| 1 |
<?php |
| 2 |
|
| 3 |
namespace FluentCart\App\Services\Report; |
| 4 |
|
| 5 |
use FluentCart\App\App; |
| 6 |
use FluentCart\App\Helpers\Status; |
| 7 |
use FluentCart\Database\Migrations\RetentionSnapshotsMigrator; |
| 8 |
|
| 9 |
class RetentionSnapshotService |
| 10 |
{ |
| 11 |
/** |
| 12 |
* Statuses to exclude from analysis (never really started) |
| 13 |
*/ |
| 14 |
protected array $excludedStatuses = [ |
| 15 |
Status::SUBSCRIPTION_PENDING, |
| 16 |
Status::SUBSCRIPTION_INTENDED, |
| 17 |
'incomplete_expired', |
| 18 |
]; |
| 19 |
|
| 20 |
/** |
| 21 |
* Active statuses |
| 22 |
*/ |
| 23 |
protected array $activeStatuses = [ |
| 24 |
Status::SUBSCRIPTION_ACTIVE, |
| 25 |
Status::SUBSCRIPTION_TRIALING, |
| 26 |
]; |
| 27 |
|
| 28 |
/** |
| 29 |
* Generate retention snapshots |
| 30 |
* |
| 31 |
* @param int|null $productIdFilter Optional product ID to filter by |
| 32 |
* @param callable|null $progressCallback Optional callback for progress updates: fn(string $message, string $level = 'info') |
| 33 |
* @return array Result with 'success', 'message', and 'stats' |
| 34 |
*/ |
| 35 |
public function generate(?int $productIdFilter = null, ?callable $progressCallback = null): array |
| 36 |
{ |
| 37 |
try { |
| 38 |
$this->log('Starting retention snapshot generation...', 'info', $progressCallback); |
| 39 |
|
| 40 |
// Step 1: Ensure table exists |
| 41 |
$this->log('Step 1: Ensuring database table exists...', 'info', $progressCallback); |
| 42 |
$this->ensureTableExists(); |
| 43 |
$this->log('Table ready.', 'success', $progressCallback); |
| 44 |
|
| 45 |
// Step 2: Truncate existing data |
| 46 |
$this->log('Step 2: Clearing existing data...', 'info', $progressCallback); |
| 47 |
$this->truncateTable(); |
| 48 |
$this->log('Data cleared.', 'success', $progressCallback); |
| 49 |
|
| 50 |
// Step 3: Get all customer-product pairs |
| 51 |
$this->log('Step 3: Fetching customer-product pairs...', 'info', $progressCallback); |
| 52 |
$pairs = $this->getCustomerProductPairs($productIdFilter); |
| 53 |
$totalPairs = count($pairs); |
| 54 |
$this->log("Found {$totalPairs} unique customer-product pairs.", 'success', $progressCallback); |
| 55 |
|
| 56 |
if ($totalPairs === 0) { |
| 57 |
$this->log('No data to process. Exiting.', 'warning', $progressCallback); |
| 58 |
return [ |
| 59 |
'success' => false, |
| 60 |
'message' => 'No data to process', |
| 61 |
'stats' => [], |
| 62 |
]; |
| 63 |
} |
| 64 |
|
| 65 |
// Step 4: Get date range |
| 66 |
$this->log('Step 4: Calculating date range...', 'info', $progressCallback); |
| 67 |
$dateRange = $this->getDateRange($productIdFilter); |
| 68 |
$this->log( |
| 69 |
"Date range: {$dateRange['first_month']} to {$dateRange['last_month']} ({$dateRange['total_months']} months)", |
| 70 |
'success', |
| 71 |
$progressCallback |
| 72 |
); |
| 73 |
|
| 74 |
// Step 5 & 6: Build timelines and aggregate snapshots |
| 75 |
$this->log('Step 5 & 6: Building timelines and aggregating snapshots...', 'info', $progressCallback); |
| 76 |
$inserted = $this->buildTimelinesAndAggregate($pairs, $dateRange, $progressCallback); |
| 77 |
$this->log("Aggregation and insertion complete. Total records: {$inserted}", 'success', $progressCallback); |
| 78 |
|
| 79 |
// Get stats |
| 80 |
$stats = $this->getStats(); |
| 81 |
|
| 82 |
return [ |
| 83 |
'success' => true, |
| 84 |
'message' => 'Retention snapshots generated successfully', |
| 85 |
'stats' => $stats, |
| 86 |
]; |
| 87 |
} catch (\Exception $e) { |
| 88 |
$this->log('Error: ' . $e->getMessage(), 'error', $progressCallback); |
| 89 |
return [ |
| 90 |
'success' => false, |
| 91 |
'message' => 'Error: ' . $e->getMessage(), |
| 92 |
'stats' => [], |
| 93 |
]; |
| 94 |
} |
| 95 |
} |
| 96 |
|
| 97 |
/** |
| 98 |
* Log a message |
| 99 |
*/ |
| 100 |
protected function log(string $message, string $level = 'info', ?callable $callback = null): void |
| 101 |
{ |
| 102 |
if ($callback) { |
| 103 |
call_user_func($callback, $message, $level); |
| 104 |
} |
| 105 |
} |
| 106 |
|
| 107 |
/** |
| 108 |
* Ensure the retention snapshots table exists |
| 109 |
*/ |
| 110 |
protected function ensureTableExists(): void |
| 111 |
{ |
| 112 |
RetentionSnapshotsMigrator::migrate(); |
| 113 |
} |
| 114 |
|
| 115 |
/** |
| 116 |
* Truncate the snapshots table |
| 117 |
*/ |
| 118 |
protected function truncateTable(): void |
| 119 |
{ |
| 120 |
App::db()->table('fct_retention_snapshots')->truncate(); |
| 121 |
} |
| 122 |
|
| 123 |
/** |
| 124 |
* Get all unique customer-product pairs |
| 125 |
*/ |
| 126 |
protected function getCustomerProductPairs(?int $productIdFilter = null): array |
| 127 |
{ |
| 128 |
$query = App::db()->query() |
| 129 |
->from('fct_subscriptions') |
| 130 |
->select(['customer_id', 'product_id']) |
