Laravel Eloquent tips that you need to know

Jan 25, 2022

This is a short list of the amazing hidden Laravel eloquent 30 tips that make the code go on smoothly.

Github repo:

#1 – Invisible Database Columns

The invisible column is a new concept in MySQL 8. What it does: when you run a `select *` query it won't retrieve any invisible column. If you need an invisible column's value you have to specify it explicitly in the `select` statement.

And now, Laravel supports these columns:

Schema::table('users', function (Blueprint $table){

$user = User::first();
$user->secret == null;

#2 – saveQuietly

If you ever need to save a model but you don't want to trigger any model events, you can use this method:

$user = User::first();
$user->name = "Hamid Afghan";


#3 – Default Attribute Values

In Laravel, you can define default values for columns in two places: Migrations and models.

Schema::create('orders', function(Blueprint $table){
  $table->string('status', 20)

This is a well-known feature. The status column will have a default draft value.

But what about this?

$order = new Order();
$order->status = null;

In this case, the status will be null, because it's not persisted yet. And sometimes it causes annoying null value bugs. But fortunately, you can specify default attribute values in the Model as well:

class Order extends Model
  protected $attributes = [
    'status' => App\Enums\OrderStatuses::DRAFT,

And now the status will be draft for a New Order:

$order = new Order();
$order->status  === 'draft';

You can use these two approaches together and you'll never have a null value bug again.

#4 – Attribute Cast

Before Laravel 8. x we wrote attribute accessors and mutators like these:

class User extends Model{
  public function getNameAttribute(string $value): string
      return Str::upper($value);

  public function setNameAttribute(string $value): string 
      $this->attributes['name'] = Str::lower($value); 

It's not bad at all, but as Taylor says in the pull request:

This aspect of the framework has always felt a bit "dated" to me. To be honest, I think it's one of the least elegant parts of the framework that currently exists. First, it requires two methods. Second, the framework does not typically prefix methods that retrieve or set data on an object with get and set

So he recreated this feature this way:

use Illuminate\Database\Eloquent\Casts\Attribute; 

class User extends Model { 
  protected function name(): Attribute { 
    return new Attribute(
       get: fn (string $value) => Str::upper($value),
       set: fn (string $value) => Str::lower($value)

The main differences:

In this example, I used PHP 8 named arguments (the get and set before the functions).

#5 – find

Everyone knows about the find method, but did you know that it accepts an array of IDs? So instead of this:

$users = User::whereIn('id', $ids)->get();

You can use this:

$users = User::find($ids);

#6 – Get Dirty

In Eloquent you can check if a model is "dirty" or not. Dirty means it has some changes that are not persisted yet:

$user = User::first();
$user->name = 'Hamid Afghan'; 
$user->isDirty() === true;
$user->getDirty === ['name' => 'Guest User'];  

The `isDirty` simply returns a bool while the `getDirty ` returns every dirty attribute.

#7 – push

Sometimes you need to save a model and its relationship as well. In this case, you can use the push method:

$employee = Employee::first();
$employee->name = 'New Name';
$employee->address->city = 'New York';


In this case, the, save would only save the name column in the employee's table but not the city column in the addresses table. The push method will save both.

#8 – Boot Eloquent Traits

We all write traits that are being used by Eloquent models. If you need to initialize something in your trait when an event happened in the model, you can boot your trait.

For example, if you have models with slug, you don't want to rewrite the slug creation logic in every model. Instead, you can define a trait, and use the creating event in the boot method:

trait HasSlug { 
  public static function bootHasSlug() { 
      static::creating(function (Model $model) {
        $model->slug = Str::slug($model->title); 

So you need to define a bootTraitName method, and Eloquent will automatically call this when it's booting a model.

#9 – updateOrCreate

Creating and updating a model often use the same logic. Fortunately Eloquent provides a very convenient method called updateOrCreate :

$flight = Flight::updateOrCreate(
  ['id' => $id],
  ['price' => 99, 'discounted' => 1], 

It takes two arrays:

And the way it works:

I want to show you a real-world example of how I handle creating and updating models

The Controller:

public function store(UpsertDepartmentRequest $request): JsonResponse {
        DepartmentResource::make($this->upsert($request, new Department()))

public function update( UpsertDepartmentRequest $request, Department $department): HttpResponse { 
    $this->upsert($request, $department);

    return response()->noContent();

private function upsert(UpsertDepartmentRequest $request, Department $department): Department {

    $departmentData = new DepartmentData(...$request->validated());
    return $this->upsertDepartment->execute($department, $departmentData);

As you can see I often extract a method called upsert . This method accepts a Department . In the store method I use an empty Department instance because in this case, I don't have a real one. But in the

update I pass the currently updated instance.

The $this->upsertDepartment refers to an Action:

class UpsertDepartmentAction {

public function execute( Department $department, DepartmentData $departmentData): Department {

  return Department->updateOrCreate(
      ['id' => $department->id],$departmentData->toArray()

It takes a Department which is the model (an empty one, or the updated one), and a DTO (a simple object that holds data). In the first array, I use the $department->id which is:

And the second argument is the DTO as an array, so the attributes of the Department.

