{"id":221319,"date":"2022-04-25T04:42:11","date_gmt":"2022-04-25T02:42:11","guid":{"rendered":"https:\/\/blog.lewagon.com\/non-classifiee\/structure-data-scale-up-payfit\/"},"modified":"2023-04-24T00:48:24","modified_gmt":"2023-04-23T22:48:24","slug":"structure-data-scale-up-payfit","status":"publish","type":"post","link":"https:\/\/blog.lewagon.com\/fr\/skills\/structure-data-scale-up-payfit\/","title":{"rendered":"Structure d’une stack data en scale-up + exemple de Payfit"},"content":{"rendered":"","protected":false},"excerpt":{"rendered":"

Certaines probl\u00e9matiques data apparaissent quand l’entreprise grandit. Julie, data engineer chez 360Learning, d\u00e9crit la structure de la stack data d’une scale-up.<\/p>\n","protected":false},"author":1,"featured_media":221320,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"inline_featured_image":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[142],"type_de_post":[],"post-format":[],"class_list":["post-221319","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-skills","tag-data-fr"],"acf":[],"acfml_non_translatable_locations":null,"_wp_attached_file":null,"_wp_attachment_metadata":null,"wpml_media_processed":null,"_elementor_inline_svg":null,"_elementor_source_image_hash":null,"_wp_attachment_image_alt":null,"_msc":null,"_elementor_is_screenshot":null,"_wp_attachment_context":null,"_oembed_4f1dc266c1e5b199d2e3668569863d35":null,"_wp_attachment_backup_sizes":null,"description_courte_du_cours":null,"_description_courte_du_cours":null,"sujets":null,"_sujets":null,"niveau":null,"_niveau":null,"duree":null,"_duree":null,"lien_externe_du_cours":null,"_lien_externe_du_cours":null,"_g_feedback_shortcode_8e54bebc0966bf7f9290fee5ef5e57c8a8c597c6":null,"_g_feedback_shortcode_atts_8e54bebc0966bf7f9290fee5ef5e57c8a8c597c6":null,"elementor_font_files":null,"elementor_font_face":null,"_elementor_location":null,"_elementor_priority":null,"_elementor_code":null,"_elementor_extra_options":null,"date_de_levent":null,"_date_de_levent":null,"heure_de_debut_de_levent":null,"_heure_de_debut_de_levent":null,"heure_de_fin_de_levent":null,"_heure_de_fin_de_levent":null,"time_zone":null,"_time_zone":null,"description_de_levent":null,"_description_de_levent":null,"format_du_cours":null,"_format_du_cours":null,"pre-requis":null,"_pre-requis":null,"langue_de_levent":null,"_langue_de_levent":null,"lien_externe_de_levent":null,"_lien_externe_de_levent":null,"_yoast_wpseo_primary_tag_events":null,"lieu_de_levent":null,"_lieu_de_levent":null,"_menu_item_type":null,"_menu_item_menu_item_parent":null,"_menu_item_object_id":null,"_menu_item_object":null,"_menu_item_target":null,"_menu_item_classes":null,"_menu_item_xfn":null,"_menu_item_url":null,"_omapi_type":null,"_omapi_ids":null,"_omapi_enabled":null,"_yoast_wpseo_content_score":null,"_pll_strings_translations":null,"_yoast_wpseo_primary_post_tag":"","_yoast_wpseo_primary_type_de_post":"","_yoast_wpseo_primary_post-format":"","_oembed_e607e81f61090d85eabdabc743a27470":null,"_oembed_196c62ccee9fda191cd6fa5f568ddc05":null,"_oembed_6b5f28c0de48cbf3a2ad766be5eb3b30":null,"_oembed_082c4292f4cd0dedc40ae746a899501c":null,"om_disable_all_campaigns":null,"wpil_sync_report3":"1","_yoast_wpseo_focuskw":null,"_yoast_wpseo_linkdex":null,"_yoast_wpseo_opengraph-title":"Structure