• Studio ATAO

Toolkit for Implementing Systematic Changes Towards Equitable Representation in Food Media Companies

Last Updated: October 19, 2021

Written By: Jenny Dorsey, Edric Huang

Edited By: Emily Chen, Osayi Endolyn, Priya Krishna, Karen Kumaki, Ximena N. Larkin, Korsha Wilson


This document is meant to be a resource for food media companies committed to remedying imbalanced systems of power and representation within their organization. This toolkit addresses both equitable representation within external-facing content and the need for internal pathways to leadership for folks with marginalized identities.

This toolkit builds off of the learnings of our Toolkit for Recognizing, Disrupting, and Preventing Tokenization in Food Media.

If you want a downloadable PDF of this document, please enter your email below and we'll send it directly to your inbox.

This free resource is made available by our regular donors. If you've learned something from our work, please consider supporting us financially so we can continue bringing free educational content to everyone. You can join our community via Patreon or send us a one-time donation via GiveLively.

Table of Contents

Click to jump to section

Introduction: Food Media’s 2020 “Reckoning”

Equitable Representation in the Content of Food Media

Ways to resist the white gaze and encourage inclusive storytelling

Equitable Representation in the People Behind Food Media

Ways to cultivate more pathways toward power for those with marginalized identities

Organizational Change

Considerations for sustainable, long-term, industry-wide solutions for achieving equity

Genesis of This Document

What’s Next?

Introduction: Food Media’s 2020 “Reckoning”

Some have referred to 2020 as a time of “reckoning” within the food media industry. After the murder of George Floyd prompted the largest movement in U.S. history with the Black Lives Matter protests, food media outlets small and large released public statements denouncing racism and promising to stand in solidarity with BIPOC. Many were immediately criticized for the performative nature of these words, as discrimination remained rampant within the organizations themselves. Most notably, Bon Appétit Magazine came under fire for a culture of toxicity and pay discrimination against BIPOC employees, leading to a public dissolution of the existing Test Kitchen team and the departure of long-time Editor-in-Chief Adam Rapoport.

Since then, food media organizations have rallied around new goals of diversity, equity, and inclusion, such as crediting recipe sources, distributing open calls for pitches from BIPOC writers, creating diversity task forces, and greenlighting more “political” food content. However, a great deal of this change has fallen on individuals within each organization, without programmatic, systematic, and institutional processes that align organizational goals with DEI initiatives.

Still untouched is the necessary unraveling of incentivization and revenue structures that reward discriminatory practices and systemically limit BIPOC access and power. To date, most national food media outlets have almost-exclusively white leadership teams; only a small handful of exceptions exist, typically as independent publications like Whetstone Magazine, LA Taco, Peddler Journal, and For the Culture Magazine.

Ultimately, our intention with this toolkit is to:

  1. Identify common challenges food media organizations face in their ongoing efforts to move toward a more sustainably equitable future.

  2. Offer potential ideas for systematic overhaul of existing organizational structures, drawing from the ideas of Salon participants, public event panelists and guests, and individuals within the food media industry.

We have divided these challenges and opportunities into three main arcs: content, people, and organizational change. The organizational change section proposes larger processes and frameworks to build across the industry, such that the changes in the Content and People sections are not one-off forms of reactionary crisis management but steps toward sustainable change management. We hope that these sections offer enough detail for short-term execution, long-term implementation, and immediate action.


1/ Reevaluating Analytics

For digital publications, analytics often drive content decisions. Traditionally, content success is defined through the quantity of engagement: page views, link clicks, and social shares. As one participant at a digital-only publication said: “We are traffic obsessive...so it’s easy to go into muscle memory and ‘play the hits’.” In fact, as the readership at most major publications has largely skewed white, using analytics to drive engagement encourages a continuous stream of white-centric and apolitical food content — in other words, the hits. It also encourages companies to build content based on popular keywords, which are typically already biased and/or Eurocentric. A vicious cycle begins: more white-centered content draws in more white readers — while potentially alienating non-white audiences — and generates data reflecting the substantial success of white-focused content over others.

This problem is further complicated when organizations lack historical data to gauge the potential for success. When the emphasis is on driving traffic that generates revenue via advertising dollars and sponsorships, the stories, features, and sections that feel potentially “risky” to leadership are scrutinized with far more analytical vigor than other content. In this way, calls for equitable representation are perceived to be in opposition to business goals. For example, several digital publication participants noted that although the summertime is always a low-traffic period, the ensuing worry about poor analytics leads to a de-emphasis on DEI-related initiatives during that same time of year.

