Sean Davis

Always a bit gratifying, if not a sign of the times, when the biggest and best-established media outlets are shining a light on best practices us folks in the Trust and Safety industry long advocated for. Good on the BBC for surfacing this. The S.I.F.T method for handing misinformation is akin to looking both ways before you cross the street. In other words, something that should be taught early, taught often, and be almost second nature in day to day life. In both instances, albeit different ways, our safety is at stake. #contentmoderation #ugc #webpurify #bbc #siftmethod #misinformation https://lnkd.in/eK-h9uj6

9

1 Comment

If you want to grow your list, use Onsite Opt-In. Simple as that

11

1 Comment

After the Gleen's Self-serve Platform and Gleen AI Shopping Assistant, we're back with the third installment in our AI trilogy – drum roll, please... 🥁 Say hello to 𝐆𝐥𝐞𝐞𝐧 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭 𝐀𝐬𝐬𝐢𝐬𝐭! While the first two interacted directly with users, Agent Assist is meant for Support Agents. It's a co-pilot that turns support agents into top performers. With Gleen AI Agent Assist, customer support agents can focus on what truly matters—building meaningful customer relationships—while the AI handles the heavy lifting. Instant summaries, sentiment analysis, and consistent, fact-checked replies reduce the pressure of high-volume inquiries and improve overall customer satisfaction. It’s not just about automation; it’s about empowering agents to deliver smarter, faster support. I couldn’t be prouder of what we’ve built, and I can’t wait to see how it helps companies elevate their support teams. #CustomerService #AgentAssist #AISupport

27

7 Comments

🚀 This is a significant step forward! 🌟 I'm hopeful this lawsuit will lead to better protections for both buyers 🛒 and sellers 🏷️, improving marketplace integrity 🛡️, blocking deceptive practices 🚫, and giving sellers a more even playing field ⚖️. #Avenue7Media #MarketplaceIntegrity #FairPlay #AmazonSellers #BuyerProtection #FightFakeReviews #eCommerce #BrandProtection #DigitalTrust #SellerSupport #OnlineSafety #ConsumerConfidence #eCommerceSolutions

7

1 Comment

in the wake of the leak, let's talk clicks again. The only thing the leak proves is that Google stores click data. This makes sense. It doesn't prove how they use it though. so what do we know? We know Google uses clicks for personalization. We've all seen it. Once we click something we see it more often. We saw tons of signals for this in the Yandex code leak too. We also know Google is great at spotting trends. Rand Fishkin proved that when he did his experiment. Google spotted a URL was trending and temporarily boosted it. But as soon as the clicks died, so did the temporary boost. Definitely useful information. But there's a click data field.... There is. That doesn't mean it's used as a ranking factor. Let's dive in and get technical. A "Factor" means it's something that gets a weight that contributes to ranking. I 100% believe that outside of personalization and trend boosts clicks are not a "direct" ranking factor. They're still used though. let's look at how: In Machine learning there's an X_train and a Y_train component. Think of this like a high school math class graph of points. In this case, X_Train are all the "factors" - everything we know about a website like pagerank, cosine similarity, title/query coverage, BM25F, Salience, link data, spam scores, ads, page speed, etc. Hundreds of things. The Y_Train is your adjusted/filtered clicks associated with every site that's used to train the model. The model then tries to give various weights to all the X values until they closely resemble the plotted graph of Y values. Once they are able to weight all those x values to resemble the y values, they save the model and can use it to score any other future website. The clicks however aren't ever weighted in the scoring process. JUST for the training. This is what I think Navboost is doing. It's predicting the clickability of a new URL based on historical data it's seen for URLs with similar features. This is also consistent with the statements Googlers have said. I haven't had a Google rep ever tell me (nor have I seen in print) a message about saying "we don't use clicks at all" - it's always carefully parsed: Clicks are not a ranking factor, we don't directly use clicks in ranking, etc. IMO that's not a lie. It's careful word parsing, and since most SEOs don't understand the process of ranking or what a factor is to fully grasp this, we were confused. So why wouldn't they use clicks as a direct factor (outside of trend boosting and personalization?) It doesn't really work. Remember the old joke about the best place to hide a body is page 2 of google? Nobody goes there let alone clicks. That means for most queries, 99% of sites would have 0 clicks. Sure you could smear that data from the domain total clicks, but that becomes a very noisy signal. One page on my site ranking really well for one query and getting clicks doesn't mean another page is relevant for another query. That wouldn't be a good signal to use.

213

48 Comments

Bots are blowing up your metrics. Everyone in DTC knows of the bot problem. What are you going to do about it? Sendlane sits on it's own hardware. We deliver billions of emails a month. You have IP pools that are the best in class when it comes to types of clients and number of clients on each IP pool. You have proprietary technology to prevent bots from hitting your lists in the first place. Deliverability has never been more important and key. You could orchestrate the most grand flows in the world, but if your email doesn't hit the inbox...well... Kind of no point. Like a Ferrari without an engine. Hit me up if you are having bot issues.

48

15 Comments

🚨 Scam Trends Alert from Amazon 🚨 Were you surprised to receive an alert about a purchase on Amazon that you never made? The message wasn’t really from Amazon. You never placed an order, but scammers claim that there is a shipping issue. These scam messages are increasing recently according to Amazon customers. Here are some important tips to identify these scams and stay safe. #ScamTrends #Amazon #ScamPrevention https://lnkd.in/d9AdY_4r

84

7 Comments

Google will not be able to resist the layout and search journey… “changes” made available by their conversational AI (subtext: SGE stopped really being generative shortly after it was rolled out and they encountered legal issues/claims/liability because of it) They’re on their way to overmonetizing their results (a big reason why they, internally, really need to stop marking ads as such). THIS is the chance an engine like Perplexity may take to gain market share… they’ll be seen as nimble/clean and “not evil” as Google used to be. (Am watching what happens w Perplexity’s recently announced plan to have ads.)

9

Scammers love targeting P2P marketplaces. How can you catch them before they wreak havoc on your platform? Experts from Reverb, @Vinted, and Incognia share in our webinar recording the challenges of buyer/seller verification and share best practices for verifying users without too much friction. Topics will include: The importance of verification – what does verification look like? Types of fraud – phishing, counterfeiting, drop shipping, opportunistic frauds, creating accounts/account takeover or farming – automatic registration Selecting the right verification methods from signals? What are some challenges to effective verification? Best practices – introducing friction and measuring its effect on users/business growth & having a culture of being data-driven You can watch it here - https://lnkd.in/eiBBHtMv Come get insights to help you improve your verification processes in 2024! #fraudprevention #frauddetection #marketplaces #data #biometrics #UX

3