d'une stack data en scale-up + exemple de Payfit","_yoast_wpseo_opengraph-description":"Certaines probl\u00e9matiques data apparaissent quand l'entreprise grandit. Julie, data engineer chez 360Learning, d\u00e9crit la structure de la stack data d'une scale-up.","_yoast_wpseo_twitter-title":"Structure d'une stack data en scale-up + exemple de Payfit","_yoast_wpseo_twitter-description":"Certaines probl\u00e9matiques data apparaissent quand l'entreprise grandit. Julie, data engineer chez 360Learning, d\u00e9crit la structure de la stack data d'une scale-up.","wpil_links_inbound_internal_count":"0","wpil_links_inbound_internal_count_data":[],"wpil_links_outbound_internal_count":"0","wpil_links_outbound_internal_count_data":[],"wpil_links_outbound_external_count":"1","wpil_links_outbound_external_count_data":[{"__PHP_Incomplete_Class_Name":"Wpil_Model_Link","link_id":0,"url":"https:\/\/www.emil.school\/programmes\/marketing-and-data-automation","host":"emil.school","internal":false,"post":null,"anchor":"Le programme se trouve ici","added_by_plugin":false,"location":"content"}],"wpil_sync_report2_time":"2023-08-11T09:31:38+00:00","gnpub_modified_count":null,"_encloseme":"1","_jetpack_memberships_contains_paid_content":null,"_oembed_df252bc3f21e94e550790f0a26013d33":null,"_oembed_7c25793225e5877909b9566a95c01f07":null,"_oembed_time_7c25793225e5877909b9566a95c01f07":null,"_oembed_c035311b2d37111892bc6336ad75c1a7":null,"_oembed_time_c035311b2d37111892bc6336ad75c1a7":null,"_yoast_wpseo_focuskw_text_input":"","_yoast_wpseo_opengraph-image":"https:\/\/blog.lewagon.com\/wp-content\/uploads\/2023\/03\/618a40dc8c5e2713e0f5cb68_pexels-lukas-669610-scaled.jpg","_yoast_wpseo_twitter-image":"https:\/\/blog.lewagon.com\/wp-content\/uploads\/2023\/03\/618a40dc8c5e2713e0f5cb68_pexels-lukas-669610-scaled.jpg","_bnitk_mfd_meta-robots-noindex":null,"_yoast_wpseo_meta-robots-noindex":null,"_bnitk_mfd_meta-robots-nofollow":null,"_yoast_wpseo_meta-robots-nofollow":null,"_oembed_9b38dff8e9475b390d223d1a13735d6e":null,"_oembed_09b03dce1da57707d8b400d6ead6cfea":null,"_oembed_time_09b03dce1da57707d8b400d6ead6cfea":null,"_oembed_761875ecd5d7c814e779a8a1669e3078":null,"_oembed_time_761875ecd5d7c814e779a8a1669e3078":null,"_oembed_60128998417fdcbaac97a635f5785efd":null,"_oembed_time_60128998417fdcbaac97a635f5785efd":null,"_oembed_bbc2c908fc7e611d82dce88119dc4350":null,"_oembed_time_bbc2c908fc7e611d82dce88119dc4350":null,"_oembed_5ada75a4373349f7b136423290505b3a":null,"_oembed_time_5ada75a4373349f7b136423290505b3a":null,"_oembed_086373a651e90a4eb1279fa1f945472c":null,"_oembed_time_086373a651e90a4eb1279fa1f945472c":null,"_wpml_post_translation_editor_native":null,"_oembed_5e24c29ce369f6b80c9eeec04f14665