Market Report

The Personalisation Market, 2000–2026: Platforms, Deals, AI and How E-commerce Teams Should Choose

Alexandre Suon · 2026-09-26

Personalisation software decides which products, pages, messages and offers each customer sees. In 25 years it has grown from rules and recommendation widgets into eight overlapping markets, a wave of acquisitions by payments, data and AI companies, and a shift towards AI agents that decide on their own. This report maps the vendors, the deals and the evidence, and shows e-commerce directors how to choose.

Executive summary

1. Personalisation is eight markets, not one. Experimentation platforms, search and recommendations, customer data and engagement, AI decisioning, content platforms, on-site pop-ups, feature flags and the big commerce and AI platforms all sell it. Gartner sizes the core personalisation-engine market at just $1.2B in 2024 (+26.1%), while suites such as Adobe's Digital Experience business ($5.86B revenue in FY2025) capture most of the spend.
2. The category was built by acquisition. Adobe entered by buying Omniture for about $1.8B in 2009, two years after Omniture bought the testing and targeting pioneers Offermatica and Touch Clarity. Salesforce, SAP, Twilio and Mastercard followed with deals for ExactTarget, Demandware, Evergage, Emarsys, Segment and Dynamic Yield.
3. Since 2024, specialists have been absorbed by neighbouring platforms. OpenAI bought Statsig for $1.1B, Braze bought OfferFit for $325M, Rokt bought mParticle for $300M and Datadog bought Eppo. Private equity is merging the rest: VWO and AB Tasty became Wingify, Monetate bought SiteSpect, and Klevu and Searchspring formed Athos Commerce.
4. AI is moving personalisation from rules to decisions, and from the website to AI assistants. Every major vendor launched AI agents in 2025–26, and visits to US retail sites from generative-AI assistants rose 393% year on year in Q1 2026, according to Adobe.
5. The evidence is weaker than the marketing. McKinsey says personalisation most often lifts revenue by 10–15%, but controlled studies show results depend heavily on the algorithm, and non-experimental measurement often overstates effects several times over. Every programme needs a holdout group that never sees the personalised experience.
6. Choose from the question, not the vendor. Most e-commerce teams should start with search and recommendations, a well-tested website and email, then add AI decisioning once the data and measurement are in place. Published prices range from free to $15,000 a month.

Section 1 · The segments

Eight segments sell personalisation, and most retailers buy from several of them

Ask ten vendors what personalisation is and you will get ten answers. A search vendor means ranking products for each shopper. An email platform means sending the right message at the right time. An experimentation platform means showing each segment the page variant that works best. All are right, which is why the market is hard to read and easy to overspend in.

What this shows. The segments answer different questions with different data. Segments 1 and 2 personalise the website; 3 and 4 personalise messages across channels; 5 personalises content at the source; 6 captures intent cheaply; 7 serves product teams; and 8 is where platforms and AI assistants build personalisation in, often for free. The lines are blurring: most large vendors now sell across three or more segments.

Section 2 · Market size

Published market sizes range from $1.2B to $455B, but most personalisation spend sits inside larger suites

Market estimates for personalisation disagree by a factor of several hundred, because they measure different things. Gartner counts only the revenue of personalisation-engine vendors. Other firms add services, customer data platforms or any software that uses AI to tailor something.

Company revenues are a better guide to where the money goes. The largest personalisation businesses are parts of broader suites, and the specialists are much smaller.

What this shows. Adobe's Digital Experience business alone earns about five times Gartner's entire engine market, and Salesforce's marketing and commerce clouds nearly as much. Most personalisation is therefore bought as part of a suite, an email platform or a commerce platform, not as a stand-alone engine. Among specialists, only a handful have passed $100M in annual revenue.

For e-commerce directors. Before buying a personalisation engine, check what your existing email, search, testing and commerce platforms already include. In our experience, many teams pay for the same capability more than once.

