WEBVTT 00:00:00.000 --> 00:00:07.250 A client sees a skin score on their phone, brings a hairstyle reference to a salon, or opens a message after an appointment. 00:00:07.250 --> 00:00:10.733 Beauty technology is starting to shape all three moments. 00:00:10.733 --> 00:00:16.562 This week, four stories about what happens between the first impression and the next useful step. 00:00:16.562 --> 00:00:19.119 I'm Mike, and this is Beauty Tech Weekly. 00:00:19.119 --> 00:00:23.030 We have richer messages from Peter Thomas Roth, research that 00:00:23.030 --> 00:00:30.551 reconstructs hair in three dimensions, skin analysis inside Qoo10, and a machine that makes personalised cosmetics. 00:00:30.551 --> 00:00:36.493 My question throughout: what does each of these actually help a client or a beauty business do? 00:00:43.493 --> 00:00:45.773 First, the message after the visit. 00:00:45.773 --> 00:00:51.931 On November tenth, Peter Thomas Roth and Listrak shared results from their use of RCS messaging. 00:00:51.931 --> 00:00:57.442 Think of a branded message with product images and buttons inside the phone's messaging app. 00:00:57.442 --> 00:01:02.771 The news is this retailer's reported experience, rather than the invention of a new messaging channel. 00:01:02.771 --> 00:01:09.251 Listrak reports an average forty percent higher conversion rate in fifty-fifty tests against SMS. 00:01:09.251 --> 00:01:12.877 That is a relative uplift, not forty percentage points. 00:01:12.877 --> 00:01:17.814 The release does not provide the absolute conversion rates or sample sizes. 00:01:17.814 --> 00:01:24.140 It is an interesting vendor-reported result, but it is not a forecast for a salon's own campaigns. 00:01:24.140 --> 00:01:28.462 For me, the useful question is what the richer message makes easier. 00:01:28.462 --> 00:01:35.439 A customer should be able to understand the offer and choose a next step without working through a wall of text. 00:01:35.439 --> 00:01:38.473 A beautiful card still needs a reason to arrive. 00:01:38.473 --> 00:01:43.705 A reminder, a relevant answer and a generic promotion are different experiences. 00:01:43.705 --> 00:01:47.027 In a salon, the question often comes after the visit. 00:01:47.027 --> 00:01:51.782 A client opens a product at home and remembers something they forgot to ask. 00:01:51.782 --> 00:01:56.538 At AI Beauty Bot, we help salons stay in touch with clients in those moments. 00:01:56.538 --> 00:02:03.558 That is the connection I care about here: making it easier to continue the conversation between appointments. 00:02:03.558 --> 00:02:10.413 If I were evaluating this for a business, I would start with one specific customer moment and one useful action. 00:02:10.413 --> 00:02:15.574 Then I would compare it with the current message, using the same audience and offer. 00:02:15.574 --> 00:02:18.154 I would also look at opt-outs and replies. 00:02:18.154 --> 00:02:23.020 More purchases alone would not tell me whether the communication was welcome. 00:02:26.020 --> 00:02:31.612 Next, a different kind of communication problem: showing someone the haircut you have in mind. 00:02:31.612 --> 00:02:36.155 SAFE-Hair is a research project whose paper was submitted on November ninth. 00:02:36.155 --> 00:02:40.936 It reconstructs editable three-dimensional hair strands from a single portrait. 00:02:40.936 --> 00:02:47.153 The authors show the input image alongside reconstructed views from the front, side and back. 00:02:47.153 --> 00:02:49.828 There is an important distinction in those images. 00:02:50.028 --> 00:02:52.333 The camera only captured one view. 00:02:52.533 --> 00:02:55.644 Parts of the hairstyle that were hidden have to be inferred. 00:02:55.644 --> 00:02:59.377 A convincing view of the back does not mean the system observed the back. 00:02:59.377 --> 00:03:04.106 The result is a reconstruction, rather than a complete measurement of that person's hair. 00:03:04.106 --> 00:03:08.151 The work is aimed at editable digital hair and graphics workflows. 00:03:08.151 --> 00:03:12.269 The public repository lists code and model releases as future work. 00:03:12.269 --> 00:03:17.615 So I would treat this as a research demonstration, not a salon tool you can install today. 00:03:17.615 --> 00:03:22.888 It also does not establish whether a real client's hair can achieve the pictured style. 00:03:22.888 --> 00:03:25.998 What interests me is a possible use in consultation. 00:03:25.998 --> 00:03:29.955 A single reference photo leaves plenty to discuss: the silhouette 00:03:29.955 --> 00:03:34.831 from the side, the length at the back, and which details the client actually likes. 00:03:34.831 --> 00:03:43.029 A three-dimensional reference could make that conversation more specific, provided everybody understands which parts were inferred. 00:03:43.029 --> 00:03:49.243 I would want the interface to keep the original photo visible and clearly mark the reconstructed views. 00:03:49.243 --> 00:03:54.886 Then the professional can ask better questions and explain what is realistic for this client. 00:03:54.886 --> 00:03:57.815 The value would be a clearer shared expectation. 00:03:57.815 --> 00:04:02.673 A polished simulation, on its own, would not be a promise of the finished haircut. 00:04:05.673 --> 00:04:09.304 Our third story moves the consultation into a shopping app. 00:04:09.304 --> 00:04:15.913 On November ninth, Perfect Corp announced that its skin-analysis technology was available on Qoo10 Japan. 