b":null,"_oembed_time_5e24c29ce369f6b80c9eeec04f14665b":null,"_oembed_8c3e7085f10b7c0fc5520e72277d797b":null,"_oembed_time_8c3e7085f10b7c0fc5520e72277d797b":null,"_oembed_1489782c3c8ef26605b6876b88f7bdda":null,"_oembed_time_1489782c3c8ef26605b6876b88f7bdda":null,"_oembed_0a3d659fc8de333aa903ab0b56018d23":null,"_oembed_time_0a3d659fc8de333aa903ab0b56018d23":null,"_oembed_d1585e059963c7dbe1909a198f64637b":null,"_oembed_da273c5e4d4993b38963204fa3018b68":null,"_oembed_time_da273c5e4d4993b38963204fa3018b68":null,"_oembed_a5e2c70342e4f17e48253b0a60b76258":null,"_oembed_time_a5e2c70342e4f17e48253b0a60b76258":null,"_oembed_83c8f84c43191797360817ecdc633650":null,"_oembed_time_83c8f84c43191797360817ecdc633650":null,"_oembed_e105b3463a1e23321ff3d1a9daf4ce77":null,"_oembed_1692f2329443bc1b18fc320e0d77cb4e":null,"_yoast_wpseo_schema_article_type":null,"_oembed_02bf1d274545a46fdad44828db802e9b":null,"_oembed_483efee027379bb1099137e135e60619":null,"_oembed_time_483efee027379bb1099137e135e60619":null,"_oembed_d9e3a2e55472c308a6d4f95f9859d57c":null,"_oembed_time_d9e3a2e55472c308a6d4f95f9859d57c":null,"_oembed_14a773041d4083fdc540105ccc635e1f":null,"_oembed_time_14a773041d4083fdc540105ccc635e1f":null,"_oembed_fd2c0295b311b878f43def139f693291":null,"_oembed_time_fd2c0295b311b878f43def139f693291":null,"_oembed_9e88339a75e081aa4aa002ac43b5917e":null,"_oembed_time_9e88339a75e081aa4aa002ac43b5917e":null,"_oembed_e58b2264fd34411178c5aead6e76fc0e":null,"_oembed_time_e58b2264fd34411178c5aead6e76fc0e":null,"_oembed_bddf29494cd341c273d2f25802785f3a":null,"_oembed_time_bddf29494cd341c273d2f25802785f3a":null,"_oembed_9fdc7ffbd7d1c456cb57323136bcf86d":null,"_oembed_time_9fdc7ffbd7d1c456cb57323136bcf86d":null,"_oembed_6882adf1f7ba34c40d8abda49c078862":null,"_oembed_time_6882adf1f7ba34c40d8abda49c078862":null,"_oembed_42ae0b6a999f98ed81c86edf9b80d360":null,"_oembed_time_42ae0b6a999f98ed81c86edf9b80d360":null,"_oembed_f4dcff6ba8f84f09cade790420ba4d80":null,"_oembed_time_f4dcff6ba8f84f09cade790420ba4d80":null,"_oembed_25071af9d5ecd73a9356fe5704bba4d8":null,"_oembed_time_25071af9d5ecd73a9356fe5704bba4d8":null,"_oembed_a86728f5efc0ca0642006bbb96f843cc":null,"_oembed_time_a86728f5efc0ca0642006bbb96f843cc":null,"_oembed_d7b5055b785ef8a0cf22c87d024bf2a4":null,"_oembed_time_d7b5055b785ef8a0cf22c87d024bf2a4":null,"_oembed_6a73e101cae7c3452b00a72cf0fc9b22":null,"_oembed_time_6a73e101cae7c3452b00a72cf0fc9b22":null,"_oembed_0aef478f0830cc2e857b9a150a7f763a":null,"_oembed_time_0aef478f0830cc2e857b9a150a7f763a":null,"_oembed_d1a3bdcdf10e4859560b01d910ad1120":null,"_oembed_time_d1a3bdcdf10e4859560b01d910ad1120":null,"_oembed_05ef2af36a31586761b7311d393d1df4":null,"_oembed_time_05ef2af36a31586761b7311d393d1df4":null,"_oembed_ed00fe1690bb2e86f1eb06acfe8a13e7":null,"_oembed_time_ed00fe1690bb2e86f1eb06acfe8a13e7":null,"_oembed_59ffcbf71f633752ca4f26b3b4a544c9":null,"_oembed_time_59ffcbf71f633752ca4f26b3b4a544c9":null,"_oembed_a7b688611478922544f14672fda18461":null,"_oembed_time_a7b688611478922544f14672fda18