#10 – upsert

Just for confusion Laravel uses the word upsert for multiple update or create operations. This is how it looks:

['departure' => 'Oakland', 'destination' => 'San Diego', 'price' =>99],
['departure' => 'Chicago', 'destination' => 'New York', 'price' => 150]
['departure', 'destination'], ['price']);

It's a little bit more complicated:

So this example will:

#11 – when

We often need to append a where clause to a query based on some conditional, for example, a Request parameter. Instead of if statements you can use the when method:

User::query()->when($request->searchTerm, function ($query) {
  $query=->where('username', 'LIKE', "%$request->searchTerm%"); })->get(); 

It will only run the callback if the first parameter returns true.

The same snippet without when :

$query = User::query();

if ($request->searchTerm) {
  $query->where('username', 'LIKE', "%$request->searchTerm%");
return $query->get();

#12 – appends

If you have an attribute accessor and you often need it when the model is converted into JSON you can use the $appends property:

class Product extends Model {

  protected $appends = ['current_price'];
  public function getCurrentPriceAttribute(): float {
    return $this->prices
        ->where('from', '<=' now())
        ->where('to', '>=', now())

Now the current_price column will be appended to the Product model every time it gets converted into JSON. It's useful when you're working with Blade templates. With APIs, I would stick to Resources.

#13 – whereRelation

Imagine you're working on a financial application like a portfolio tracker, and you have the following models:

A Holding belongs to a Stock and a Stock has many Holdings. You can write a simple join to get every Apple holdings, for example:

$apple = Holding::select('holdings.*')
->leftJoin('stocks', '', 'holdings.stock_id')
->where('stocks.ticker', 'AAPL')

Or you can use the whereRelation helper:

$apple = Holding::whereRelation('stock', 'ticker', 'AAPL')->get();

It reads like this: give every Holdings where the Stock relation's ticker column is equal to AAPL (this is the ticker symbol of Apple). Under the hood, it will run an EXISTS query. So if you actually need data from the stocks table it's not a good solution.

#14 – whereBelongsTo

Consider this snippet:

public function index(User $user) {
  $sum = Order::where('user_id', $user->id)

It looks good. How about this?

public function index(User $user) {
  $sum = Order::whereBelongsTo($user)

WIth Eloquent you almost speak in English.

#15 – oldestOfMany

There's a special relationship called oldestOfMany . You can use it if you constantly need the oldest model from a hasMany relationship.

In this example, we have an Employee and a Paycheck model. An employee has many paychecks, so this is the basic relationship:

class Employee extends Models {

  public function paychecks() {

      return $this->hasMany(Paycheck::class); 

If you want to get the oldest paycheck you don't have to write a custom query every time, you can create a relationship for it.

class Employee extends Models {

  public function oldestPaycheck() {
    return $this->hasOne(Paycheck::class)

In this case, you have to use the hasOne method because it will return only one paycheck, the oldest one. The oldest paycheck is the one with the smallest auto-increment ID. So it won't work if you are using UUIDs as foreign keys.

#16 – latestOfMany

Similarly to oldestOfMany we can use the newestOfMany as well:

class Employee extends Models {

  public function latestPaycheck() {
    return $this->hasOne(Paycheck::class)->latestOfMany();

The latest paycheck is the one with the largest auto-increment ID.

#17 – ofMany

You can also use the ofMany relationship with some custom logic, for example:

class User extends Authenticable { 
  public function mostPopularPost() {
      return $this->hasOne(Post::class)->ofMany('like_count', 'max'); 

This will return the post with the highest like_count .

So instead of writing a custom query every time, you can use these relationships:

#18 – hasManyThrough

Often we have relationships like $parent->child->child . For example, a department has employees, and each employee has paychecks. This is a simple hasMany relationship:

class Department extends Model {

    public function employees(): HasMany {
        return $this->hasMany(Employee::class); 
class Employee extends Model {

    public function paychecks(): HasMany {
         return $this->hasMany(Paycheck::class);

If we need all the paychecks within a department we can write:


Instead of this, you can define a paychecks relationship on the Department model with the hasManyThrough :

class Department extends Model {

  public function employees(): HasMany {
    return $this->hasMany(Employee::class); 
  public function paychecks(): HasManyThrough {  
     return $this->hasManyThrough(Paycheck::class, Employee::class); 

The second argument is the "through" model. And now you simply write:


#19 – hasManyDeep

Okay, this is clickbait, because there is no hasManyDeep relationship in Laravel but there is an excellent package called eloquent-has-many-deep.

Consider the following relationships:

Country -> has many -> User -> has many -> Post -> has many -> Comment

If you try to get every comment from a country with Eloquent it will cost you around 1 billion database queries. But with this package you can query every comment in one query using Eloquent:

class Country extends Model {

use \Staudenmeir\EloquentHasManyDeep\HasRelationships;

  public function comments() {
    return $this->hasManyDeep(Comment::class, [User::class, Post::class]); 

So it can handle additional levels of relationships. Under the hood it uses subqueries.

