MoineauWorld.com – Where Cooking Meets Tech & Everything Between

Tag: price experiments

Tell Google Discover What You Want | Analysis by Brian Moineau

Posted on by Brian Moineau

Tell Google What You Want: “Tailor your feed” Brings Prompt-Powered Control to Discover

Imagine opening Google Discover and being able to say, in plain English, “Show me cozy home-cooking videos, but only dairy-free recipes,” or “Keep politics out for a while — show me science and college basketball instead.” That’s the idea behind Google’s new experimental Labs feature called “Tailor your feed,” spotted in testing this week.

Why this feels different

For years, Discover has quietly learned from what you search, click, and ignore. It nudges you toward topics it thinks you’ll like, but the control panel has always been a bit clunky: tap three dots, mark something “not interested,” or favorite a source. “Tailor your feed” moves that control into natural language prompts — you talk to Discover like you would a helpful friend, and its AI updates your recommendations instantly.

This is not a full public rollout. It’s a Search Labs experiment in the Google app, currently limited to early testers (US English was reported), but the approach signals a bigger shift in how Google wants us to manage passive, algorithmic content.

What to know right now

  • The feature appears in the Google app’s Search Labs (tap the beaker icon in the top-left).
  • You open a prompt box labeled “Ask for the kind of content you want,” type a request, and Discover updates your feed instantly.
  • Prompts can include topics, formats, tones or “vibes,” publishers to prioritize, or content to avoid (e.g., “Stop showing me negative news”).
  • Google says Discover will remember these preferences and you can adjust them anytime; activity links back to My Activity.
  • The experiment is early and rolling out slowly — not everyone will see it yet. (Reported Dec 15–16, 2025.)

The practical examples that caught attention

  • Add a project-based topic: “I signed up for my first half marathon; give me training advice.”
  • Remove a stale topic: “I’m back from a NY trip — stop showing me travel tips.”
  • Narrow formats or dietary constraints: “Show me meal-prep videos that are dairy-free.”
  • Adjust tone: “Make my feed feel calm and cozy.”
  • Favor publishers: “Show more from The Washington Post.”

These examples illustrate how specific you can be — goals, formats, sources, and even mood are fair game.

Why Google is doing this

  • Personalization, made faster: Natural-language prompts shortcut the months-long feedback loop of behavior-based learning.
  • Engagement and retention: If people get what they want, they’ll spend more time in Discover (and the Google app).
  • Better signals for relevance (and ad targeting): More explicit preferences are valuable for content ranking — and for ad relevance.
  • Experimentation culture: Google Labs lets the company try riskier UI and AI ideas without committing to a wide release.

The potential upside

  • Faster, clearer control: Users can correct misfires quickly without hunting through menus.
  • Useful for life changes: Short-term goals (training for a race, planning a move) become easier to surface.
  • Better format discovery: If you want videos, explain it — Discover can prioritize that format.
  • Reduces noise: If you need a break from heavy topics like politics, you can simply say so.

The trade-offs and concerns

  • Filter bubbles deepen: Explicitly asking to favor certain topics or tones may reduce exposure to diverse viewpoints.
  • Publisher discoverability: Smaller outlets might lose traction if users ask for a narrow set of sources or vibes.
  • Privacy and activity linking: The prompt history links to My Activity; anything you tell Discover becomes another personalization signal.
  • Misunderstanding and misuse: Natural-language interfaces can misinterpret vague prompts, requiring additional back-and-forth.

How this changes the Discover experience

Think of Discover sliding along a spectrum from passive surfacing to semi-curated reading list. “Tailor your feed” pushes it closer to a hybrid: still recommendation-driven, but with on-demand curation. That could make Discover feel more intentional for users who want it — and more “sticky” for Google.

My take

Giving users a conversational way to tweak their feed is a smart move. It matches how people already describe preferences — in goals, vibes, and formats — and it reduces friction. But expect the usual tension: personalization makes life easier and more pleasant, yet it also tightens your content bubble. Ideally, Google will offer nudges that encourage variety and let users reset or explore outside their requested tastes.

If you’re curious and see the Labs beaker in your Google app, it’s worth trying — it’s an experiment, after all. Use it deliberately: try a goal-based prompt for a few weeks, then toggle it off to see how much Discover relied on that instruction.