| 131 |
->whereNotNull('customer_id') |
| 132 |
->whereNotNull('product_id') |
| 133 |
->where('customer_id', '>', 0) |
| 134 |
->where('product_id', '>', 0) |
| 135 |
->whereNotIn('status', $this->excludedStatuses) |
| 136 |
->groupBy(['customer_id', 'product_id']); |
| 137 |
|
| 138 |
if ($productIdFilter) { |
| 139 |
$query->where('product_id', $productIdFilter); |
| 140 |
} |
| 141 |
|
| 142 |
return $query->get()->toArray(); |
| 143 |
} |
| 144 |
|
| 145 |
/** |
| 146 |
* Get the date range for analysis |
| 147 |
*/ |
| 148 |
protected function getDateRange(?int $productIdFilter = null): array |
| 149 |
{ |
| 150 |
$query = App::db()->query() |
| 151 |
->from('fct_subscriptions') |
| 152 |
->selectRaw('MIN(created_at) as first_date, MAX(created_at) as last_date') |
| 153 |
->whereNotIn('status', $this->excludedStatuses); |
| 154 |
|
| 155 |
if ($productIdFilter) { |
| 156 |
$query->where('product_id', $productIdFilter); |
| 157 |
} |
| 158 |
|
| 159 |
$result = $query->first(); |
| 160 |
|
| 161 |
$firstDate = new \DateTime($result->first_date); |
| 162 |
$lastDate = new \DateTime(); // Use today as the end date |
| 163 |
|
| 164 |
$firstMonth = $firstDate->format('Y-m'); |
| 165 |
$lastMonth = $lastDate->format('Y-m'); |
| 166 |
|
| 167 |
// Calculate total months |
| 168 |
$interval = $firstDate->diff($lastDate); |
| 169 |
$totalMonths = ($interval->y * 12) + $interval->m + 1; |
| 170 |
|
| 171 |
// Generate all months |
| 172 |
$months = []; |
| 173 |
$current = new \DateTime($firstMonth . '-01'); |
| 174 |
$end = new \DateTime($lastMonth . '-01'); |
| 175 |
|
| 176 |
while ($current <= $end) { |
| 177 |
$months[] = $current->format('Y-m'); |
| 178 |
$current->modify('+1 month'); |
| 179 |
} |
| 180 |
|
| 181 |
return [ |
| 182 |
'first_month' => $firstMonth, |
| 183 |
'last_month' => $lastMonth, |
| 184 |
'total_months' => $totalMonths, |
| 185 |
'months' => $months, |
| 186 |
]; |
| 187 |
} |
| 188 |
|
| 189 |
/** |
| 190 |
* Build customer timelines - for each customer-product pair, determine their state in each month |
| 191 |
*/ |
| 192 |
protected function buildCustomerTimelines(array $pairs, array $dateRange, ?callable $progressCallback = null): array |
| 193 |
{ |
| 194 |
$timelines = []; |
| 195 |
|
| 196 |
// OPTIMIZATION: Fetch ALL subscriptions in one query, then group in memory |
| 197 |
$this->log('Fetching all subscriptions in one query...', 'info', $progressCallback); |
| 198 |
|
| 199 |
$allSubscriptions = App::db()->query() |
| 200 |
->from('fct_subscriptions') |
| 201 |
->whereNotNull('customer_id') |
| 202 |
->whereNotNull('product_id') |
| 203 |
->where('customer_id', '>', 0) |
| 204 |
->where('product_id', '>', 0) |
| 205 |
->whereNotIn('status', $this->excludedStatuses) |
| 206 |
->orderBy('created_at', 'ASC') |
| 207 |
->get(); |
| 208 |
|
| 209 |
$this->log('Fetched ' . count($allSubscriptions) . ' subscriptions.', 'info', $progressCallback); |
| 210 |
|
| 211 |
// Fetch last payment dates for all subscriptions in one query |
| 212 |
$this->log('Fetching last payment dates...', 'info', $progressCallback); |
| 213 |
|
| 214 |
$lastPayments = App::db()->query() |
| 215 |
->from('fct_order_transactions') |
| 216 |
->select(['subscription_id', App::db()->raw('MAX(created_at) as last_payment')]) |
| 217 |
->whereNotNull('subscription_id') |
| 218 |
->where('status', 'succeeded') |
| 219 |
->groupBy('subscription_id') |
| 220 |
->get(); |
| 221 |
|
| 222 |
// Index last payments by subscription_id |
| 223 |
$lastPaymentMap = []; |
| 224 |
foreach ($lastPayments as $lp) { |
| 225 |
$lastPaymentMap[$lp->subscription_id] = $lp->last_payment; |
| 226 |
} |
| 227 |
unset($lastPayments); |
| 228 |
|
| 229 |
$this->log('Found last payment dates for ' . count($lastPaymentMap) . ' subscriptions.', 'info', $progressCallback); |
| 230 |
|
| 231 |
// Attach last_payment to each subscription |
| 232 |
foreach ($allSubscriptions as &$sub) { |
| 233 |
$sub->last_payment = isset($lastPaymentMap[$sub->id]) ? $lastPaymentMap[$sub->id] : null; |
| 234 |
} |
| 235 |
unset($sub); |
| 236 |
unset($lastPaymentMap); |
| 237 |
|
| 238 |
$this->log('Grouping by customer-product pairs...', 'info', $progressCallback); |
| 239 |
|
| 240 |
// Group subscriptions by customer_id + product_id in memory |
| 241 |
$grouped = []; |
| 242 |
foreach ($allSubscriptions as $sub) { |
| 243 |
$key = "{$sub->customer_id}_{$sub->product_id}"; |
| 244 |
if (!isset($grouped[$key])) { |
| 245 |
$grouped[$key] = []; |
| 246 |
} |
| 247 |
$grouped[$key][] = $sub; |
| 248 |
} |
| 249 |
|
| 250 |
// Free memory |
| 251 |
unset($allSubscriptions); |
| 252 |
|
| 253 |
$this->log('Grouped into ' . count($grouped) . ' customer-product pairs.', 'info', $progressCallback); |
| 254 |