#20 – withDefault

Let's say you have a Post model in your application that has an Author relationship which is a User model. Users can be deleted but often we need to keep their data. This means Author is a nullable relationship, and probably cause some code like this:

$authorName = $post->author ? $post->author->name
: 'Guest Author'; 

We obviously want to avoid code like this. Fortunately PHP provides us the nullable operator, so we can write this:

$authorName = $post->author?->name || 'Guest Author';

The hardcoded Guest Author seems like an anti-pattern, and you have to write this snippet every time you

need the author's name. In this situation you can use the withDefault :

class Post extends Model {

  public function author(): BelongsTo {
    return $this->belongsTo(User::class)
      ->withDefault(['name' => 'Guest Author']);

Two things happening here:

If the author_id in the Post is null , the author relationship will not return null but a new User model.

The new User model's name is Guest User .

So now, you don't need to check null values anymore:

// Either a real name or 'Guest Author'
$authorName = $post->author->name;

#21 – Order by Related Model's Average

In this example, we have a Book and a Rating model. A book has many ratings. Let's say we need to order the books by the average rating. We can use the withAvg method:

public function bestBooks() {
    ->withAvg('ratings as average_rating', 'rating')

There's a lot of ratings here so let's clear this up:

ratings as average_rating
  - We want the averages from the ratings table
  - We want an alias called average_rating
  - This is the column in the ratings table that want to average

#22 – Eager Loading Specific Columns

select * queries can be slow and memory-consuming. If you want to eager a relationship but you don't need every column, you can specify which ones you want to load:


In this case, Eloquent will run a select id, name from categories query.

#23 – saveMany

With the saveMany function you can save multiple related models in one function call.

Consider the following relationships: a Product has many Prices.

When you update a Product you may want to delete all the prices and save the new ones. In this scenario you can use the saveMany :

$productPrices = collect($prices)
 ->map(fn (array $priceData) => new ProductPrice([ 
    'from_date' => $priceData['fromDate'],
    'to_date' => $priceData['toDate'],
    'price' => $priceData['price'], 


It will create all the prices in one query, and you don't have to deal with loops:

foreach ($productPrices as $price) {

#24 – createMany

Similarly to saveMany, you can also use the createMany if you don't have models, but arrays instead:

$prices = [
    'from' => '2022-01-10',
    'to' => '2022-02-28',
    'price' => 9.99
    'from' => '2022-03-01',
    'to' => null,
    'price' => 14.99

#25 – foreignId & constrained

This is the default way you write a foreign key in a migration:


It's not bad, but here's a more cool approach:


constrained will call: references('id')->on('categories')

Even better, you can do this:


#26 – nullOnDelete

If you have a nullable relationship just use the nullOnDelete helper:


#27 – afterCreating

There's an afterCreating method on the Factory class that you can use to do something after a Model has been created.

When I have Users with profile pictures, I always use this feature:

class UserFactory extends Factory {

  public function definition() {
    return ['username' => $faker->username];

  public function configure()
     return $this->afterCreating(function (User $user) { 

        $faker = FakerFactory::create();
        $dir = storage_path('images');
        $path = $faker->image($dir, 640, 640, null, false);
        $user->profile_picture_path = $dir . DIRECTORY_SEPARATOR . $path;
        $user->save(); });  

#28 – Factory For

Let's say you want to create a Product with a Category:

$category = Category::factory()->create();

$product = Product::factory([
  'category_id' => $category->id

Instead of creating the category and passing it as an attribute, you can do this:

$product = Product::factory()

You can use this method for belongs to relationships.

#29 – Factory Has

With the has method you can do the inverse of the relationship. So you can use this for has many relationships:


You can also specify the relationship's name if it's different from the table name:

->has(ProductPrice::factory(), 'prices');

In this example, the prices is the name of the relationship on the Product model.

#30 – Factory States

When working with factories in tests (or seeders) we often need a specific 'state' in a given model. Let's say we have a Product model and it has an active column.

You can do this:

class ProductFactory extends Factory {

    public function definition() {
        return [
            'name' => $this->faker->words(3, true),
            'description' => $this->faker->paragraph(3),
            'category_id' =>  fn () => Category::factory()->create()->id, 
            'active' => !!rand(0, 1),

With this setup you have to specify the active flag every time you want to set it:

$product = Product::factory([ 'active' !& false])->create();

But factories provide a way to define 'states' for your model. A state can be something like an inactive product:

class ProductFactory extends Factory {

    public function definition() {
        return [
            'name' => $this->faker->words(3, true),
            'description' => $this->faker->paragraph(3),
            'category_id' =>  fn () => Category::factory()->create()->id, 
            'active' => !!rand(0, 1),

    public function inactive(): ProductFactory {
        return $this->state(fn (array $attributes) => [
            'active' => false,

And now you can create an inactive product like this:

$product = Product::factory()->state('inactive')->create();

It's very useful when your state affects 3-4 columns.

Good luck

The end