address Book.
FAIL
Reasons for failure
- Firefox is my browser and if your software doesn't work on my browser, I ain't gonna try hard for you
- Linkages and creating connections is the whole point of Viadeo (a business social media site). If Outlook only works with IE you fail doubly because I can't tell my network about you and so you lose the opportunity of one of them also joining up.
- Make it easy for the simplest user. This wasn't. Unsubscribe [for now]
Anyone want to persuade me to change my mind?
This could have been a useful tool for Relationship Development , Creating Opportunities and Making New Business Happen .
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_c.png?x-id=efd135ed-7f59-451a-bcc8-c85c87d0172d)