$this->log('Processing timelines...', 'info', $progressCallback); |
| 255 |
|
| 256 |
foreach ($grouped as $key => $subscriptions) { |
| 257 |
if (empty($subscriptions)) { |
| 258 |
continue; |
| 259 |
} |
| 260 |
|
| 261 |
$firstSub = $subscriptions[0]; |
| 262 |
$customerId = $firstSub->customer_id; |
| 263 |
$productId = $firstSub->product_id; |
| 264 |
|
| 265 |
// Determine cohort (first subscription month) |
| 266 |
$cohort = (new \DateTime($firstSub->created_at))->format('Y-m'); |
| 267 |
|
| 268 |
// Build monthly state |
| 269 |
$monthlyState = []; |
| 270 |
foreach ($dateRange['months'] as $month) { |
| 271 |
$state = $this->getCustomerStateForMonth($subscriptions, $month); |
| 272 |
$monthlyState[$month] = $state; |
| 273 |
} |
| 274 |
|
| 275 |
$timelines[$key] = [ |
| 276 |
'customer_id' => $customerId, |
| 277 |
'product_id' => $productId, |
| 278 |
'cohort' => $cohort, |
| 279 |
'monthly' => $monthlyState, |
| 280 |
]; |
| 281 |
} |
| 282 |
|
| 283 |
return $timelines; |
| 284 |
} |
| 285 |
|
| 286 |
/** |
| 287 |
* Build timelines and aggregate - optimized to fetch all data once but insert in batches |
| 288 |
*/ |
| 289 |
protected function buildTimelinesAndAggregate( |
| 290 |
array $pairs, |
| 291 |
array $dateRange, |
| 292 |
?callable $progressCallback |
| 293 |
): int { |
| 294 |
// Fetch ALL subscriptions in one query |
| 295 |
$this->log('Fetching all subscriptions...', 'info', $progressCallback); |
| 296 |
|
| 297 |
$allSubscriptions = App::db()->query() |
| 298 |
->from('fct_subscriptions') |
| 299 |
->whereNotNull('customer_id') |
| 300 |
->whereNotNull('product_id') |
| 301 |
->where('customer_id', '>', 0) |
| 302 |
->where('product_id', '>', 0) |
| 303 |
->whereNotIn('status', $this->excludedStatuses) |
| 304 |
->orderBy('created_at', 'ASC') |
| 305 |
->get(); |
| 306 |
|
| 307 |
$this->log('Fetched ' . count($allSubscriptions) . ' subscriptions.', 'info', $progressCallback); |
| 308 |
|
| 309 |
// Fetch last payment dates |
| 310 |
$this->log('Fetching last payment dates...', 'info', $progressCallback); |
| 311 |
|
| 312 |
$lastPayments = App::db()->query() |
| 313 |
->from('fct_order_transactions') |
| 314 |
->select(['subscription_id', App::db()->raw('MAX(created_at) as last_payment')]) |
| 315 |
->whereNotNull('subscription_id') |
| 316 |
->where('status', 'succeeded') |
| 317 |
->groupBy('subscription_id') |
| 318 |
->get(); |
| 319 |
|
| 320 |
// Index last payments by subscription_id |
| 321 |
$lastPaymentMap = []; |
| 322 |
foreach ($lastPayments as $lp) { |
| 323 |
$lastPaymentMap[$lp->subscription_id] = $lp->last_payment; |
| 324 |
} |
| 325 |
unset($lastPayments); |
| 326 |
|
| 327 |
// Attach last_payment to each subscription |
| 328 |
foreach ($allSubscriptions as &$sub) { |
| 329 |
$sub->last_payment = $lastPaymentMap[$sub->id] ?? null; |
| 330 |
} |
| 331 |
unset($sub, $lastPaymentMap); |
| 332 |
|
| 333 |
$this->log('Grouping by customer-product pairs...', 'info', $progressCallback); |
| 334 |
|
| 335 |
// Group subscriptions by customer_id + product_id |
| 336 |
$grouped = []; |
| 337 |
foreach ($allSubscriptions as $sub) { |
| 338 |
$key = "{$sub->customer_id}_{$sub->product_id}"; |
| 339 |
if (!isset($grouped[$key])) { |
| 340 |
$grouped[$key] = []; |
| 341 |
} |
| 342 |
$grouped[$key][] = $sub; |
| 343 |
} |
| 344 |
unset($allSubscriptions); |
| 345 |
|
| 346 |
$this->log('Processing timelines and aggregating...', 'info', $progressCallback); |
| 347 |
|
| 348 |
// Build aggregates directly without building full timelines array |
| 349 |
$aggregates = []; |
| 350 |
$processedCount = 0; |
| 351 |
$totalCount = count($grouped); |
| 352 |
|
| 353 |
foreach ($grouped as $key => $subscriptions) { |
| 354 |
if (empty($subscriptions)) { |
| 355 |
continue; |
| 356 |
} |
| 357 |
|
| 358 |
$firstSub = $subscriptions[0]; |
| 359 |
$customerId = $firstSub->customer_id; |
| 360 |
$productId = $firstSub->product_id; |
| 361 |
$cohort = (new \DateTime($firstSub->created_at))->format('Y-m'); |
| 362 |
|
| 363 |
// Get cohort baseline |
| 364 |
$cohortState = $this->getCustomerStateForMonth($subscriptions, $cohort); |
| 365 |
|
| 366 |
// Only include customers who were active at their cohort month |
| 367 |
if (!$cohortState['is_active']) { |
| 368 |
continue; |
| 369 |
} |
| 370 |
|
| 371 |
$cohortMrr = $cohortState['mrr']; |
| 372 |
$cohortDate = new \DateTime($cohort . '-01'); |
| 373 |
|
| 374 |
// Track for each period from cohort onwards |
| 375 |
foreach ($dateRange['months'] as $period) { |
| 376 |
$periodDate = new \DateTime($period . '-01'); |
| 377 |
|
| 378 |
// Only track periods from cohort month onwards |
| 379 |
if ($periodDate < $cohortDate) { |
| 380 |
continue; |
| 381 |
} |
| 382 |
|
| 383 |