461":null,"_oembed_266f9a151e3a09eeee918af2d28fd24f":null,"_oembed_time_266f9a151e3a09eeee918af2d28fd24f":null,"_oembed_9f52726dd229470e47d40ff0c8aeb5f8":null,"_oembed_time_9f52726dd229470e47d40ff0c8aeb5f8":null,"_oembed_5547f713058fed19783e0533ce393a31":null,"_oembed_time_5547f713058fed19783e0533ce393a31":null,"_oembed_68ff8dd20e77e109ad084c2252ca040d":null,"_oembed_time_68ff8dd20e77e109ad084c2252ca040d":null,"_oembed_819f09a569080db91bf5dadda0a30d9f":null,"_oembed_time_819f09a569080db91bf5dadda0a30d9f":null,"_oembed_4fc081fbdf0afd05a624ba852a63e133":null,"_oembed_time_4fc081fbdf0afd05a624ba852a63e133":null,"_oembed_d66baf1301b99db8b19a0924319cf03e":null,"_oembed_time_d66baf1301b99db8b19a0924319cf03e":null,"_oembed_14a5f80e3eef4903e482b0ef3799659e":null,"_oembed_time_14a5f80e3eef4903e482b0ef3799659e":null,"_oembed_c24c715adb3853482526a794a5f3c3bb":null,"_oembed_time_c24c715adb3853482526a794a5f3c3bb":null,"_oembed_9c493e098be2852715cc990b6e6243f3":null,"_oembed_time_9c493e098be2852715cc990b6e6243f3":null,"_oembed_b60b7e3028d326711916099f54e986fa":null,"_oembed_time_b60b7e3028d326711916099f54e986fa":null,"_oembed_5283e19fcec44e632de8d436d46ec2f7":null,"_oembed_time_5283e19fcec44e632de8d436d46ec2f7":null,"_oembed_057692cb4479cf5c77363896a28db44f":null,"_oembed_time_057692cb4479cf5c77363896a28db44f":null,"_oembed_e8cb0f41d141fe5eb9d69fc32a98cd72":null,"_oembed_time_e8cb0f41d141fe5eb9d69fc32a98cd72":null,"_oembed_75dcbca80ec1829a77578390fae8fa56":null,"_oembed_time_75dcbca80ec1829a77578390fae8fa56":null,"_oembed_e49318dfeabd5f16d7967e2fcc746fbd":null,"_oembed_time_e49318dfeabd5f16d7967e2fcc746fbd":null,"_oembed_76116b03fe5a013341210d278021184b":null,"_oembed_time_76116b03fe5a013341210d278021184b":null,"_oembed_6c81e1c9b634b5066fb37aaccfd71053":null,"_oembed_time_6c81e1c9b634b5066fb37aaccfd71053":null,"_oembed_abbc1aa97e554f334c2ce25f0d883702":null,"_oembed_time_abbc1aa97e554f334c2ce25f0d883702":null,"_oembed_ba421363b36dd97c087665d5025396cf":null,"_oembed_time_ba421363b36dd97c087665d5025396cf":null,"_oembed_a98f236eb48eb9cd4ede7f319e9cfbc7":null,"_oembed_time_a98f236eb48eb9cd4ede7f319e9cfbc7":null,"_oembed_d61fec7d73397687995fec8d7bd9852f":null,"_oembed_time_d61fec7d73397687995fec8d7bd9852f":null,"_oembed_2e9e46a0176fe90e49cc5fff32ac6912":null,"_oembed_time_2e9e46a0176fe90e49cc5fff32ac6912":null,"_oembed_8a304f49b3aa1c3eec8d0b0577e05e3d":null,"_oembed_time_8a304f49b3aa1c3eec8d0b0577e05e3d":null,"_oembed_ad30824ca1e37cffdad2dbce46c4998a":null,"_oembed_time_ad30824ca1e37cffdad2dbce46c4998a":null,"_oembed_6c3dacc8273d4877c65378c1f324c9cc":null,"_oembed_time_6c3dacc8273d4877c65378c1f324c9cc":null,"_oembed_1d0e7d3c6d826ca1e56a24218abba6f2":null,"_oembed_time_1d0e7d3c6d826ca1e56a24218abba6f2":null,"_oembed_50929cd1300f7cdee57e6e5451afb186":null,"_oembed_time_50929cd1300f7cdee57e6e5451afb186":null,"_oembed_72dc139d079681a2c3d759a86af48ccc":null,"_oembed_time_72dc139d079681a2c3d759a86af48ccc":null,"_oembed_d359fdb1fa5da9e62dd3505937f46fb6":null,"_oembed_time_d359fdb