For investors. Market-size reports are not a sound basis for a personalisation investment case. Anchor valuations on disclosed revenue, net retention and the share of spend at risk of being bundled into a suite.

Section 3 · 2000–2012

Recommendation engines and testing tools created the category, and Adobe bought its way in

Modern personalisation has two roots. The first is the recommendation engine. In 2003 Amazon engineers Greg Linden, Brent Smith and Jeremy York published "Item-to-Item Collaborative Filtering", describing how Amazon recommended products similar to the ones a customer had already bought or rated, instead of searching for similar customers. In 2006 Netflix offered $1M to anyone who could improve its recommendations by 10%; the prize was awarded in September 2009 to the team BellKor's Pragmatic Chaos.

Netflix later put numbers on the value. In a 2015 paper, two of its executives wrote that recommendations influence choice for about 80% of hours streamed, and that personalisation and recommendations together "save us more than $1B per year", mostly by reducing cancellations. A widely quoted claim that 35% of Amazon's sales come from recommendations goes back to a 2013 McKinsey article that cites no source; we found no primary evidence for it.

The second root is website testing and targeting. In 2007 the analytics company Omniture bought Offermatica, an A/B and multivariate testing tool, for $65M, and Touch Clarity, a UK behavioural-targeting company, for about $50M. In September 2009 Adobe agreed to buy Omniture for about $1.8B. Their technology lives on in Adobe Target, which still anchors Adobe's personalisation offer.

A new generation of specialists appeared around the same time: Monetate (2008), Evergage and Optimizely (2010) and Dynamic Yield (around 2011). They made testing and targeting usable by marketers without developers, and set up the fight between suites and specialists that still shapes the market.

Section 4 · 2013–2021

Suites assembled personalisation stacks through a decade of $1–5B acquisitions

Between 2013 and 2021 the big software companies bought the pieces of a full personalisation stack: messaging, commerce, customer data and targeting. At the same time, investors valued specialists at up to $2B or more.

What this shows. Two waves stand out. From 2009 to 2020, software suites paid $1–5B to add marketing, commerce and customer-data capabilities. From 2024, the buyers changed: payments, data, advertising and AI companies are buying specialists, at lower prices, to add personalisation to platforms of their own.

• Salesforce bought ExactTarget for $2.5B (2013), which became Marketing Cloud, Demandware for $2.8B (2016), which became Commerce Cloud, and Evergage (2020), now Marketing Cloud Personalization.
• Adobe added Magento for $1.68B and Marketo for $4.75B in 2018, then built Real-Time CDP and Journey Optimizer on Adobe Experience Platform.
• Insight Partners bought the content-management company Episerver for $1.16B in 2018; Episerver bought Optimizely in 2020 and took its name in January 2021. Optimizely reported $400M of annual recurring revenue in May 2024.
• McDonald's bought Dynamic Yield in 2019 for a reported $300M, then sold it to Mastercard, which completed the deal in April 2022.
• Twilio bought the customer data platform Segment for about $3.2B in stock (2020), and SAP bought Emarsys the same year.
• Investors valued Algolia at $2.25B (2021) and Bloomreach at $2.2B (2022) after it bought Exponea. Braze (2021), Coveo (2021) and Klaviyo (2023, at about $9.2B) listed on public markets.

The period also brought the first backlash. In December 2019 Gartner predicted that by 2025, 80% of marketers who had invested in personalisation would abandon their efforts "due to lack of ROI, the perils of customer data management or both". Spending did not collapse: Gartner's own 2026 report says engine revenue grew 26% in 2024. But the prediction captured a real problem, which Section 7 returns to: many programmes cannot prove what they earn.

Privacy changes also began to bite. Apple's Safari blocked third-party cookies by default from March 2020, and Apple's App Tracking Transparency, introduced in April 2021, made iPhone apps ask before tracking users across other companies' apps and websites. Personalisation based on data collected by other companies became harder, and first-party data became more valuable.