This lack of data in support of more diverse content can be used to delay or deny opportunities for representation and create a negative cycle where not giving the content a chance ensures it can never generate any data. In these scenarios, when BIPOC-centric content is approved, it is often given only “one shot” to prove itself worthy of continued investment. After a poor showing, leadership may use these very analytics to axe future content without seeming discriminatory. One participant offered as an example: “If shellfish recipes don't seem to do well, when these recipes are proposed the response is, ‘Oh, people will not cook that’ — and that can lead to discounting entire cuisines.”

This lens of content analysis ignores many underlying conditions, such as a pre-existing white audience unaccustomed to certain recipes, ingredients, or faces. As one participant explained, “Our YouTube presence is 90% white men and they are looking for a specific kind of video...so we’ve been trying to train that audience [to be closer to the on-site audience, which is more diverse].” However, this audience education requires dedicated dollars and commitment from upper management, especially because some attrition should be expected when shifting content focus. As one participant asked of their team:

“Are we willing to sacrifice traffic for a better story that our readership might not have heard of?”

The underlying assumption by white-majority leadership that diverse content doesn't perform is also proven to not be true. As one BIPOC participant explained, “POC have a lot of pride. When they see themselves reflected [in the publication], they will come out and support. But even with past success, I still have to prove it every time. Even if it made crazy numbers last time, I have to prove I can do it again.

(It is worth noting that participants from digital publications with a paywall or subscription model expressed less concern or attention to these analytics because of existing, recurring revenue streams. However, they did note audience sentiment and feedback - generally collected via surveys - factored into content decisions. Participants from print media outlets, on the other hand, explained that “sales and ads dictate the issue themes.”)

Opportunities for Change

1/ Shift to engagement-based analytics.

Instead of focusing solely on quantity-based analytics, participants suggested also incorporating statistics that highlight the extent of the audience’s engagement with the content, such as:

  • Time spent on page

  • Channels of shares (e.g., content was shared on channels where the publication is not as active)

  • Content viewed per session (e.g., watching 5 videos on the same channel)

These reports are most helpful when paired with qualitative feedback; one participant explained how it is useful when the insights team includes instances when “a story started a conversation here, or helped draw in these people who never read [our publication].” With these metrics in mind, participants still suggested that editorial takes a “data-informed, but not data-driven, approach” when it comes to assigning or prioritizing content. In particular, organizations should set aside a budget for pieces that are clearly non-SEO, non-social content because not every single piece is meant for virality.

2/ Set longer timelines for results.

Overhauling biased systems requires time and resources, particularly when data is involved. This extends beyond editorial’s purview, into departments like audience development and sales operations. The organization needs more time (several participants recommended a year or longer) to see results from “training” readership on the publication’s point of view and to gather longer-tailed analytics around new types of content.

That means stakeholders who manage budgets must know that acquiring traction takes time and continued support from all departments (e.g., social media, sales, operations). One participant framed this time period as a “double approach of training consumers and getting new ones” — meaning, educating existing consumers and attracting those who support the organization’s overall goals for more representative content.

3/ Experiment with alternative streams of revenue.

For example, branch into eCommerce (e.g., Food52 Provisions) or live events (e.g., Infatuation Eeeeeatscon). Some participants mentioned that because these opportunities are not predicated on advertising revenue, they become less beholden to traditional, numerical-only analytics that are implicitly white-biased.

2/ Implicit or Explicit Content Quotas

To promote diversity in editorial content, executives will create implicit or explicit content quotas. Yet the existence of these quotas can limit how much “diverse” (a term up for interpretation itself) content is allowed. Quotas can “other” this content as distinct from “regular” content and they can fluctuate depending on who oversees the quotas in the first place.

As one participant recounted, the editor for a column proposed a quota that “50% of the content would represent BIPOC — but I pointed out that would mean 50% is still white.” Multiple other participants stated they felt uncomfortable while researching whether a brand or business was BIPOC-owned, as a way to meet quotas. As one participant asked the group, “How are we considering intersectionality when it comes to something like a gift guide?”