1fa5da9e62dd3505937f46fb6":null,"_oembed_589d6065ccac39334963be5bd8656631":null,"_oembed_time_589d6065ccac39334963be5bd8656631":null,"_oembed_715be02b6ebadc1be6dd638a57d01eb7":null,"_oembed_time_715be02b6ebadc1be6dd638a57d01eb7":null,"_oembed_f5139a9df267d3aa532f5b56409c90b0":null,"_oembed_time_f5139a9df267d3aa532f5b56409c90b0":null,"_oembed_3d9a26be229b27005ca74596fb62cc08":null,"_oembed_time_3d9a26be229b27005ca74596fb62cc08":null,"_oembed_a7d6f0ce79de685fc6f7f4d447e15468":null,"_oembed_time_a7d6f0ce79de685fc6f7f4d447e15468":null,"_oembed_20849479e832b5312335e3086c5f972b":null,"_oembed_time_20849479e832b5312335e3086c5f972b":null,"_oembed_2be4d392f66846deb7624c1bf3bc293e":null,"_oembed_time_2be4d392f66846deb7624c1bf3bc293e":null,"_oembed_9bcf546ec292041d23df28a719840688":null,"_oembed_time_9bcf546ec292041d23df28a719840688":null,"_oembed_0858ffd7404bcd7d82896f942481fe63":null,"_oembed_time_0858ffd7404bcd7d82896f942481fe63":null,"_oembed_4a2dd0ecf082a7c7d82731c06a644799":null,"_oembed_time_4a2dd0ecf082a7c7d82731c06a644799":null,"_oembed_8a1e46bc730cb21c1085d157ae6fa8b6":null,"_oembed_time_8a1e46bc730cb21c1085d157ae6fa8b6":null,"_oembed_cae7deb931d75f809e459e912e71429a":null,"_oembed_time_cae7deb931d75f809e459e912e71429a":null,"_oembed_6fcb4f8ab72abccda75674f71a946f97":null,"_oembed_time_6fcb4f8ab72abccda75674f71a946f97":null,"_oembed_7f0bfd584f844117a64fe73d83b43245":null,"_oembed_time_7f0bfd584f844117a64fe73d83b43245":null,"_oembed_d0aea105cffdf6e004979cf1e4f51e96":null,"_oembed_time_d0aea105cffdf6e004979cf1e4f51e96":null,"_oembed_e53b85b8766b3dc91338f6c1b95a2541":null,"_oembed_time_e53b85b8766b3dc91338f6c1b95a2541":null,"_oembed_48832be87b763046d232c1f2b5ac5fb3":null,"_oembed_time_48832be87b763046d232c1f2b5ac5fb3":null,"_oembed_32a9a34fcf91c466be9a08e1dba3c84e":null,"_oembed_time_32a9a34fcf91c466be9a08e1dba3c84e":null,"_oembed_fb50b2dc07acfa2418f00cdc2ad4c4c2":null,"_oembed_time_fb50b2dc07acfa2418f00cdc2ad4c4c2":null,"_oembed_3089e1d62f944a73b38c996e49abd2fc":null,"_oembed_time_3089e1d62f944a73b38c996e49abd2fc":null,"_oembed_1ecc99e75cf5d001c8da4f4d600703a8":null,"_oembed_time_1ecc99e75cf5d001c8da4f4d600703a8":null,"_oembed_b9c03ca24b382c17966861d251dc3b8d":null,"_oembed_time_b9c03ca24b382c17966861d251dc3b8d":null,"_wp_trash_meta_status":null,"_wp_trash_meta_time":null,"_wp_desired_post_slug":null,"_oembed_031755ba22f2b639d88c78254a99ff39":null,"_oembed_time_031755ba22f2b639d88c78254a99ff39":null,"description_de_la_ressource":null,"_description_de_la_ressource":null,"titre_du_fichier":null,"_titre_du_fichier":null,"_edit_last":"1","_edit_lock":"1682289967:1","_wp_old_slug":null,"_top_nav_excluded":null,"_cms_nav_minihome":null,"inline_featured_image":null,"_yoast_wpseo_estimated-reading-time-minutes":"0","_yoast_wpseo_wordproof_timestamp":"","_wpml_media_duplicate":"1","_thumbnail_id":"221320","_wpml_media_featured":"1","_wpml_word_count":"{\"total\":1461,\"to_translate\":{\"en\":1461,\"de\":1461,\"ja\":1461,\"pt-br\":1461,\"es\":1461}}","_last_translation