Dirty data and how to prevent it

Andrew Bonar 0:01
All right. Sorry for the delay. There's Raymond. Thanks again. And please do join us in the chatroom because I think there's lots of people. Hello. Thank you, Raymond. So right now I want to say thank you for agreeing to come on stage in box Expo, and I'm really looking forward to your presentation. Are you looking to screenshare this morning, apologies to watching. We're gonna do a quick test to make sure everything works live. Just everyone knows this is live. We're running a bunch of fun recorded videos have helped. And yeah, quite a few could have a look for the screen share button. And yes, yeah. Working on this, yep, it's one second, and I will just bring it online. So you see my page now? Yep. is up and it slides as well. So again, if you could take a couple minutes to a few seconds, at least, to introduce yourself, and your agency, and everything that you do, that will really be appreciated. I'm sure everyone who's listening, couldn't agree that the last moment to speak for us, I'm very, very grateful. But we're not as well prepared. And we might be for this. So I'm going to let Quinn introduce himself, because I know that Yeah, well, he's brought an awful lot of fantastic delegates. So I do have to say thank you for that right off the bat. And you can also watch queens and Dutch day. So over to you Quinn's and

Quint Wapenaar 1:42
this one. Thank you. Thank you, Andrew, for inviting me as well. So my name is Quint Wapenaar, I'm campaigning for pride 100% email, I'm sure he was an agency. We are at like 30 people and cover everything about email marketing, and even a bit broader. My function is your campaign developer is pretty technical. So I built templates, SQL data, connections, landing pages, and everything that comes with it. Over the past few months, I'm really into the data science. So a lot of cleaning processes and fixing stuff that should have should not have been fixed. So when I was asked to give a talk here, I had my outline in like 15 minutes, I'm going to be talking about dirty data, and how to prevent it. so dirty data, what is dirty data. Dirty data is often something we don't immediately spot. Most of the times in my case, we find it through reporting or complaints of customer service. And I hope with this session, I could help you guys to prevent a few mistakes in the future. So the common mistakes that I see most of the times is wrong sanitation. So the agenda isn't right, the first name or the last name, it is not right. Sometimes you even have informal or formal solicitations based on an age and when the age is not a right you maybe have a wrong interpretation not to want to sending the wrong order info. So when you send an order confirmation with there's a lot of data in there a lot of personal information and a lot of information about the order. And sometimes the match to the to the order data edition, right? So you send the whole order info of the wrong version. And this is one of the most painful mistakes I've ever found is that we still mailing someone who had died. Most of the times when someone dies, relative calls the clients service or customer service desk. And instead of a this burden is light come out of your system. And then most of the connections there's a specific CRM system specific CRM field in a system where you can say someone dies and most of the mailing lists don't opt out. The person I'm still meeting them. And it's really painful. If you send someone a retention email once a long time, we had seen you or we have missed you. And then a relative, she's that email. And so I wanted to make sure there's a really good call to action to preference page where people can easy out. Meaning this happens also, quite often, most of the times is going wrong when you're not using it the email address property. The last one is triggered based on behaviour. For example, when some, when you send an email reminder based on click activity, and the person had to click the unsubscribe link. So actually, there's a lot of time many times in campaigns where they just send out a reminder or or an extra email based on a click, and then just say click on any any link. So when they hit the unsubscribe link, they got an email as well. So why do we want to prevent this? Because good quality data is better marketing and sales and increasing your goal. Most of the times your goal is to get more money. But you often see also campaigns or creating less report tickets or more awareness for your brand.