However, some saw the use of quotas as an imperfect, but necessary, interim step. One participant expressed, “I don’t think quotas are useless, because they can give us a landing point for our goals.” Others agreed, saying that “numbers are a tangible way of holding ourselves accountable as opposed to just saying ‘we want to do this’.” Yet while one participant pointed out that “numbers are good to see what direction the company is moving in”, they also expressed concern that:

“When diversity is met with a number, the response becomes, ‘We took care of it for the year. Let’s move on.’”

With or without a quota to hit, the final decision-makers defining how much and what qualifies as “diverse content,” are also still predominantly white. This naturally gives rise to certain preferences. As Bon Appetit’s Elyse Inamine wrote in her public letter: “Early on, I noticed how [other BA editors] gravitated toward the same aesthetic, cuisines, and chefs — and how little we had in common.” In the same vein, BIPOC-focused stories tend to be grouped in specific ways — as stories of overcoming hardship or growing up with a specific cultural heritage. As one speaker stated in our public panel about tokenization, “I would have loved to have seen someone who looked like me [a BIPOC] making 30-minute meals or living it up in the Hamptons.”

Correspondingly, even when publications begin to diversify theirconf stories, stories rooted in non-western cultures are often essentialized or categorized through a Eurocentric lens. As one participant explained, “It’s an endless stream of ‘differentiated’ content when it comes to [the south of] France or [individual cities in] Italy...but then it’s like ‘We can’t run this roti recipe from the southern part of India, because we ran a recipe from the northern part of India a month ago.” As a result, certain types of content are designated by leadership as repetitive. Even though no diversity quota is explicitly stated in these types of exchanges, the implication is that certain people, cultures, and cuisines are seen as too niche and/or unimportant to be covered with the level of detail as others.

Ultimately, quotas are an imperfect metric, and are not a means to drive change. Rather, the focus needs to be on the people who decide these quotas and the full process of how content is brought in and nurtured at a publication.

Opportunities for Change

1/ Redistribute the power to greenlight and veto content.

The criteria for evaluating pitches and greenlighting content will always be subjective. Upending existing power dynamics is necessary for change that extends beyond the goodwill of one, or a few, leaders at an organization. To do this, organizations can visibly promote more BIPOC staffers to decision-making positions, give more BIPOC staffers veto or override power, or disperse currently concentrated power by offering more staffers the ability to greenlight content independent of management. (We also discuss other ways to offer more transparency within organizations under the People section.)

2/ Encourage diverse content via pitches (versus assignments).

Participants suggested a bottoms-up approach. Encourage diverse sets of contributors (both staffers and freelancers) to organically pitch stories, rather than allowing management to assign what they believe are diverse stories. However, this still requires the definition, purpose, and intent of “diverse content” to be made clear. Otherwise, as one participant said, “We just greenlight the first three pieces that fit [a general diversity bucket like] ‘Chinese’ or ‘Black.’”

For example, a call for pitches centered on an ingredient, cooking style, or even equipment could also be complemented with a statement that the publication is prioritizing the item’s history with BIPOC communities, in order to challenge existing narratives around that item. (Read more about pitch guidelines under People > The Problem with Networks.) This offers intentionality, framing, and the viewpoint of the organization from the get-go, making it clear to writers and assigning editors what does or does not make sense for this content package. Another participant suggested regular columns or series that “intentionally allows for diverse perspectives,” such as a “Way We Eat” series, each spotlighting a different person and their approach to food.

In particular, participants agreed that greenlighting more personal essays is a good place to start, as more diverse stories often emerge here. However, this is only the first step in integrating a wider range of content throughout the publication. Contextualizing non-white stories as only fitting for first-person essays (versus reported pieces that may force editors to learn more individually about subject matter, for example) also limits BIPOC contributors in the pitches they feel will be accepted. Thus, intentionally diversifying how BIPOC foodways are presented within a publication is also an important part of this process.

3/ Incorporate data-informed, quantitative means to confirm efficacy of changes.

For example, one participant noted that using a random diversity audit could be a good starting point to determine if internal changes had actually resulted in tangible, numerical outcomes (e.g. increases in % of BIPOC contributors). At their organization, a random selection of stories for both digital and print from the last 2 or 5 years, respectively, are currently being analyzed to determine the publication’s overall composition of diverse content.