00:04:15.913 --> 00:04:21.111 Shoppers use a phone camera and receive a report linked to product recommendations. 00:04:21.111 --> 00:04:26.910 The same announcement covers colour-contact-lens try-on, which has already been available since August. 00:04:26.910 --> 00:04:31.306 The system checks the capture conditions and reports on ten skin concerns. 00:04:31.306 --> 00:04:34.237 Those scores are presented in a shopping context. 00:04:34.237 --> 00:04:37.726 I would be careful about treating a score as a diagnosis, or a 00:04:37.726 --> 00:04:43.448 repeatable measurement as proof that a product recommendation is right for a particular person. 00:04:43.648 --> 00:04:45.962 Those are separate questions. 00:04:46.162 --> 00:04:51.131 For salons, I think the interesting scenario is a client arriving with a screenshot. 00:04:51.131 --> 00:04:56.939 They may already have a score, a product shortlist and a strong impression of what their skin needs. 00:04:56.939 --> 00:05:01.068 That gives the professional a starting point: what was photographed, 00:05:01.068 --> 00:05:05.336 under what conditions, and what does the client actually want to change? 00:05:05.336 --> 00:05:07.610 There is also a relevant piece of context. 00:05:07.610 --> 00:05:13.587 On November twelfth, Qoo10 published a notice about a security incident and personal-information leakage. 00:05:13.587 --> 00:05:18.589 That notice's existence does not establish a connection with this skin-analysis feature. 00:05:18.789 --> 00:05:21.173 I am keeping those facts separate. 00:05:21.373 --> 00:05:25.790 It does make data handling a sensible question to ask of any photo-based service. 00:05:25.790 --> 00:05:29.820 Before introducing a similar experience, I would want clear answers 00:05:29.820 --> 00:05:35.239 about what images are retained, who can access them, and how a customer can have them deleted. 00:05:35.239 --> 00:05:39.130 I would also want to understand how recommendations are selected. 00:05:39.130 --> 00:05:45.591 A useful consultation should help a person understand the result, including what the system cannot tell them. 00:05:48.591 --> 00:05:52.483 Finally, from a recommendation to a physical product. 00:05:52.483 --> 00:06:00.242 Lillycover's ENIMA Light won the Home and Professional Devices and Tools category at the Cosmoprof Asia Awards this week. 00:06:00.242 --> 00:06:06.746 This is an award story about an existing system, rather than a claim that the machine launched for the first time this week. 00:06:06.746 --> 00:06:15.293 The award description connects measurements from Lillycover's Muilli device with SkinDia software and the ENIMA Light manufacturing system. 00:06:15.293 --> 00:06:22.229 The idea is to turn a customer's skin information into a personalised formula, then make the skincare product on site. 00:06:22.229 --> 00:06:28.114 The company describes production in around four minutes, and profiles can be saved for reordering. 00:06:28.114 --> 00:06:31.664 That production time is only one part of the appointment. 00:06:31.664 --> 00:06:37.064 For a salon, I would also count the consultation, preparation, cleaning and handover. 00:06:37.064 --> 00:06:41.428 The useful number is the time and cost of delivering the whole service. 00:06:41.428 --> 00:06:46.975 I would want to see that workflow with real staff before building a service menu around it. 00:06:46.975 --> 00:06:49.065 My next questions would be practical. 00:06:49.265 --> 00:06:50.940 What do refills cost? 00:06:51.140 --> 00:06:53.361 How are ingredients stored and tracked? 00:06:53.561 --> 00:06:55.890 What training does the team need? 00:06:56.090 --> 00:06:59.682 What happens if the machine is unavailable when a client returns? 00:06:59.682 --> 00:07:01.968 An award makes this worth investigating. 00:07:01.968 --> 00:07:05.887 It does not answer the operating questions for an individual business. 00:07:05.887 --> 00:07:08.712 The part I find most interesting is continuity. 00:07:08.712 --> 00:07:14.998 If a customer likes the product, can the business explain what was made and manage the next order reliably? 00:07:14.998 --> 00:07:20.083 Personalisation should be understandable to the client and manageable for the team. 00:07:20.083 --> 00:07:25.662 The experience after the first bottle matters as much as the moment it comes out of the machine. 00:07:25.662 --> 00:07:28.902 Across these four stories, I see the same practical test. 00:07:28.902 --> 00:07:32.143 Does the technology make the next conversation clearer? 00:07:32.143 --> 00:07:34.708 Richer messages can offer a useful next step. 00:07:34.708 --> 00:07:37.746 Hair reconstruction can give people more to discuss. 00:07:37.946 --> 00:07:40.336 A skin report can open questions. 00:07:40.536 --> 00:07:43.574 On-site production can create a continuing service. 00:07:43.574 --> 00:07:46.612 Each still needs an honest explanation of its limits. 00:07:46.612 --> 00:07:49.814 Which of these would be most useful in your business: better 00:07:49.814 --> 00:07:55.834 follow-up, a clearer hairstyle reference, help interpreting a skin report, or personalised products on site? 00:07:56.034 --> 00:07:57.174 Let me know. 00:07:57.374 --> 00:07:59.615 I'm Mike, and this was Beauty Tech Weekly. 00:07:59.815 --> 00:08:01.276 See you next week.