Where does this data come from? So we could create cheapo scripts are modified from memory they the incoming data and change it. But this is not a long term solution, I think we have to act on this mistakes in the beginning, the early stage or aquasource. So what types of data are there, there's contact data, where first name, last name, email address, etc. So fields, their sales data, this depends on the company, sometimes its sales data, sometimes customer service data, or both. Then you have demographic data. behaviour data, this is really nice for your journeys. And then you have third party data. And I will be focusing on the first three, that's those three are the types of data where lots of mistakes happens and the most painful mistakes. The other two behaviour data third party data is most of the times pretty consistent, and it's filled by machines. So if it's wrong, it's consistently wrong. And it's most of the times fixable by code. And I only got 30 minutes, 20 minutes left. So the first three are filled in by a human and human make mistakes, and even on purpose. Most of the times later on, they're going to be explaining a lot more about that. So you will make mistakes. And it's our job to help help them. I think it's part of our job because we know all the different systems we have contact with the different stakeholders of every different system that is almost every different system there if Shall we know where we can fix the problem for the salespeople and the customer service. So the most common mistakes are misinterpreted, misinterpreted form fields, incorrect use of the system, and mistakes on purpose. I will cover the first one. How do we prevent misinterpreted fields? How do we do this is by user experience, I have to make these points. I hope everybody is already stopped using this. So we can skip this point fast but I still see a lot and our forms with placeholder text as just based on the text. So it's nice small that is not very practical. Because once you prefilled this kind of forms you don't know if it's right or wrong. So if you go to a preference page, for example, and asked them to change their information, or extra information and they see this you don't know that john doe is not right. be that way around. So how do we fix this pretty simple use labels, it makes it a bit larger your form and use validation. Also, positive validation is really, really helping increases the how much a customer leaves on a on a forum, what you see is that if you give positive feedback, you can also unlock more information of your clients because they are willing to feel like rewarded and really happy and, and so in more information. For people who really like those placeholder text fields, please use this fancy ones. There are a lot of libraries out there. This example I made with bootstrap with auto materialised facets. So you can still have your fancy nice forms. But then, once prefilled you still she was supposed to be in the input field. So we've covered already a bit validation. This is something that is really underestimated when it comes to clean data. I think there are all kinds of theming processes where you can have fixed things MySQL or manipulate stuff. But I think validation is one of the key things to make sure that the data is getting correct to the shores.

So how does this works? There are two types of validation, you have client side and server side validation. blindside validation is done through JavaScript and JavaScript, this is one of the big advantages of JavaScript, he can give feedback based on behaviour on the site itself. So when clients focus on input field or leaving it or scrolling, or whatever, you can check the data or even manipulate the data. On client side, the biggest con of of client side is that it's based on JavaScript and JavaScript can be turned off. And so you can fully rely on client side validation. So you need to have sharps right for the nation as well. And server side validation is done on a surfer, obviously. So this can only check the data that is already sent. So no instant feedback with a chauffeur is the gate get to database shows the most important validation, there's done that you're wondering, when it's not going well to the database. There's a lot more detail. So when someone fills in a form, they go from, on the client side, they fill their information, it goes sharp charts, and then sharpside. If it's all passed, it goes to your database. Check.

It's pretty logical process. And what's really cool is the solution for validation for Sherpa and client side, it's both

run by FX References The thing that checks the value that someone puts into an input fields. And what's really nice is that the client side and the server side, both use the same structure, they both use functions, and they both use ref ref x. The only difference is that most of the times client side JavaScript and server side is PHP, but there are no other programming languages like Angular or react that our server side JavaScript. So what we can create now is a shared foundation library. We should I think we should keep the foundation the foundation roof you have on your on your forms should be the same as in your CRM system or at your distribution platform. So all the data is consistently and you could not easily believe you can make a set of rules. Same as the design library you use. in your home in your company you can use if your condition as well. Yeah, there's just nothing Nothing new. There are a lot of validation numbers out there, there's even a big chance that your company is already using a validation libraries, most of the companies I worked for, have multiple libraries used and different checks. So the first thing that I do when I come to a new company is to check, okay, what kind of gradations are there, and what's the difference? A lot of companies use different validations. Between client side and server side. Sharp side is most of the time it's built by an IT department and the clients affiliation is most of the times built by a front end department. And sometimes, if the company is really big, they don't talk that much to each other. So they have a different set of rules. And I think it's part of our job to point that out and help them. What's also really nice, if if the data is clean, you don't only help yourself with it, but I found this chart of data scientists have been perfect productivity breakdown for the data scientists. And what you see is 60% of their time goes to cleaning and organising data. So when you actually datian and fix problems at the shores, you will not only help your shell, but also different departments in your company. So one down to go and this to have a lot of common. This mistakes are most of the time made by other departments, obviously not by us, which is most of the time to CRM or support system. And we have a connection. The ESB has multiple types of connection with a CRM system or support ticketing system system. And what I find out is that, sadly enough, most of the mistakes that I found, were not necessary mistakes, but they were there on purpose. So I will cover these two together. I made this journey. And not only I the whole team made this journey, at a really big moment agency, I will explain that in fit can be pretty fast. So what explain what the campaign does, is a company puts a vacancy online on any site. And we scrape this within a database, an algorithm, try to find a match between CDC and the company in their database. So in the database of the employment agency, if we find a company, we check if there are contact persons, if there are contact versions, we are allowed to mill, we check if there are possible talents. And then we sent out an email to the company that vacancy online in the name of the name of a consultant who works at the agency. So we made the match, and that sent the email.