_edit_mode":null,"_last_editor_used_jetpack":"classic-editor","_elementor_edit_mode":null,"_elementor_template_type":null,"_wp_page_template":"default","_elementor_page_settings":null,"_elementor_data":null,"_elementor_page_assets":[],"_elementor_version":null,"_elementor_pro_version":null,"_elementor_elements_default_values":null,"_elementor_css":null,"_elementor_source":null,"_elementor_conditions":null,"_wpml_location_migration_done":null,"_elementor_controls_usage":null,"_elementor_screenshot_failed":null,"_elementor_screenshot":null,"_elementor_popup_display_settings":null,"_yoast_wpseo_primary_category":"","_wp_old_date":null,"_elementor_template_widget_type":null,"_icl_lang_duplicate_of":null,"_yoast_wpseo_metadesc":"Certaines probl\u00e9matiques data apparaissent quand l'entreprise grandit. Julie, data engineer chez 360Learning, d\u00e9crit la structure de la stack data d'une scale-up.","_yoast_wpseo_title":"Structure d'une stack data en scale-up + exemple de Payfit","contenu_de_larticle":"On vous en avait d\u00e9j\u00e0 parl\u00e9 dans cet article, le choix de vos outils data d\u00e9pend de la taille de votre entreprise et surtout de votre niveau de maturit\u00e9 sur les sujets data.\r\n\r\nJ'aborderai donc dans ce contenu les probl\u00e9matiques, les challenges et les m\u00e9thodes de travail li\u00e9s au niveau de maturit\u00e9 de la scale-up<\/em> sp\u00e9cifiquement.\r\n\r\nCommen\u00e7ons par poser les bases : qu'est-ce qu'une scale-up<\/em> ?\r\n\r\nIl s'agit d'un statut interm\u00e9diaire entre la start-up et la licorne. G\u00e9n\u00e9ralement, l'objectif de ce type d'entreprise n'est plus d'it\u00e9rer pour rechercher un business model viable, mais plut\u00f4t de suivre ce business model pour g\u00e9n\u00e9rer de la croissance.\r\n\r\nC'est le cas de Payfit, de 360Learning ou encore de Tiller, dont je partagerai les exemples plus loin.\r\n

La structure d'une stack data id\u00e9ale en scale-up<\/h2>\r\nVoici un exemple de structure de stack data classique, inspir\u00e9e par l'exemple de SumUp - ex-Tiller Systems. L'avantage, c'est que les outils data utilis\u00e9s varient rarement d'une entreprise \u00e0 l'autre. Vous pouvez donc consid\u00e9rer que ce sch\u00e9ma repr\u00e9sente la stack data de 80% des scale-up.\r\n\r\n\u200d\r\n
\r\n
<\/div><\/figure>\r\n\u200d\r\n\r\nSi l'on regarde ce sch\u00e9ma, on y voit diff\u00e9rentes \u00e9tapes. Pour chacune d'elle, vous serez amen\u00e9 \u00e0 choisir les outils adapt\u00e9s.\r\n
    \r\n \t
  1. Connectez vos data sources<\/em><\/strong> : ce sont les outils op\u00e9rationnels depuis lesquels va \u00eatre collect\u00e9e la data. Par exemple, si vous utilisez Salesforce comme CRM, Salesforce sera une de vos data sources<\/em>. Si vous utilisez Facebook Ads pour du marketing, m\u00eame chose...<\/li>\r\n \t