3/ The White Gaze

By and large, the editorial staff at most food media publications are white. When these same editors interface with BIPOC contributors, freelance or not, there is a power imbalance and sociocultural difference in how content is understood and interpreted. The way certain foods and food experiences are framed often assume the white experience to be the neutral, unbiased, and universal one.

For example, we noted in our initial toolkit that non-white-centric language (such as dish names and ingredients) are often assumed to be “new” and require additional explanation; in many instances, these are biases that can be perpetuated by the editor themselves. Multiple participants pointed to NYT’s now infamous Thai fruit article as an overt case of irresponsible writing; for example, at one point in the piece, the writer describes dragonfruit as “a bland mush with tiny seeds that can require floss to dislodge.” Corn on the cob, as one example, is generally spared such critical framing.

In fact, terms like “holiday,” “restaurant,” and “chef” are generally understood to favor a western point of view. As one participated noted, “holiday” typically refers to white holiday content (e.g., recipes for Thanksgiving or Christmas), which neglects holidays like Diwali that fall roughly during the same time period and ignores major holidays during other parts of the year like Lunar New Year or Eid al-Adha.

Similarly, the notion that a restaurant is an institution inside four walls — versus a street cart, food stand, etc. — is built upon western interpretations of dining, and favors higher-income, often white, entrepreneurs who have access to the capital necessary to open brick-and-mortars. As Khushbu Shah of Food & Wine commented in The Washington Post, it was important that her search for Best New Restaurants “[included] food trucks, pop-ups and food halls — ‘things that don’t require a lease on the space.’”

These foundational definitions are, again, often decided unilaterally by white leadership. For example, one participant explained that an internal conversation on “balancing brand with current, relevant topics” opened discussions around how white leaders had decided what is “on brand.” This also extends to the visual identity of a publication. When a publication already has a set aesthetic (with its styling, lighting, props, etc.), it can be challenging to retrofit different types of dishes into this look.

As one participant pointed out, not everything should be served with a glass of wine on the side, even if that may be visually appealing for some audiences. In other cases, the prop styling or imagery may lean on stereotypes held by the white staffers developing the visual content. For example, some news outlets used images of Chinatown for all COVID-related coverage, even though the coverage had no connection to the neighborhood, perpetuating dangerous stereotypes of yellow peril. Food media similarly allows “a group to stand in for a larger idea,” as one participant put it, often oversimplifying or creating static, timeless images of non-white cultures.

As an initial first step to combating the white gaze (and other sources of bias or inequity), some organizations employ sensitivity readers to review content before it goes live. Unfortunately, these readers may be selected arbitrarily — e.g., they may be one of the more senior editors — or the task may be assigned to a BIPOC staffer as additional, unpaid labor. As one BIPOC participant shared:

“Just because I’m a person of color does not mean that I’m qualified to edit every story [by or about people of color].”

While the process of having a sensitivity reader has recently become formalized across publications, the process of training and educating sensitivity readers has not.

Additionally, because many editors also share similar backgrounds — e.g., similar alma maters, fields of study, and former workplaces — there are commonalities in the style of editing favored and seen as “good.” Multiple participants agreed they were taught to “rip a piece apart.” This is not only counterproductive in establishing different narrative voices at a publication, it also destroys trust in an already unequal relationship. Despite this, one’s editing style is generally at the sole discretion of the assigning editor, and no clear systems seem to exist to mediate potential conflicts in these relationships.

Opportunities for Change

1/ Normalize individual research in the editing process.

If content doesn't immediately make sense to the editor that is solely the result of their lack of familiarity with a culture (as opposed to a reporting or continuity issue, for example), it should be their responsibility to research it instead of expecting the writer to extrapolate more in the assignment. As one participant recounted, “When a writer I was working with mentioned an ingredient and referenced a part of India where the ingredient comes from, I was confused about the verbiage, and initially wanted to highlight the passage and ask them to clarify, but instead I did some Googling, reading, and learning on my own so I could properly edit without assuming everyone would have my point of view.”

2/ Proactively test the assumptions of staffers.

“What exactly is a ‘universal experience?’” one participant asked the group, pointing out how visuals for pieces on food and marriage, for example, typically center a white, heterosexual couple. As part of their organization’s onboarding process, new team members are now asked a series of questions, like the above, to “consider their own biases.” Writers are also encouraged to avoid language that makes collective assumptions (e.g., “everyone has heard of…”).