And what I find out is that john doe is a consultant at an employment agency. And he is performing. Yeah, not so very well. 86% of his clients don't have a high enough match rate to send out an email. So when we send out an email, we check in the algorithm if there's a match rate higher than 90%. So when it's high 9% we sent out an email. But when it's less, we don't send out the email because we think that the match is not good enough. The second point is that 23% of his clients, we can no we're hard buttons. So a lot of his clients that we were mailing were hard bounced. 87% of his clients that had opened the email with a personalised offer didn't show any interest. So an interest was measured by going through the resume page. So all the tenants of a resume page where they can see what kind of work they did and stuff. And they could also click on a button where they can get in contact with john doe. punishments were measured and we use that a lot of his clients weren't reacting or they are shown any interest. So it was time to investigate. So maybe our algorithm wasn't good enough. Or maybe we miss mature data to the scrape the data that we're seeing this result author shunning 1000 mils, so we decided to change our code. We didn't think that john doe was doing things wrong, maybe it was our algorithm, which went by Andrew, john doe was still performing best. And he was not alone, we saw that not only john doe, was performing bad, but his whole team was doing bad.

So time to investigate again, shall he here's a list of the things that we have found.

They use a hash tag and email address. pretty strange. They used client fields to leave notes for internal use. So what they did is the first name, the last name were obviously filled with, but the eccentric fields in the CRM system, they were used to fill in things like notes for internal use. So don't email disperse, or, or we don't work with this guy anymore. That's very kind to us, or whatever. Then point three, they didn't register all the moments of contact with a client. So when I found this, I get all my insights. And we're ready to confront john doe, with old age mistakes. So I got on the phone, and we talked, we talked. And the strange thing was, he was very kind. And he explained why they work this way. And I felt sorry for him. They were misusing the system for their own benefits. So here's why. They use actually the human rescues. This was done because the out field in the sheer interesting wasn't working. He he put the opt out fields, a lot of times for different clients. And what they got is really they got calls from clients that they said, I called you and I wanted to get knocked out and I still got meal meals from you. They asked the IT departments to fix this, but they didn't fix this, or they thought they fixed it. But they didn't know about the connection they have with the ESP system that they use. They use fewest leaf nodes. So what they did is they have an internal dashboard. So they use these notes to get an overview. So every morning, they get it around like a standard. And they feel to the shearim system with the with a dashboard. And they use the internal notes to filters. There wasn't a proper dashboard show they misused those fields to help their process. Yeah, but we use those fields for our algorithm and match for matching stuff. And so we had to fix that. This was also very painful one, they didn't register all the contact moments, there was simply no time their target target was 30 calls a day. So of course 15 to 20 minutes. So this is seven and a half hours of set 10 hours a day. And registering a call takes over 10 minutes if it's certainly if it's a long haul. So there was simply no time. And what they did is just registered, the most important golf. So how do we fix this. And what we did is we added a step in this process. So we created a dashboard to give more control to the consultants. What consultants could see which companies we were going to meal, email them email the next day out of their name. And we also show them all the fields that we used in our algorithm, and they could change that in the dashboard. So before we sending out a date for sending out They had the opportunity to change those fields. And we also show the match rates. So also the clients that with a mess metric metric less than 90%, were shown. So if they tweak those data or clean that data, the metrics will maybe get higher. And then there's a big change, we could email those persons as well. There's another big change. So the 30 calls, they have to view was now changed to 30 contact moments. And we agreed that every person over we made by our algorithm was a new contact moments show this combination with the dashboard was really changing the refill really shows with the campaign.

So here are some results. It's pretty, pretty good results. So one is the controls, we're adapting this campaign. So ancient consultancies that they get a lot more contact with potential customers, we saw a massive increase in contact moments, also, because we inserted of contact moments in their system of active leads, get a lot more active leads. And because I'm logging, our logging for our contact moments, we could easily measure how much clients were reactivated by this campaign. So I think all these results are very nice, but I think this was the biggest win, the biggest win were that there was no more misuse of the system, at least not not that we could find. And maybe they are still misused, but not on the fields that we monitor. So and we had to get your contact profiles. So the match rate of the algorithm got a lot a lot higher, and because of the richer profiles, and what we saw is that not only our algorithm was increasing or getting better, but also other services within the company, were benefiting of this rich, richer contact profile. And this is the biggest one, I think, is that we created more data awareness with the consultants, and maybe even the whole company. And this is one of the biggest things that we all should do is, is check who's the end user, and make sure that they are also becoming a stakeholder in this in this whole process. So what we did in this case is they had over 50 departments. And in every department, we asked someone to be a super user. So there should be user dedicated time, a few hours a week, or a few hours a month to get us more information short, we had a call every month to share a dish or the things on our roadmap. This is things we would want to change in our algorithm. And we still see problems with the data you feel you guys fill fill in. What do you what do you use that do that. And this really made our process a lot a lot more. Better. And already the industry is growing and growing, and the data is getting better and better. So we covered I think all the points. So here's a quick round up. So start with you. I hope everybody already does that. Then we have validation. And I think clients are validation. I think every company uses server side validation. I think also this can get better and better when you start with clients has validation and if you want to take it to the next level, make a library of it and document this very well and spread this over to all companies and every department should know about your validation rules. Then you have a monitor and what I do that over the last year or three years before I start building a campaign, I also think of the monitoring or reporting or dashboarding or whatever you want to call it. This is a really big part of your whole journey. Are your your whole campaign. And I think this can really help shop through most of the times either a campaign and it's running and it's working. And we don't want to earn anything. So after a year, or maybe the first two or three months, and after that, it's just running, and we don't look at it again. And with monitoring, this can get much better and better and your campaign is getting better and better as well. The last thing is