Section 5 · 2022–2026

Since 2024, neighbouring platforms and private equity have absorbed the specialists

The last three years have reshaped every segment. Specialists that raised money at high valuations in 2021 have been bought by companies whose main business is elsewhere, merged by private-equity owners, or folded into larger platforms.

The customer data platform segment shows the pattern most clearly. In 2024, Gartner placed CDPs in the "trough of disillusionment" of its hype cycle, reporting that 67% of marketers had adopted one but only 17% reported high use, according to AdExchanger. The CDP Institute reported in July 2025 that acquisitions of independent CDPs were a clear trend, even as the industry returned to growth. Buyers want customer data to live in their own warehouse, and vendors are responding by becoming "composable" or by selling out. Engagement platforms, by contrast, are still raising large rounds: MoEngage completed a $280M Series F in December 2025, led in its second tranche by ChrysCapital and Dragon Funds, and says it serves more than 1,350 consumer brands.

For investors. The exits of 2022–26 went mostly to strategic buyers from adjacent markets: payments (Mastercard), advertising (Rokt), observability (Datadog), AI (OpenAI) and data movement (Fivetran). Few specialists reached a large independent exit. Underwrite personalisation assets on who would buy them, not on a future listing.

Section 6 · AI and agents

AI is moving personalisation from rules written by marketers to decisions made by agents

Personalisation has gone through four technology stages, and each still exists in the market. Rule-based targeting lets marketers define segments and choose what each one sees. Machine-learning recommendations predict products for each person. AI decisioning uses reinforcement learning to choose offers, timing and channels for each customer and learns from every response. AI agents now go further, proposing hypotheses, building variants and running campaigns with limited human input.

What this shows. The first two stages took two decades to mature; the last two went mainstream in about two years, mostly in 2025–26. Almost every vendor in this report now markets AI agents, but most agent features are less than a year old, and published evidence of their incremental value comes mainly from the vendors themselves.

From rules to decisioning

• Braze bought OfferFit, whose reinforcement-learning agents "replace the manual work of A/B testing", for $325M, and now sells it as OfferFit by Braze.
• Hightouch, whose AI Decisioning product assigns learning agents to individual customers across email, push, in-app and ads, raised $150M at a $2.75B valuation in April 2026.
• Salesforce added a Personalization Decisioning agent to Marketing Cloud Next (June 2025). Adobe made six AI agents generally available in September 2025, including Journey, Audience and Experimentation agents. Tealium launched AI Decisioning in May 2026, and Pega, whose Customer Decision Hub has sold next-best-action decisioning for years, reported $1.75B of 2025 revenue.
• MoEngage, an engagement platform, launched Merlin AI custom agents in June 2026: marketers set rules on audience, channels, content and budget, and choose whether agents act alone or wait for approval. It also opened an MCP server so that outside AI assistants such as ChatGPT and Claude can use MoEngage data.

From the website to AI assistants

A second shift is happening off the retailer's website. Shoppers increasingly ask AI assistants what to buy, and the assistants personalise the answer before the shopper arrives, if they arrive at all.

• Traffic is growing fast from a small base. Adobe reports that visits to US retail sites from generative-AI sources rose 393% year on year in Q1 2026, and that in March 2026 those visitors converted 42% better than other traffic. Adobe does not publish AI's share of total visits.
• Assistants are adding checkout. OpenAI launched Instant Checkout in ChatGPT in September 2025 with an open protocol built with Stripe, then scaled native checkout back in March 2026, according to trade press. Google launched the Universal Commerce Protocol with Shopify, Etsy, Wayfair, Target and Walmart in January 2026. Perplexity added shopping with PayPal in November 2025.
• Retail platforms are building their own. Amazon said its Rufus assistant was used by 250 million shoppers in 2025. Shopify's Winter '26 release added "agentic storefronts" that list merchants' products in ChatGPT, Perplexity and Microsoft Copilot.