$periodState = $this->getCustomerStateForMonth($subscriptions, $period); |
| 384 |
|
| 385 |
// Calculate period offset |
| 386 |
$interval = $cohortDate->diff($periodDate); |
| 387 |
$periodOffset = ($interval->y * 12) + $interval->m; |
| 388 |
|
| 389 |
// Aggregate for specific product |
| 390 |
$this->addToAggregate( |
| 391 |
$aggregates, |
| 392 |
$cohort, |
| 393 |
$period, |
| 394 |
$productId, |
| 395 |
$periodOffset, |
| 396 |
$cohortMrr, |
| 397 |
$periodState |
| 398 |
); |
| 399 |
|
| 400 |
// Aggregate for all products |
| 401 |
$this->addToAggregate( |
| 402 |
$aggregates, |
| 403 |
$cohort, |
| 404 |
$period, |
| 405 |
'all', |
| 406 |
$periodOffset, |
| 407 |
$cohortMrr, |
| 408 |
$periodState |
| 409 |
); |
| 410 |
} |
| 411 |
|
| 412 |
$processedCount++; |
| 413 |
if ($processedCount % 1000 == 0) { |
| 414 |
$this->log("Processed {$processedCount}/{$totalCount} customer-product pairs", 'info', $progressCallback); |
| 415 |
} |
| 416 |
} |
| 417 |
unset($grouped); |
| 418 |
|
| 419 |
// Insert aggregates in batches |
| 420 |
$this->log('Inserting snapshots to database...', 'info', $progressCallback); |
| 421 |
$totalInserted = $this->insertAggregates('fct_retention_snapshots', $aggregates, $progressCallback); |
| 422 |
|
| 423 |
unset($aggregates); |
| 424 |
if (function_exists('gc_collect_cycles')) { |
| 425 |
gc_collect_cycles(); |
| 426 |
} |
| 427 |
|
| 428 |
return $totalInserted; |
| 429 |
} |
| 430 |
|
| 431 |
/** |
| 432 |
* Build timelines and aggregate in batches to minimize memory usage |
| 433 |
* |
| 434 |
* Instead of building ALL timelines in memory, we: |
| 435 |
* 1. Fetch subscriptions for a batch of customer-product pairs |
| 436 |
* 2. Build timelines for that batch |
| 437 |
* 3. Aggregate and insert to DB |
| 438 |
* 4. Clear memory and repeat |
| 439 |
*/ |
| 440 |
protected function buildAndAggregateInBatches( |
| 441 |
array $pairs, |
| 442 |
array $dateRange, |
| 443 |
?callable $progressCallback, |
| 444 |
?int $productIdFilter |
| 445 |
): int { |
| 446 |
$table = 'fct_retention_snapshots'; |
| 447 |
$batchSize = 100; // Process 100 customer-product pairs at a time |
| 448 |
$totalPairs = count($pairs); |
| 449 |
$processedPairs = 0; |
| 450 |
|
| 451 |
// Global aggregates accumulator (across all batches) |
| 452 |
$globalAggregates = []; |
| 453 |
|
| 454 |
// Fetch last payment dates for all subscriptions in one query (this is memory-efficient) |
| 455 |
$this->log('Fetching last payment dates...', 'info', $progressCallback); |
| 456 |
$lastPayments = App::db()->query() |
| 457 |
->from('fct_order_transactions') |
| 458 |
->select(['subscription_id', App::db()->raw('MAX(created_at) as last_payment')]) |
| 459 |
->whereNotNull('subscription_id') |
| 460 |
->where('status', 'succeeded') |
| 461 |
->groupBy('subscription_id') |
| 462 |
->get(); |
| 463 |
|
| 464 |
$lastPaymentMap = []; |
| 465 |
foreach ($lastPayments as $lp) { |
| 466 |
$lastPaymentMap[$lp->subscription_id] = $lp->last_payment; |
| 467 |
} |
| 468 |
unset($lastPayments); |
| 469 |
|
| 470 |
$this->log('Processing in batches of ' . $batchSize . ' customer-product pairs...', 'info', $progressCallback); |
| 471 |
|
| 472 |
// Process pairs in batches |
| 473 |
foreach (array_chunk($pairs, $batchSize) as $batchIndex => $batchPairs) { |
| 474 |
$batchNum = $batchIndex + 1; |
| 475 |
$this->log("Processing batch {$batchNum} ({$processedPairs}/{$totalPairs} pairs)...", 'info', $progressCallback); |
| 476 |
|
| 477 |
// Get customer IDs and product IDs from this batch |
| 478 |
$customerIds = array_unique(array_column($batchPairs, 'customer_id')); |
| 479 |
$productIds = array_unique(array_column($batchPairs, 'product_id')); |
| 480 |
|
| 481 |
// Fetch subscriptions for this batch only |
| 482 |
$query = App::db()->query() |
| 483 |
->from('fct_subscriptions') |
| 484 |
->whereIn('customer_id', $customerIds) |
| 485 |
->whereIn('product_id', $productIds) |
| 486 |
->whereNotNull('customer_id') |
| 487 |
->whereNotNull('product_id') |
| 488 |
->where('customer_id', '>', 0) |
| 489 |
->where('product_id', '>', 0) |
| 490 |
->whereNotIn('status', $this->excludedStatuses) |
| 491 |
->orderBy('created_at', 'ASC'); |
| 492 |
|
| 493 |
$batchSubscriptions = $query->get(); |
| 494 |
|
| 495 |
// Attach last payment dates |
| 496 |
foreach ($batchSubscriptions as &$sub) { |
| 497 |
$sub->last_payment = $lastPaymentMap[$sub->id] ?? null; |
| 498 |
} |
| 499 |
unset($sub); |
| 500 |
|
| 501 |
// Group by customer-product key |
| 502 |
$grouped = []; |
| 503 |
foreach ($batchSubscriptions as $sub) { |
| 504 |
$key = "{$sub->customer_id}_{$sub->product_id}"; |
| 505 |
if (!isset($grouped[$key])) { |
| 506 |
$grouped[$key] = []; |
| 507 |
} |
| 508 |
$grouped[$key][] = $sub; |
| 509 |
} |
| 510 |