to include the end user as a polish. I think this creates more data awareness, and also helps you to get rid of your dirty data. So this was it's Thanks for joining. And if you have any questions, please feel free to contact me. I think all my info is in my account, I think.

Andrew Bonar 31:13
Thank you. For that one time. browser. Quint. Sorry about that I have some slight technical issues this way. I really enjoyed that presentation, I had the luxury of actually getting to watch all of it for the first time and or not

been able to get to watch the whole thing in the studio. There's lots of really useful insight there into, you know, data hygiene and managing it and Raven two came before you would have approved, we're not talking about you know, waiting till the last minute and then kind of wash talking about being smart from the beginning. So I think you touched on a lot of great points. But anyone wanting to do anything with the points that you raised, are really going to have to get a lot more information about it. So I would suggest they reach out to yourself and 100% email, see if you guys can help. Because they were brilliant points. But I think each each and every one of them. I mean, you talked about JavaScript for a few minutes, but could have probably talked about that for hours I imagined. Yeah. And yeah, there was so much there. And I really encourage delegates that are there to reach out to you, I'd know that you're probably around for most of the day for a touchdown and stuff. And I'm really, really keen to join one of the Dutch sessions. So I want to say thank you for your time. Thank you for the presentation. Thank you for everything that 100% email did to help make this event a success, because you guys did a lot for the Dutch day. And the event overall, you brought some brilliant delegates along. You delivered some great content yourself. And certainly Mr. Wilson, on stage. So just one say thank you from everyone that email, email expert and the team inbox Expo especially for everything I did. And I really encourage everyone to check you guys out. And thanks for your time again.

Quint Wapenaar 34:02
Yeah, thank you. Thank you for having me, was really nice to do

Andrew Bonar 34:08
this. And so it's probably 40 minutes before the next session come out. Haley is going to be talking to us shortly. I'm about to wrap up the studio for the day before diamond comes in, to take over. So Devin will be running the studio for the rest of the day. polished currently, but yet, we're live and I want to thank everyone for having come along. I want to thank everyone for making inbox Express such as success exceeded our expectations. When we started, we can't be we can't be physical. We try to be in Spain this year. That didn't happen. It is what it is. But when we thought we were To go online, we we were talking publicly about 300 delegates. And, you know, hoping privately between myself and Eddie that we were going to see five or 600. And we actually went out to sponsors and told them that this event was capped at 900 delegates and that we wouldn't have more than 100 delegates. So we exceeded that and we got 1000s. As of 1060, last time I checked, I don't know what it is right now. People are still registering. A lot of people registering with limited access tickets. It gives you live access, access to everything you saw on stage or the workouts or the breakouts. But what it doesn't give you is access to the replays and downloads. Everything that's been hidden been delivered over the course of the event is fantastic. And email geeks community deserves to have it and habits at no cost. But there is a huge cost involved in doing this and doing this right. So I hope people have noticed that you know, we've put a fair amount of effort into the production. Thank you very much 2131 photography in the UK. We've been running studios from the UK for us. Thank you very much to do our studios, Diamond news team in the US will be running the US studios for us. Thank you to Jennifer thank you to other people that are working on the team behind the scenes. Big thanks to them, we do need to to help pay their bills, we need to pay our own bills we need to keep the lights on. So whilst the content will eventually be available for free. If you would like it quickly, then you can pay for all access upgrade the all access upgrades available through the booth area. Do check out the other exhibitors in the booth area while you're there, to pop in and say hi, I know there's at least two vendors offering Cash Money type voucher as well. Amazon ventures at least almost as good as cash money. So do look through and see what's going on and Matic have a session that they've decided at the last moment as a TPM inside their base. And you can learn all about their open source, a system that we actually use ourselves. So it's what inbox Expo uses the messages you're sending from puppet. It's a good system, it's not for everyone, but it does work. And it's very powerful. So if you want access to warnings, then go into the vendor area, click on this tool recordings and click the upgrade, it is half price. So the price at the moment will go up to full price in a couple of hours time. But for the next couple of hours it is half price, what you get access to is the recordings and the slide decks and the PowerPoints we do have not needed we don't necessarily have all of them will reach out to speaks. If they're happy to share them with you, we will share them with you. Some of the speakers may well be giving extra goodies that we're not going to announce or disclose. But they may well be added to the pack in good time. The other thing that you get to do is get certified so if you've missed any of the certified sessions and there was a whole bunch, there's some over a half dozen certifications then get those certifications allow you to add them to your LinkedIn profile. They're verifiable clickable. And so they carry a great deal of weight.