Sources



Related update: We recently published an article that expands on this topic: read the latest post.

Instacart’s Algorithm Inflates Grocery | Analysis by Brian Moineau

Posted on by Brian Moineau

The grocery price you see might not be the grocery price someone else sees

Imagine loading your cart with the same staples you always buy — eggs, peanut butter, cereal — and watching the total quietly climb depending on who’s logged into the app. That’s the unsettling picture painted by a new investigation into Instacart’s pricing experiments. The findings suggest algorithmic pricing on grocery delivery platforms is no longer hypothetical: it’s affecting the bills people pay for essentials.

Why this matters right now

  • Grocery affordability is a top concern for many households in the U.S., and small percentage differences compound quickly.
  • The findings come from a coordinated investigation by Groundwork Collaborative, Consumer Reports, and labor group More Perfect Union that tested live prices across hundreds of Instacart users in multiple cities.
  • The study’s headline figure — that average pricing variation could cost a four-person household roughly $1,200 a year — is what turned heads and spurred debate about transparency, fairness, and the role of algorithmic experiments in everyday commerce.

What the investigation found

  • Across tests in four U.S. cities, nearly three-quarters of items showed multiple prices to different shoppers for the exact same product at the exact same store and time. (groundworkcollaborative.org)
  • Price differences for individual items were often sizable — the highest price was as much as 23% above the lowest for the same SKU. Examples included peanut butter, deli turkey and eggs. (groundworkcollaborative.org)
  • Average basket totals for identical carts differed by about 7% in the study’s sample. Using Instacart’s own estimates of household grocery spending, that swing could translate to roughly $1,200 extra per year for a household of four exposed to the typical price variance observed. (consumerreports.org)

How it works (the mechanics, in plain language)

  • Instacart and some retailers use dynamic pricing tools and experimentation platforms (including technology Instacart acquired in 2022) to run price tests.
  • These systems can show different “original” or “sale” prices and can test multiple price points simultaneously across users to see what increases purchases or revenue.
  • The troubling element isn’t experimentation per se — price testing exists in physical stores too — but the lack of disclosure and the fact that shoppers trying to comparison-shop or budget are effectively excluded from seeing consistent prices. (consumerreports.org)

Responses and pushback

  • Instacart has acknowledged running pricing experiments in some cases but has asserted it does not use personal or demographic data to set prices and that most retailers do not use their pricing tools. The company also said it had stopped running experiments for some retailers named in coverage. (consumerreports.org)
  • Retail partners gave mixed reactions: some distanced themselves or said they were not involved, while others did not comment. Lawmakers and consumer advocates have seized on the report to call for clearer disclosures or limits on “surveillance pricing.” (consumerreports.org)

Broader implications

  • Algorithmic pricing can amplify existing inequalities if certain groups are more likely to be exposed to higher-priced experiments — even if a company insists it’s not using demographic targeting. The opacity of models and the complexity of A/B tests make oversight difficult. (consumerreports.org)
  • The grocery sector is already under regulatory and public scrutiny for price transparency. States and federal policymakers are beginning to consider rules about algorithmic price disclosures and “surveillance pricing” bans. Expect legislative activity and watchdog attention to grow. (wcvb.com)
  • For consumers, the convenience of home delivery may now come with a hidden premium that is not obvious at checkout.

What shoppers can do now

  • Compare with in-store prices when possible. If an item looks markedly higher in the app, check the store shelf price or call the store before completing a large order. (wcvb.com)
  • Use price-tracking habits: keep receipts, note repeated price differences, and report discrepancies to the retailer or Instacart. Consumer complaints create records that regulators and journalists can use.
  • Consider pickup (if available) or buying directly through a retailer’s own online service when price transparency matters most. Some retailers still control and publish consistent prices on their own platforms. (wcvb.com)

My take

Algorithmic testing can be a useful business tool — it can tune pricing to demand, clear inventory, or optimize promotions. But when the item is a family’s food staples, the ethical and practical bar for transparency should be higher. Consumers budgeting for essentials need predictable, comparable prices. If pricing experiments are going to be run on grocery purchases, they should be disclosed clearly, limited in scope for essentials, and subject to guardrails so that convenience doesn’t become a stealth surcharge.

Sources



Related update: We recently published an article that expands on this topic: read the latest post.