3/ Formalize expectations of writers and offer explicit guidelines around DEI work.

To codify editorial and visual standards beyond one-off examples, many publications have begun working on internal DEI style guides. One participant actively working on a diversity-specific editorial style guide explained how there are “process checklists for each beat” that ask writers to evaluate everything from the diversity of their experts interviewed to word choice, to the framing of certain experiences.

As we describe in more detail in our whitepaper, Well+Good’s 16-page internal DEI Style Guide offers guidelines on how to write across race, gender, orientation, and disability, based on the recommendations from experts and advocacy organizations working in those fields. While one executive editor approves and adds all major changes or additions to the guide, it is a collaborative effort among multiple editors; we believe this is important to ensure such a document encapsulates multiple perspectives and retains institutional value even as employees cycle out.

Well+Good’s guide also links to specific past examples to learn from; for example, they point out that their own Sweet Potato Noodle Pad Thai recipe from 2017 “was explicitly positioned as a ‘healthier’ version of takeout solely because it used sweet potato noodles instead of rice noodles, when there’s nothing necessarily wrong with rice noodles.” Many publications are still learning how to continually refresh their guides and ensure consistent adherence to the guidelines as new writers and editors join the organization.

4/ Establish a formal sensitivity review process that begins at the start of the content pipeline.

Multiple participants voiced that late-stage “sensitivity reviews don’t cover fundamental flaws” and rely on one or a handful of sensitivity readers to catch all potential issues, which is unrealistic. Instead, a participant spoke of earlier-stage meetings where editors would lead their own and other departments through scheduled stories, during which anyone could comment on potential red flags, items to consider, visual assets, or a framing to avoid. This also can be a helpful exercise in building cultural competency across the department.

To add multiple layers of feedback into their process, participants also recommended designating a separate, internal Slack channel to ask for peer review. While this was recommended by all — and is helpful to give more junior employees a voice — it also presented efficiency challenges. As one participant asked, “How do we decide if that’s a personal opinion versus the best choice?” Another pointed out: “It’s hard to incorporate 6 peoples’ feedback into 2 lines of text...and everyone is afraid of making the final call in case it is the ‘wrong’ call.”

During our discussions, participants repeatedly stated there should be dedicated personnel that can fact-check and review content full-time. We believe there should be a transparent process for selecting who is suitable and well-equipped for a sensitivity read, as well as who manages the relationship and edits with these readers. Additionally, multiple other participants stated that it was on organizations to provide “proper training for everyone [so they can] self-diagnose [issues] along the way,” have a budget available for external sensitivity reads, and/or have the research team (if one exists) support the review of origin stories of selected pieces.

Even with these processes in place, sensitivity reads require time to be effective. With writers and editors pressed to create so much content each month, they often do not have the bandwidth to comprehensively research and vet every piece that crosses their desks. One participant recounted a time when their organization made a particularly public mistake: Even though many people they “consider to be very good at sensitivity reads” saw the piece before publication, no one had flagged the mistake.

5/ Open communication channels with the original creator.

Participants pointed out that creative briefs for photo or video shoots tend to come from the editors and not the original creator, writer, or recipe developer. This prioritizes the editors’ lens in the project, which can complicate or occasionally misconstrue the original preferences or intent of the creator. While this is somewhat inevitable, participants said that more channels of communication to the original creator are helpful when questions arise; the more specifics in the brief, the better. One participant added: “Including what not to do in the visual guides and creative briefs [is] really useful.”

6/ Create formal conflict mediation channels.

When it comes to editor and writer relationships, no formal system for addressing disagreements or mediating conflict seems to exist. Participants were aware of the reality that “the more you work with [someone] the more they are willing to say.” As part of cultivating a mutually respectful relationship, participants agreed it was important for editors to use the track changes function and “explain yourself in the edits” by providing specific, concrete reasons for changes. In the case of potentially sensitive edit requests, participants would also send prescribed edits to colleagues for a fresh review.

While these are helpful individual changes, a more systematic approach should be explored. For example, designating and training specific staff members at the organization as mediators for potential conflicts. This way, the two (or more) parties have a structured space to deliver direct feedback, and have this mediator document the appropriate action steps for resolution that can be delivered to HR or other departments if necessary. We talk more about conflict mediation in People > Building Inclusive Cultures.