For e-commerce directors. In an AI-assistant world, personalisation starts with product data: complete attributes, accurate stock and prices, and clear policies that an assistant can read. Treat product feeds and structured data as personalisation infrastructure, not an SEO chore.

For investors. If assistants personalise before the visit, on-site personalisation vendors risk losing part of their value to whoever controls the assistant. Vendors that feed product data and offers into assistants, or run the retailer's own assistant, are better placed.

Section 7 · Does it pay?

Personalisation can pay well, but the evidence is thinner than vendor claims, so every programme needs a holdout

The case for personalisation rests on a few widely quoted studies. They are useful, but most are surveys or company-reported figures. Controlled experiments tell a more cautious story.

Three lessons follow. First, personalisation is not one thing: the same budget can produce a large gain or none, depending on the algorithm, the data and the placement. Second, vendor case studies without a randomised control group are not reliable evidence of incremental value. Third, customers notice when personalisation goes wrong. A 2015 study in the Journal of Retailing found that personalised ads raised click intentions when data was collected openly, but backfired when customers felt it had been collected covertly.

Testing culture explains why. Most ideas fail when tested properly: at Microsoft only about one third of ideas improved the metrics they were designed to improve. Personalisation rules and models are ideas too, and should be tested in the same way. Amazon's own history shows the upside: its early shopping-cart recommendations were launched as a controlled experiment against a senior executive's objection, and won by a wide margin.

For e-commerce directors. Keep a permanent holdout group, typically a small random share of visitors or customers who never see personalisation, and report incremental revenue against it every month. If a vendor cannot support a holdout, you will never know what you are paying for.

For investors. Ask targets for the share of customers who measure results against a holdout, and for renewal rates among them. Programmes that cannot prove value are the first to be cut or bundled.

Section 8 · Privacy and regulation

Regulators are targeting profiling, pricing and transparency, while third-party cookies survive in Chrome

Personalisation depends on data, and the rules on data have changed several times since 2020. The picture in 2026 is less dramatic than predicted for cookies, but tighter on transparency and pricing.

Two further rules apply across the EU. Under the GDPR, people can object at any time to profiling for direct marketing, and a decision based solely on automated processing is restricted only where it has legal or similarly significant effects, which ordinary product recommendations usually do not. Under the ePrivacy Directive, storing or reading information on a visitor's device needs consent unless it is strictly necessary. In the US, about 20 states had comprehensive privacy laws in force by early 2026.

Personalised pricing is the most sensitive area. In July 2025 Delta Air Lines faced questions from US senators about AI-based fares and stated that it does not target customers with individualised prices based on personal data. In December 2025 Instacart ended item price tests after an investigation led by Consumer Reports found different shoppers seeing different prices; Instacart said the tests were randomised and did not use personal data.

For e-commerce directors. Personalise products, content and service freely, with consent; be very careful before personalising prices. Document what data each personalisation uses, keep a non-personalised option where required, and review any price variation with legal counsel.

Section 9 · Choosing a tool

E-commerce directors should choose personalisation tools from the question they need answered

Most personalisation buying mistakes start with a vendor demo rather than a business question. A team buys an enterprise engine to "personalise the experience", then runs a handful of homepage banners while the biggest opportunities, search and product recommendations, stay generic. The framework below starts from six questions e-commerce teams actually ask.

What this shows. Each question maps to a different segment and a different data source. In our experience, most online retailers need the first three rows and one of the last three. AI decisioning pays off only once the messaging programme, the data and a holdout measurement are in place.

Step 1: Match the stack to the size of your team and business

Step 2: Compare what you will actually pay

Pricing models differ widely. Search vendors charge by queries or records, engagement platforms by contacts or monthly active users, experimentation platforms by traffic, and feature-flag tools by seats, events or service connections. Published prices run from free to $15,000 a month; most enterprise vendors publish none.