unset($batchSubscriptions); |
| 511 |
|
| 512 |
// Build timelines for this batch |
| 513 |
$timelines = []; |
| 514 |
foreach ($grouped as $key => $subscriptions) { |
| 515 |
if (empty($subscriptions)) { |
| 516 |
continue; |
| 517 |
} |
| 518 |
|
| 519 |
$firstSub = $subscriptions[0]; |
| 520 |
$customerId = $firstSub->customer_id; |
| 521 |
$productId = $firstSub->product_id; |
| 522 |
$cohort = (new \DateTime($firstSub->created_at))->format('Y-m'); |
| 523 |
|
| 524 |
// Build monthly state |
| 525 |
$monthlyState = []; |
| 526 |
foreach ($dateRange['months'] as $month) { |
| 527 |
$state = $this->getCustomerStateForMonth($subscriptions, $month); |
| 528 |
$monthlyState[$month] = $state; |
| 529 |
} |
| 530 |
|
| 531 |
$timelines[$key] = [ |
| 532 |
'customer_id' => $customerId, |
| 533 |
'product_id' => $productId, |
| 534 |
'cohort' => $cohort, |
| 535 |
'monthly' => $monthlyState, |
| 536 |
]; |
| 537 |
} |
| 538 |
unset($grouped); |
| 539 |
|
| 540 |
// Aggregate into global aggregates (don't insert yet) |
| 541 |
$this->accumulateAggregates($globalAggregates, $timelines, $dateRange); |
| 542 |
$processedPairs += count($batchPairs); |
| 543 |
|
| 544 |
$this->log("Batch {$batchNum} complete. Processed {$processedPairs}/{$totalPairs} pairs", 'info', $progressCallback); |
| 545 |
|
| 546 |
// Clear memory |
| 547 |
unset($timelines); |
| 548 |
if (function_exists('gc_collect_cycles')) { |
| 549 |
gc_collect_cycles(); |
| 550 |
} |
| 551 |
} |
| 552 |
|
| 553 |
// Now insert all aggregates in batches |
| 554 |
$this->log('All batches processed. Inserting snapshots to database...', 'info', $progressCallback); |
| 555 |
$totalInserted = $this->insertAggregates($table, $globalAggregates, $progressCallback); |
| 556 |
|
| 557 |
unset($globalAggregates); |
| 558 |
if (function_exists('gc_collect_cycles')) { |
| 559 |
gc_collect_cycles(); |
| 560 |
} |
| 561 |
|
| 562 |
return $totalInserted; |
| 563 |
} |
| 564 |
|
| 565 |
/** |
| 566 |
* Accumulate timelines into global aggregates array |
| 567 |
* This merges data from multiple batches without creating duplicates |
| 568 |
*/ |
| 569 |
protected function accumulateAggregates(array &$globalAggregates, array $timelines, array $dateRange): void |
| 570 |
{ |
| 571 |
foreach ($timelines as $timeline) { |
| 572 |
$cohort = $timeline['cohort']; |
| 573 |
$productId = $timeline['product_id']; |
| 574 |
$monthly = $timeline['monthly']; |
| 575 |
|
| 576 |
// Get cohort baseline (state at cohort month) |
| 577 |
$cohortState = $monthly[$cohort] ?? ['is_active' => false, 'mrr' => 0]; |
| 578 |
|
| 579 |
// Only include customers who were active at their cohort month |
| 580 |
if (!$cohortState['is_active']) { |
| 581 |
continue; |
| 582 |
} |
| 583 |
|
| 584 |
$cohortMrr = $cohortState['mrr']; |
| 585 |
|
| 586 |
// Track for each period from cohort onwards |
| 587 |
$cohortDate = new \DateTime($cohort . '-01'); |
| 588 |
|
| 589 |
foreach ($dateRange['months'] as $period) { |
| 590 |
$periodDate = new \DateTime($period . '-01'); |
| 591 |
|
| 592 |
// Only track periods from cohort month onwards |
| 593 |
if ($periodDate < $cohortDate) { |
| 594 |
continue; |
| 595 |
} |
| 596 |
|
| 597 |
$periodState = $monthly[$period] ?? ['is_active' => false, 'mrr' => 0]; |
| 598 |
|
| 599 |
// Calculate period offset |
| 600 |
$interval = $cohortDate->diff($periodDate); |
| 601 |
$periodOffset = ($interval->y * 12) + $interval->m; |
| 602 |
|
| 603 |
// Aggregate for specific product |
| 604 |
$this->addToAggregate( |
| 605 |
$globalAggregates, |
| 606 |
$cohort, |
| 607 |
$period, |
| 608 |
$productId, |
| 609 |
$periodOffset, |
| 610 |
$cohortMrr, |
| 611 |
$periodState |
| 612 |
); |
| 613 |
|
| 614 |
// Aggregate for all products (product_id = NULL) |
| 615 |
$this->addToAggregate( |
| 616 |
$globalAggregates, |
| 617 |
$cohort, |
| 618 |
$period, |
| 619 |
'all', // Will be converted to NULL |
| 620 |
$periodOffset, |
| 621 |
$cohortMrr, |
| 622 |
$periodState |
| 623 |
); |
| 624 |
} |
| 625 |
} |
| 626 |
} |
| 627 |
|
| 628 |
/** |
| 629 |
* Insert aggregates into database in batches |
| 630 |
*/ |
| 631 |
protected function insertAggregates(string $table, array $aggregates, ?callable $progressCallback): int |
| 632 |
{ |
| 633 |
$snapshots = []; |
| 634 |
$batchInsertThreshold = 10000; |
| 635 |
$totalInserted = 0; |
| 636 |
|
| 637 |
// Convert aggregates to final snapshot format and insert in batches |
| 638 |
foreach ($aggregates as $cohort => $periods) { |
| 639 |
foreach ($periods as $period => $products) { |
| 640 |
foreach ($products as $productKey => $data) { |
| 641 |
$productId = $productKey === 'all' ? null : $productKey; |
| 642 |
|
| 643 |
$retentionRateCustomers = $data['cohort_customers'] > 0 |
| 644 |