  2. Extrayez et chargez les donn\u00e9es<\/strong> :<\/em> ici, vous collectez la donn\u00e9e depuis vos data sources<\/em> et l'envoyez dans un espace de stockage unique. Vous pouvez le faire soit en utilisant du code, plus pr\u00e9cis\u00e9ment le langage Python, soit avec des logiciels cl\u00e9s en main que l'on appelle ETL<\/strong> comme Fivetran ou Stitch.<\/li>\r\n \t
  3. Stockez vos donn\u00e9es<\/strong> : l'id\u00e9e est de n'avoir qu'une source de donn\u00e9es unique. C'est en fait un lieu unique o\u00f9 vous centralisez l'ensemble de votre data : c'est ce qu'on appelle un data warehouse<\/em>, \u00e0 l'image de BigQuery ou de Snowflake.<\/li>\r\n \t
  4. Transformez les donn\u00e9es<\/strong> **: lors de cette \u00e9tape, vous formatez les donn\u00e9es pour les rendre lisibles et utilisables<\/em>par les \u00e9quipes m\u00e9tiers. **Des outils comme DBT vous permettent de le faire simplement, m\u00eame si vous pouvez aussi utiliser Python ou votre ETL.<\/li>\r\n \t
  5. Analysez la donn\u00e9e<\/strong> en l'envoyant dans des outils de business intelligence<\/em> utilis\u00e9s pour le reporting, tels que Tableau ou Salesforce. Les \u00e9quipes m\u00e9tiers peuvent ainsi facilement prendre les bonnes d\u00e9cisions, fond\u00e9es sur la data et non pas sur de simples ressentis.<\/li>\r\n<\/ol>\r\n

    Les probl\u00e9matiques de structure d'une stack data d'une scale-up<\/h2>\r\n

    Une base de donn\u00e9es scalable<\/h3>\r\nPlus l'entreprise grandit, plus les utilisateurs et les donn\u00e9es sont nombreux, et donc plus le nombre de requ\u00eates est important. Il est donc essentiel que la base de donn\u00e9es puisse supporter la charge.\r\n\r\nL'une des techniques utilis\u00e9es par Julie consiste \u00e0 cr\u00e9er des noeuds : les calculs interm\u00e9diaires sont ainsi trait\u00e9s s\u00e9par\u00e9ment sur chaque noeud, puis les calculs finaux sont rassembl\u00e9s sur le noeud central. C'est notamment ce que permet de faire BigQuery.\r\n\r\nVous pouvez ainsi travailler en parall\u00e8le sur de nombreuses requ\u00eates et gagner ainsi du temps.\r\n

    Vers l'ELT plut\u00f4t que l'ETL<\/h3>\r\nComme je le disais plus haut, l'ETL est un logiciel qui permet de r\u00e9colter, de transformer et de stocker les donn\u00e9es de l'entreprise. Dans le cas de l'utilisation d'un ETL, je formate ma donn\u00e9e entre la data source<\/em> et le data warehouse<\/em>. D\u00e8s que j'ai besoin d'un nouveau type d'analyse, j'ai donc besoin d'aller de nouveau chercher la donn\u00e9e dans ma data source<\/em> puis de la transformer.\r\n\r\nDe plus en plus d'entreprises envisagent de passer \u00e0 un mod\u00e8le ELT, c'est-\u00e0-dire que je formate ma donn\u00e9e une fois qu'elle a \u00e9t\u00e9 stock\u00e9e dans le data warehouse<\/em>. Ainsi plut\u00f4t que de r\u00e9aliser l'\u00e9tape de transformation en amont de l'\u00e9tape de chargement, elle est r\u00e9alis\u00e9e apr\u00e8s. Cela permet de d\u00e9corr\u00e9ler les deux actions et apporte une plus grande flexibilit\u00e9.\r\n\r\nAinsi si l'on souhaite de nouveau apporter une transformation, je n'ai pas besoin de charger de nouveau l'ensemble des donn\u00e9es et gagne donc un temps pr\u00e9cieux.\r\n