What this shows. A small store can personalise search, recommendations, email and on-site messages for under $100 a month. Prices jump when you add enterprise web personalisation, unified customer data and AI decisioning, and most of those vendors only quote. Free tiers are generous in search (Algolia, Recombee), feature flags (LaunchDarkly, Statsig, GrowthBook, PostHog) and content platforms (Contentful, Contentstack), although content-platform entry plans exclude personalisation modules, which are sold separately or in enterprise plans.

Step 3: Ask ten questions before you sign

1. Which decision will this improve first? Name the page, message or ranking and the metric it should move in the next 90 days.
2. Can we keep a permanent holdout? Insist on a random group that never sees personalisation, and on reporting against it.
3. Do we already own this capability? Check your commerce platform, email tool, search and testing contracts.
4. What data does it need, and where does it live? Prefer tools that use your own first-party data and warehouse over copying data into another silo.
5. How does it handle consent and privacy? Check consent integration, EU hosting, data-processing terms and non-profiling options.
6. Will it help us in AI assistants? Ask how the vendor feeds products, prices and offers to ChatGPT, Gemini and other agents.
7. What does the AI actually decide, and can we see why? You need to explain decisions to colleagues, customers and, increasingly, regulators.
8. How is it priced as we grow? Model the cost at two and five times today's traffic, contacts or events.
9. Who will run it? Personalisation needs content, data and testing capacity; a tool without an owner produces a few banners and no results.
10. Who owns the vendor, and what is its roadmap? After a merger or buyout, ask which products will be kept and how pricing will change.

For e-commerce directors. The best first personalisation programme is usually unglamorous: better site search, product recommendations tested against a control, email flows triggered by behaviour, and a holdout to measure it all. Add one-to-one web experiences and AI decisioning once those are proven. That is how personalisation turns into a better customer experience, higher revenue and customer lifetime value.

Section 10 · The private-equity view

Investors should back data ownership, measurable decisioning and exposure to AI assistants

Four investment theses stand out in 2026, each with different risks:

1. Suite consolidation. Adobe and Salesforce dominate spend and keep adding AI agents. The opportunity is in companies that suites need to buy; the risk is being bundled away.
2. Specialist roll-ups. Private-equity owners are merging mid-sized specialists: Everstone with Wingify, PSG with Athos Commerce, and investor-backed Monetate with SiteSpect and Simon AI. Scale and cross-selling can work, but integration and product overlap are real risks.
3. AI decisioning. In our view, investors are paying premiums for AI that replaces manual testing and rules: Braze paid $325M for OfferFit, and Hightouch, which sells AI Decisioning, was valued at $2.75B. The risk is that suites and engagement platforms build the same capability.
4. Personalisation inside other platforms. Payments, advertising, observability and AI companies bought specialists in 2022–26. Assets that plug into these platforms, or into AI shopping assistants, have more buyers.

Section 11 · Implications

Personalise what customers search for and buy first, prove it with a holdout, and prepare for AI assistants

Personalisation software is abundant, consolidating and increasingly automated. The winners, for vendors and for retailers, will be those that can show incremental value and adapt to shopping that starts in AI assistants. Our recommendations:

1. Start with search, recommendations and lifecycle messages

In our experience these three touch the most revenue for most online retailers. Get them working and tested before investing in one-to-one web experiences.

2. Measure everything against a holdout

Keep a permanent random group without personalisation and report incremental revenue against it. Treat vendor case studies as hypotheses, not proof. Our essential guide to A/B testing explains how to design tests you can trust.

3. Build on first-party data you control

Collect data with consent, keep it in systems you own, and make sure every tool can read from it and export to it. This protects you from browser changes, vendor mergers and product sunsets.

4. Make your products readable by AI assistants

Complete product attributes, accurate prices and stock, and clear policies are now personalisation assets, because assistants use them to decide what to recommend. Track traffic and conversion from AI assistants separately.

5. Use the customer's voice to decide what to personalise

The best personalisation ideas come from knowing why different customers behave differently. Our essential guide to Voice of Customer shows how to find those reasons and turn them into tests.