? round(($data['retained_customers'] / $data['cohort_customers']) * 100, 2) |
| 645 |
: 0; |
| 646 |
|
| 647 |
$retentionRateMrr = $data['cohort_mrr'] > 0 |
| 648 |
? round(($data['retained_mrr'] / $data['cohort_mrr']) * 100, 2) |
| 649 |
: 0; |
| 650 |
|
| 651 |
$snapshots[] = [ |
| 652 |
'cohort' => $cohort, |
| 653 |
'period' => $period, |
| 654 |
'product_id' => $productId, |
| 655 |
'period_offset' => $data['period_offset'], |
| 656 |
'cohort_customers' => $data['cohort_customers'], |
| 657 |
'cohort_mrr' => $data['cohort_mrr'], |
| 658 |
'retained_customers' => $data['retained_customers'], |
| 659 |
'retained_mrr' => $data['retained_mrr'], |
| 660 |
'new_customers' => 0, |
| 661 |
'churned_customers' => $data['cohort_customers'] - $data['retained_customers'], |
| 662 |
'retention_rate_customers' => $retentionRateCustomers, |
| 663 |
'retention_rate_mrr' => $retentionRateMrr, |
| 664 |
'created_at' => current_time('mysql'), |
| 665 |
'updated_at' => current_time('mysql'), |
| 666 |
]; |
| 667 |
|
| 668 |
// Batch insert when threshold is reached |
| 669 |
if (count($snapshots) >= $batchInsertThreshold) { |
| 670 |
$inserted = $this->insertBatch($table, $snapshots); |
| 671 |
$totalInserted += $inserted; |
| 672 |
$this->log("Inserted batch of {$inserted} records (total: {$totalInserted})", 'info', $progressCallback); |
| 673 |
|
| 674 |
// Clear memory |
| 675 |
$snapshots = []; |
| 676 |
if (function_exists('gc_collect_cycles')) { |
| 677 |
gc_collect_cycles(); |
| 678 |
} |
| 679 |
} |
| 680 |
} |
| 681 |
} |
| 682 |
} |
| 683 |
|
| 684 |
// Insert remaining snapshots |
| 685 |
if (!empty($snapshots)) { |
| 686 |
$inserted = $this->insertBatch($table, $snapshots); |
| 687 |
$totalInserted += $inserted; |
| 688 |
$this->log("Inserted final batch of {$inserted} records", 'info', $progressCallback); |
| 689 |
} |
| 690 |
|
| 691 |
return $totalInserted; |
| 692 |
} |
| 693 |
|
| 694 |
/** |
| 695 |
* Determine customer's state for a given month |
| 696 |
*/ |
| 697 |
protected function getCustomerStateForMonth($subscriptions, string $month): array |
| 698 |
{ |
| 699 |
$monthStart = new \DateTime($month . '-01'); |
| 700 |
$monthEnd = (clone $monthStart)->modify('last day of this month')->setTime(23, 59, 59); |
| 701 |
|
| 702 |
$isActive = false; |
| 703 |
$mrr = 0; |
| 704 |
|
| 705 |
foreach ($subscriptions as $sub) { |
| 706 |
$createdAt = new \DateTime($sub->created_at); |
| 707 |
|
| 708 |
// Skip subscriptions created after this month |
| 709 |
if ($createdAt > $monthEnd) { |
| 710 |
continue; |
| 711 |
} |
| 712 |
|
| 713 |
// Check if subscription was active during this month |
| 714 |
$wasActiveThisMonth = $this->wasSubscriptionActiveInMonth($sub, $monthStart, $monthEnd); |
| 715 |
|
| 716 |
if ($wasActiveThisMonth) { |
| 717 |
$isActive = true; |
| 718 |
// Use the subscription's recurring amount as MRR (normalized to monthly) |
| 719 |
$mrr = max($mrr, $this->normalizeToMonthlyMrr($sub)); |
| 720 |
} |
| 721 |
} |
| 722 |
|
| 723 |
return [ |
| 724 |
'is_active' => $isActive, |
| 725 |
'mrr' => $mrr, |
| 726 |
]; |
| 727 |
} |
| 728 |
|
| 729 |
/** |
| 730 |
* Check if a subscription was active during a given month |
| 731 |
* |
| 732 |
* A subscription is considered active in a month if: |
| 733 |
* - It was created on or before the end of that month, AND |
| 734 |
* - It had not ended before the start of that month |
| 735 |
* |
| 736 |
* For ended subscriptions, we use the END date (expire_at, or canceled_at + billing period) |
| 737 |
* For active subscriptions, we check if they existed during that month |
| 738 |
*/ |
| 739 |
protected function wasSubscriptionActiveInMonth($sub, \DateTime $monthStart, \DateTime $monthEnd): bool |
| 740 |
{ |
| 741 |
$createdAt = new \DateTime($sub->created_at); |
| 742 |
|
| 743 |
// Subscription must have started by end of month |
| 744 |
if ($createdAt > $monthEnd) { |
| 745 |
return false; |
| 746 |
} |
| 747 |
|
| 748 |
// Determine the subscription's effective end date |
| 749 |
$endDate = $this->getSubscriptionEndDate($sub); |
| 750 |
|
| 751 |
// If no end date, subscription is still ongoing |
| 752 |
// But we need to check if it was created by this month |
| 753 |
if (!$endDate) { |
| 754 |
return $createdAt <= $monthEnd; |
| 755 |
} |
| 756 |
|
| 757 |
// Subscription was active in this month if: |
| 758 |
// - It started on or before the end of this month, AND |
| 759 |
// - It ended on or after the start of this month |
| 760 |
return $createdAt <= $monthEnd && $endDate >= $monthStart; |
| 761 |
} |
| 762 |
|
| 763 |
/** |
| 764 |
* Get the effective end date of a subscription |
| 765 |
* Returns null if subscription is still ongoing |
| 766 |
* |
| 767 |
* We use last_payment + billing_interval as the effective end date. |
| 768 |