    Vers les ETL manag\u00e9s<\/h3>\r\nG\u00e9n\u00e9ralement, l'entreprise peut g\u00e9rer toute sa collecte et sa transformation de donn\u00e9es en autonomie, en utilisant le langage Python, assorti de l'outil Airflow. Cet outil permet notamment de programmer des workflows et donc d'automatiser certains traitements de la donn\u00e9e.\r\n\r\nN\u00e9anmoins, quand les \u00e9quipes grandissent et demandent de plus en plus d'analyses, cela n\u00e9cessite une bande passante importante des data engineers.\r\n\r\nJe recommande donc d'utiliser des ETL manag\u00e9s, \u00e0 l'image de Fivetran. Ce sont des logiciels permettant de simplifier et d'automatiser le traitement de la data. Par exemple, si un collaborateur vous demande une donn\u00e9e sp\u00e9cifique, vous avez simplement \u00e0 cocher une case, plut\u00f4t que de cr\u00e9er une nouvelle requ\u00eate\r\n

    Vers l'incr\u00e9mental plut\u00f4t que le full refresh<\/h3>\r\nDans la plupart des entreprises, les requ\u00eates quotidiennes chargent l'ensemble des donn\u00e9es, ce qui demande une bande passante significative ainsi qu'un d\u00e9lai important. C'est ce qu'on appelle le full refresh.\r\n\r\nEn scale-up, on adopte de plus le chargement incr\u00e9mental. Ce dernier permet d'actualiser uniquement les donn\u00e9es modifi\u00e9es depuis le dernier chargement, ce qui repr\u00e9sente un gain de temps consid\u00e9rable.\r\n\r\nPour gagner du temps, les requ\u00eates ne chargent pas l'ensemble des donn\u00e9es chaque matin, mais uniquement les donn\u00e9es modifi\u00e9es depuis le dernier chargement.\r\n

    Collaborer sur une stack data en scale-up<\/h2>\r\nPlus le nombre de personnes avec lesquelles les data engineers sont amen\u00e9s \u00e0 collaborer augmente, plus la formalisation des process et l'automatisation prend de l'ampleur.\r\n\r\nAinsi, la gestion des acc\u00e8s \u00e0 la base de donn\u00e9es peut se transformer en une t\u00e2che tr\u00e8s chronophage si de nouveaux salari\u00e9s arrivent chaque semaine. Cette action peut donc \u00eatre automatis\u00e9e.\r\n\r\nDe m\u00eame, tout le monde n'aura pas le m\u00eame niveau d'acc\u00e8s aux donn\u00e9es. Par exemple, les \u00e9quipes sales ne voudront pas que d'autres d\u00e9partements aient acc\u00e8s \u00e0 leur variable. Il faut donc d\u00e9finir des r\u00e8gles de visualisation claires.\r\n\r\nPour que l'ensemble des parties prenantes puisse avoir une vision sur ce qui existe et qui est responsable de telle ou telle donn\u00e9e ou requ\u00eate, Julie conseille de cr\u00e9er un data catalog<\/em> - un peu comme un content manager pourrait cr\u00e9er un catalogue de contenus.\r\n\r\nEnfin, pour que les data analysts et data scientists puissent travailler en toute autonomie, les data engineers doivent pr\u00e9parer un framework optimal.\r\n

    De nouveaux challenges quant \u00e0 la structure d'une stack data mature<\/h2>\r\nAvec l'atteinte de la maturit\u00e9 data d'une scale-up, de nouvelles probl\u00e9matiques se posent.\r\n