Our view. Personalisation is a means, not the goal. The goal is a better customer experience that grows revenue and lifetime value. Teams of every size, from a two-person Shopify store to a multi-brand retailer, can get there by personalising what matters most to customers, testing it properly and scaling only what works.

FAQ

Frequently asked questions about personalisation software

Frequently asked questions

What is a personalisation engine?

A personalisation engine is software that decides in real time which content, products or offers each visitor sees, based on their behaviour and profile, and usually tests the results. Examples include Adobe Target, Dynamic Yield, Optimizely, Kameleoon and Salesforce Personalization.

What are the best personalisation tools for e-commerce in 2026?

It depends on the question. Search and recommendation platforms (Algolia, Bloomreach, Constructor, Nosto, Athos Commerce) personalise products; experimentation platforms (Optimizely, Dynamic Yield, Kameleoon, Wingify) personalise pages; engagement platforms (Klaviyo, Braze, Insider One) personalise messages. Small stores can start with Shopify's free Search & Discovery app and Klaviyo.

How big is the personalisation market?

Gartner sized the personalisation-engine market at $1.2B in 2024, up 26.1%. Broader estimates run from $3.7B to $14B or more, depending on whether they include services, customer data platforms and other software. Suite vendors such as Adobe and Salesforce capture most of the spend.

Does personalisation increase revenue?

Often, but not always. McKinsey reports a typical lift of 10–15%, but controlled experiments show results depend on the algorithm and placement, and non-experimental measurement can overstate effects several times over. Measure every programme against a holdout group.

What is the difference between personalisation and A/B testing?

A/B testing finds the version that works best for everyone on average. Personalisation shows different versions to different people. The two work together: personalisation rules and models should themselves be tested against a control.

Is personalised pricing legal?

It is heavily scrutinised. The US FTC has studied "surveillance pricing", New York requires a disclosure when prices are set by an algorithm using personal data, and the EU plans to address unfair personalisation practices in its Digital Fairness Act. Seek legal advice before personalising prices.

Are third-party cookies going away?

Not in Chrome. Google abandoned its plan to remove them in 2024 and retired most Privacy Sandbox technologies in October 2025. Safari and Firefox block them by default, and Apple's App Tracking Transparency limits tracking in apps, so first-party data remains the safer foundation.

Key terms

Personalisation

Showing different content, products, prices or messages to different people based on what is known about them. Done well, it makes shopping easier; done badly, it feels intrusive or wrong.

Personalisation engine

Software that decides, in real time, which experience each visitor sees, and usually tests it. Gartner's narrow market definition covers these engines only.

Rule-based targeting

Personalisation written by people as "if this segment, show that". Easy to understand, but hard to scale beyond a few dozen rules.

Recommendation engine

Software that predicts which products or content a person is likely to want, often from what similar customers bought or viewed (collaborative filtering).

Search and merchandising

On-site search, category ranking and product placement. In our experience, it is where personalisation touches the most revenue for most retailers.

Customer data platform (CDP)

Software that unifies customer data from many sources into one profile per customer, then sends audiences to other tools. It is the data layer behind cross-channel personalisation.

Composable CDP

A CDP that works on top of the company's own data warehouse instead of copying data into the vendor's system. It gives more control, but needs more data expertise.

AI decisioning

AI that chooses, for each customer, which offer, message, channel and timing to use, and learns from the results. It replaces many hand-built rules and A/B tests.

Multi-armed bandit

A testing method that shifts traffic towards better-performing variants while the test runs. Faster to exploit a winner, but less suited to learning why it won.

Holdout group

A random share of customers who never receive the personalised experience. Comparing them with everyone else is the only reliable way to measure what personalisation adds.

First-party data

Data a company collects directly from its own customers, with their consent. It matters more as browsers and regulators restrict tracking across sites.

Feature flag

A switch in software that turns a feature on or off for chosen users. Product teams use flags to release, test and target features.