* This represents when the subscription actually stopped providing revenue. |
| 769 |
*/ |
| 770 |
protected function getSubscriptionEndDate($sub): ?\DateTime |
| 771 |
{ |
| 772 |
// If subscription is currently active/trialing, it's still ongoing |
| 773 |
if (in_array($sub->status, $this->activeStatuses)) { |
| 774 |
return null; |
| 775 |
} |
| 776 |
|
| 777 |
// For non-active subscriptions, use last_payment + billing_interval as end date |
| 778 |
// This is the most accurate indicator of when the customer stopped paying |
| 779 |
if (!empty($sub->last_payment) && $sub->last_payment !== '0000-00-00 00:00:00') { |
| 780 |
$lastPayment = new \DateTime($sub->last_payment); |
| 781 |
$interval = $this->getBillingIntervalMonths($sub->billing_interval ?? 'yearly'); |
| 782 |
return (clone $lastPayment)->modify("+{$interval} months"); |
| 783 |
} |
| 784 |
|
| 785 |
// Fallback: if no payment history, use created_at + billing_interval |
| 786 |
// (they paid at creation) |
| 787 |
$createdAt = new \DateTime($sub->created_at); |
| 788 |
$interval = $this->getBillingIntervalMonths($sub->billing_interval ?? 'yearly'); |
| 789 |
return (clone $createdAt)->modify("+{$interval} months"); |
| 790 |
} |
| 791 |
|
| 792 |
/** |
| 793 |
* Get billing interval in months |
| 794 |
*/ |
| 795 |
protected function getBillingIntervalMonths(string $interval): int |
| 796 |
{ |
| 797 |
switch ($interval) { |
| 798 |
case 'yearly': |
| 799 |
case 'annual': |
| 800 |
return 12; |
| 801 |
case 'half_yearly': |
| 802 |
return 6; |
| 803 |
case 'quarterly': |
| 804 |
return 3; |
| 805 |
case 'weekly': |
| 806 |
return 1; // Approximate to 1 month |
| 807 |
case 'daily': |
| 808 |
return 1; // Approximate to 1 month |
| 809 |
default: |
| 810 |
return 1; // monthly |
| 811 |
} |
| 812 |
} |
| 813 |
|
| 814 |
/** |
| 815 |
* Normalize recurring amount to monthly MRR |
| 816 |
*/ |
| 817 |
protected function normalizeToMonthlyMrr($sub): int |
| 818 |
{ |
| 819 |
$amount = (int) ($sub->recurring_amount ?? 0); |
| 820 |
$interval = $sub->billing_interval ?? 'monthly'; |
| 821 |
|
| 822 |
switch ($interval) { |
| 823 |
case 'yearly': |
| 824 |
case 'annual': |
| 825 |
return (int) round($amount / 12); |
| 826 |
case 'half_yearly': |
| 827 |
return (int) round($amount / 6); |
| 828 |
case 'quarterly': |
| 829 |
return (int) round($amount / 3); |
| 830 |
case 'weekly': |
| 831 |
return (int) round($amount * 4.33); |
| 832 |
case 'daily': |
| 833 |
return (int) round($amount * 30); |
| 834 |
default: |
| 835 |
return $amount; // monthly |
| 836 |
} |
| 837 |
} |
| 838 |
|
| 839 |
/** |
| 840 |
* Aggregate timelines into cohort snapshots |
| 841 |
* Returns the number of snapshots inserted (not the snapshots themselves to save memory) |
| 842 |
*/ |
| 843 |
protected function aggregateSnapshots(array $timelines, array $dateRange, ?callable $progressCallback = null): int |
| 844 |
{ |
| 845 |
$table = 'fct_retention_snapshots'; |
| 846 |
|
| 847 |
$snapshots = []; |
| 848 |
$batchInsertThreshold = 10000; // Insert every 10k records to manage memory |
| 849 |
$totalInserted = 0; |
| 850 |
|
| 851 |
// Initialize snapshot structure |
| 852 |
// We need: by cohort, by period, by product_id (and NULL for all) |
| 853 |
$aggregates = []; |
| 854 |
|
| 855 |
foreach ($timelines as $timeline) { |
| 856 |
$cohort = $timeline['cohort']; |
| 857 |
$productId = $timeline['product_id']; |
| 858 |
$monthly = $timeline['monthly']; |
| 859 |
|
| 860 |
// Get cohort baseline (state at cohort month) |
| 861 |
$cohortState = $monthly[$cohort] ?? ['is_active' => false, 'mrr' => 0]; |
| 862 |
|
| 863 |
// Only include customers who were active at their cohort month |
| 864 |
if (!$cohortState['is_active']) { |
| 865 |
continue; |
| 866 |
} |
| 867 |
|
| 868 |
$cohortMrr = $cohortState['mrr']; |
| 869 |
|
| 870 |
// Track for each period from cohort onwards |
| 871 |
$periodOffset = 0; |
| 872 |
$cohortDate = new \DateTime($cohort . '-01'); |
| 873 |
|
| 874 |
foreach ($dateRange['months'] as $period) { |
| 875 |
$periodDate = new \DateTime($period . '-01'); |
| 876 |
|
| 877 |
// Only track periods from cohort month onwards |
| 878 |
if ($periodDate < $cohortDate) { |
| 879 |
continue; |
| 880 |
} |
| 881 |
|
| 882 |
$periodState = $monthly[$period] ?? ['is_active' => false, 'mrr' => 0]; |
| 883 |
|
| 884 |
// Calculate period offset |
| 885 |
$interval = $cohortDate->diff($periodDate); |
| 886 |
$periodOffset = ($interval->y * 12) + $interval->m; |
| 887 |
|
| 888 |
// Aggregate for specific product |
| 889 |
$this->addToAggregate( |
| 890 |
$aggregates, |
| 891 |
$cohort, |
| 892 |
$period, |
| 893 |
$productId, |
| 894 |
$periodOffset, |
| 895 |
$cohortMrr, |
| 896 |
$periodState |
| 897 |
); |
| 898 |
|
| 899 |