You've got to really want to want certification, for example from validity, so very well respected company. So all of these certifications are done by email expert in association with very credible brands and leaders in this space. So there's literally 101 is certified by Valencian email expert. We also have running right now and I don't know if it's a good time to crash the session. But maybe if you're quick, we have the advanced email authentication yakob is delivering a multi hour workshop today depending on how many people turned up and how much help they need. That workshop could run anywhere from two to three hours or more. And he will take you through everything that you need to know about authenticating email and SPF Deacon demark and all the rest of that good stuff. And you know you need to know how to get deep into DNS records and the like. Now if you've missed that and you've missed the Liberty one or one workshop, or you've missed the email accessibility session or in the the how to make email testing work properly session, um all of those are certified as an all access pass holder. You're able to access those recordings and also do the testing those attended the sessions live, we'll be able to get tested and receive this certificates, there's no need to upgrade to that. But if you've missed out or you want to rewatch, then you will need the upgrade. Gift with the upgrade, you'll be able to watch those videos as many times as you want and the recordings, you'll be able to review the related materials for the workshops that will allow you to see the test with far more confidence. And the other advantage of the all access with the download upgrade is the fact that you can retest. So if you fail the test, depending on which test is anywhere from 24 hours to seven days later, you can reset your test, review the materials that was so may well have a to a recap, type workshop with some of the speakers for some of the certifications. If that's necessary. We'll be available to support you as a paid upgrade. candle holder. What else would you get? But essentially, that's what you're going to get. So you're going to get access to all of the recordings that we made over the past four days, nobody could humanly possibly washed everything. There was an incredible amount of good content, you also get the apps to get certifications, you get hot thanks. You get to allow us to put these types of events on again in future. So if you like what you've seen, you want to see more of it. Please do support us. Please do take the pain upgrade option. Otherwise, you know wait around a few weeks of content will be released will be available in the mail expert.com. It's a big job. But more than anything right now I want to say thank you. Thank you to the delegates, thank you to the speakers, thank you to the sponsors. Thank you to the team. Thank you more than anything to Nellie, who held this all together. Thank you to the innumerable partners who helped us get the word out how big thank you to the DMA, thanks to the DMA, UK for letting their membership know about us for putting together the panel for sharing that information. Thank you to the espc who did the same thing. Big organisations have put their support behind the mock Expo and that is a big reason why we we've been able to grow to where we have, it's early days, we do hope to finally realise the dream of a hybrid event soon. And anyone who has got a page, all access upgrades to the downloads will be sent special offers, you'll be the first to be able to register for the physical events, we believe the hybrid events when they do happen, are going to be micro events. So we're not going to see 1000 1500 delegates, anytime soon. Anywhere where we plan to hold an event anytime soon, we don't believe but what we'll have is limited socially distanced events, I believe, I believe that's where we're going. In any case, those that have financially supported us always get first dibs on tickets get the best rates or future tickets.

And indeed, there will be other online events in the interim, and you will be able to get at least access to one of those events at no cost. And the details will be revealed. After you make the upgrade. Have I made the point the upgrade really, really, really would help us at this point. And aside from that, I'm tempted to start thanking individuals. But that would be a really bad road to get to town because there are so many of them that there's over with 130 speakers and performance and the team. And the teams teams that we've hired. There's over 150 160 people that have invested a minimum of five hours each and some of them invested anything up to 200,000 this project. So we're incredibly grateful. There was so much hard work that went into this. So I think I've spoken enough. I do want to say thanks to everyone. I'm humbled truly, truly humbled. It's grown faster, and bigger than I could have hoped. And, you know, this is fast becoming the everything email conference that we dreamed about. So everyone, thank you very much.