Agentic commerce

Shopping done or assisted by AI agents, such as ChatGPT, Gemini or a retailer's own assistant, which search, compare and sometimes buy for the customer.

Surveillance pricing

Setting individual prices from personal data such as location, browsing or demographics. US regulators are studying it, and New York requires it to be disclosed.

Sources

Company facts, deals and financials come from press releases, investor relations pages and filings, cross-checked with reputable outlets (TechCrunch, Axios, Business Wire, PR Newswire). Deal values marked "reported" come from press coverage rather than the companies. Prices were checked on vendor pricing pages on 26 September 2026 and exclude taxes. Market-size estimates use different definitions and are not comparable. Research findings were checked against the original papers or publishers' pages. Henkan & Partners works with several vendors named in this report, including as a certified partner of Kameleoon and AB Tasty (now Wingify); this report does not rank vendors.

1. Gartner, Magic Quadrant for Personalization Engines, February 2026
2. SAP, a Leader in the 2026 Gartner Magic Quadrant for Personalization Engines
3. Mordor Intelligence, personalization engine software market
4. Research and Markets, personalization software market
5. Grand View Research, AI-based personalization engines market
6. Adobe, Q4 and FY2025 results (SEC)
7. Salesforce, Q4 FY2026 results
8. Pega, Q4 2025 results (SEC)
9. Klaviyo, 2025 results
10. Braze, fiscal 2026 results
11. Optimizely reaches $400M ARR, May 2024
12. Bloomreach surpasses $260 million ARR, February 2026
13. LaunchDarkly expands leadership team, January 2026
14. Coveo, fiscal 2026 results
15. TechCrunch, Everstone combines Wingify and AB Tasty, January 2026
16. TechCrunch, Everstone acquires Wingify for $200M, January 2025
17. AB Tasty and VWO unite under Wingify, September 2026
18. Linden, Smith and York, Amazon.com recommendations: item-to-item collaborative filtering, 2003
19. Netflix Prize (overview)
20. Gomez-Uribe and Hunt, The Netflix Recommender System, ACM TMIS, 2015
21. McKinsey, How retailers can keep up with consumers, 2013
22. Search Engine Watch, Omniture acquires Offermatica, 2007
23. Adobe to acquire Omniture, September 2009 (SEC)
24. Insight Venture Partners invests in Episerver, 2018
25. Episerver to acquire Optimizely, 2020
26. Axios, McDonald's agrees to buy Dynamic Yield, 2019
27. Mastercard closes Dynamic Yield acquisition, April 2022
28. Salesforce acquires Evergage, 2020
29. VentureBeat, Salesforce acquires Demandware for $2.8B, 2016
30. TechCrunch, Adobe buys Marketo for $4.75B, 2018
31. Twilio completes acquisition of Segment, 2020
32. SAP completes acquisition of Emarsys, 2020
33. TechCrunch, Algolia raises $150M at $2.25B valuation, 2021
34. Bloomreach valuation reaches $2.2B, 2022
35. CNBC, Klaviyo IPO, September 2023
36. Gartner predicts 80% of marketers will abandon personalization efforts, 2019
37. WebKit, full third-party cookie blocking, 2020
38. TechCrunch, Apple's App Tracking Transparency has arrived, 2021
39. Monetate acquires SiteSpect, June 2025
40. Monetate acquires Simon AI, July 2026
41. Kibo spins off Monetate, 2022
42. Harness completes acquisition of Split, 2024
43. TechCrunch, Datadog acquires Eppo, 2025
44. TechCrunch, OpenAI acquires Statsig, September 2025
45. Amplitude and Statsig partnership, May 2026
46. Klevu and Searchspring form Athos Commerce, January 2025
47. Attraqt backs £63.2M Crownpeak takeover offer, 2022
48. Rokt and mParticle merge, January 2025
49. AdExchanger, Rokt acquires mParticle for $300 million
50. Uniphore to acquire ActionIQ, December 2024
51. TechCrunch, Fivetran acquires Census, May 2025
52. Contentstack acquires Lytics, January 2025
53. PYMNTS, Hightouch valued at $2.75 billion, April 2026
54. Braze announces agreement to acquire OfferFit, March 2025
55. Braze completes acquisition of OfferFit, June 2025
56. Contentful to acquire Ninetailed, August 2024
57. Sitecore unveils SitecoreAI, November 2025
58. Publicis Groupe acquires Yieldify, January 2023
59. SalesTechStar, Fanplayr rebrands to Verada, September 2026
60. AdExchanger, CDPs in Gartner's trough of disillusionment, 2024
61. CDP Institute, CDP industry is growing again, 2025
62. AWS, Amazon Personalize generally available, June 2019
63. TechRadar, Google Recommendations AI, 2020
64. Hightouch Series C and AI Decisioning, February 2025
65. MoEngage, additional $180 million in Series F funding, December 2025
66. MoEngage, Merlin AI custom agents and open MCP architecture, June 2026
67. Salesforce announces Marketing Cloud Next, June 2025
68. Adobe, general availability of AI agents, September 2025
69. Tealium newsroom (AI Decisioning, May 2026)
70. Mastercard, Dynamic Yield unveils Shopping Muse, November 2023
71. Optimizely Opal transformation, May 2025
72. Kameleoon, introducing prompt-based experimentation, June 2025
73. Nosto introduces Huginn, October 2025
74. Wingify, Wingz AI
75. Adobe, AI traffic to retail sites, 2026
76. Stripe, Instant Checkout in ChatGPT, September 2025
77. Checkout.com, OpenAI's agentic commerce shift, April 2026
78. Google, Universal Commerce Protocol, January 2026
79. PayPal and Perplexity launch Instant Buy, November 2025
80. Yahoo Finance, Amazon's Rufus AI shopping assistant, November 2025
81. Shopify, Winter '26 Edition
82. McKinsey, The value of getting personalization right—or wrong—is multiplying, 2021
83. BCG, personalization programs increase revenues by 6% to 10%, 2017
84. Zielnicki et al., The Value of Personalized Recommendations: Evidence from Netflix, 2025
85. Lee and Hosanagar, Impact of Recommender Systems on Sales Volume and Diversity, ICIS 2014
86. Bernardi et al., 150 Successful Machine Learning Models, Booking.com, KDD 2019
87. Gordon et al., A Comparison of Approaches to Advertising Measurement, Marketing Science, 2019
88. Gartner, personalization can triple the likelihood of customer regret, June 2025
89. Aguirre et al., Unraveling the Personalization Paradox, Journal of Retailing, 2015
90. Kohavi, Tang and Xu, Trustworthy Online Controlled Experiments, chapter 1
91. Google Privacy Sandbox, a new path, July 2024
92. Google Privacy Sandbox, next steps, April 2025
93. Google Privacy Sandbox, update on plans, October 2025
94. European Commission, very large online platforms and search engines
95. DSA Article 38
96. EU AI Act implementation timeline
97. EU AI Act Annex III
98. GDPR Article 21
99. GDPR Article 22
100. European Parliament Legislative Train, Digital Fairness Act
101. FTC, surveillance pricing study, January 2025
102. Jones Day, New York's algorithmic pricing disclosure law, November 2025
103. MultiState, 20 state privacy laws in effect in 2026
104. Delta, response on AI pricing, July 2025
105. Grocery Dive, Instacart ends price tests, December 2025
106. Vendor pricing pages checked 26 September 2026: Salesforce Personalization, LaunchDarkly, Statsig, GrowthBook, PostHog, Algolia, Recombee, OptiMonk, Privy, Justuno, Wisepops, CleverTap, Contentful, Contentstack, Uniform
107. Omnisend, Klaviyo pricing (secondary source), May 2026