// Aggregate for all products (product_id = NULL) |
| 900 |
$this->addToAggregate( |
| 901 |
$aggregates, |
| 902 |
$cohort, |
| 903 |
$period, |
| 904 |
'all', // Will be converted to NULL |
| 905 |
$periodOffset, |
| 906 |
$cohortMrr, |
| 907 |
$periodState |
| 908 |
); |
| 909 |
} |
| 910 |
} |
| 911 |
|
| 912 |
// Convert aggregates to final snapshot format and insert in batches |
| 913 |
foreach ($aggregates as $cohort => $periods) { |
| 914 |
foreach ($periods as $period => $products) { |
| 915 |
foreach ($products as $productKey => $data) { |
| 916 |
$productId = $productKey === 'all' ? null : $productKey; |
| 917 |
|
| 918 |
$retentionRateCustomers = $data['cohort_customers'] > 0 |
| 919 |
? round(($data['retained_customers'] / $data['cohort_customers']) * 100, 2) |
| 920 |
: 0; |
| 921 |
|
| 922 |
$retentionRateMrr = $data['cohort_mrr'] > 0 |
| 923 |
? round(($data['retained_mrr'] / $data['cohort_mrr']) * 100, 2) |
| 924 |
: 0; |
| 925 |
|
| 926 |
$snapshots[] = [ |
| 927 |
'cohort' => $cohort, |
| 928 |
'period' => $period, |
| 929 |
'product_id' => $productId, |
| 930 |
'period_offset' => $data['period_offset'], |
| 931 |
'cohort_customers' => $data['cohort_customers'], |
| 932 |
'cohort_mrr' => $data['cohort_mrr'], |
| 933 |
'retained_customers' => $data['retained_customers'], |
| 934 |
'retained_mrr' => $data['retained_mrr'], |
| 935 |
'new_customers' => 0, // Can be calculated separately if needed |
| 936 |
'churned_customers' => $data['cohort_customers'] - $data['retained_customers'], |
| 937 |
'retention_rate_customers' => $retentionRateCustomers, |
| 938 |
'retention_rate_mrr' => $retentionRateMrr, |
| 939 |
'created_at' => current_time('mysql'), |
| 940 |
'updated_at' => current_time('mysql'), |
| 941 |
]; |
| 942 |
|
| 943 |
// Batch insert when threshold is reached |
| 944 |
if (count($snapshots) >= $batchInsertThreshold) { |
| 945 |
$inserted = $this->insertBatch($table, $snapshots); |
| 946 |
$totalInserted += $inserted; |
| 947 |
$this->log("Inserted batch of {$inserted} records (total: {$totalInserted})", 'info', $progressCallback); |
| 948 |
|
| 949 |
// Clear memory |
| 950 |
$snapshots = []; |
| 951 |
if (function_exists('gc_collect_cycles')) { |
| 952 |
gc_collect_cycles(); |
| 953 |
} |
| 954 |
} |
| 955 |
} |
| 956 |
} |
| 957 |
} |
| 958 |
|
| 959 |
// Insert remaining snapshots |
| 960 |
if (!empty($snapshots)) { |
| 961 |
$inserted = $this->insertBatch($table, $snapshots); |
| 962 |
$totalInserted += $inserted; |
| 963 |
$this->log("Inserted final batch of {$inserted} records", 'info', $progressCallback); |
| 964 |
} |
| 965 |
|
| 966 |
return $totalInserted; |
| 967 |
} |
| 968 |
|
| 969 |
/** |
| 970 |
* Add data to aggregate array |
| 971 |
*/ |
| 972 |
protected function addToAggregate( |
| 973 |
array &$aggregates, |
| 974 |
string $cohort, |
| 975 |
string $period, |
| 976 |
$productId, |
| 977 |
int $periodOffset, |
| 978 |
int $cohortMrr, |
| 979 |
array $periodState |
| 980 |
): void { |
| 981 |
if (!isset($aggregates[$cohort][$period][$productId])) { |
| 982 |
$aggregates[$cohort][$period][$productId] = [ |
| 983 |
'period_offset' => $periodOffset, |
| 984 |
'cohort_customers' => 0, |
| 985 |
'cohort_mrr' => 0, |
| 986 |
'retained_customers' => 0, |
| 987 |
'retained_mrr' => 0, |
| 988 |
]; |
| 989 |
} |
| 990 |
|
| 991 |
// Always increment cohort baseline |
| 992 |
$aggregates[$cohort][$period][$productId]['cohort_customers']++; |
| 993 |
$aggregates[$cohort][$period][$productId]['cohort_mrr'] += $cohortMrr; |
| 994 |
|
| 995 |
// Increment retained if active in this period |
| 996 |
if ($periodState['is_active']) { |
| 997 |
$aggregates[$cohort][$period][$productId]['retained_customers']++; |
| 998 |
$aggregates[$cohort][$period][$productId]['retained_mrr'] += $periodState['mrr']; |
| 999 |
} |
| 1000 |
} |
| 1001 |
|
| 1002 |
/** |
| 1003 |
* Insert a batch of records |
| 1004 |
*/ |
| 1005 |
protected function insertBatch(string $table, array $batch): int |
| 1006 |
{ |
| 1007 |
if (empty($batch)) { |
| 1008 |
return 0; |
| 1009 |
} |
| 1010 |
|
| 1011 |
App::db()->table($table)->insert($batch); |
| 1012 |
|
| 1013 |
return count($batch); |
| 1014 |
} |
| 1015 |
|
| 1016 |
/** |
| 1017 |
* Get statistics from the generated snapshots |
| 1018 |
*/ |
| 1019 |
public function getStats(): array |
| 1020 |
{ |
| 1021 |
$stats = App::db()->table('fct_retention_snapshots') |
| 1022 |
->selectRaw('COUNT(*) as total_records') |
| 1023 |
->selectRaw('COUNT(DISTINCT cohort) as unique_cohorts') |
| 1024 |
->selectRaw('COUNT(DISTINCT period) as unique_periods') |
| 1025 |
->selectRaw('COUNT(DISTINCT product_id) as unique_products') |
| 1026 |
->first(); |
| 1027 |
|
| 1028 |
return [ |
| 1029 |
'total_records' => (int) ($stats->total_records ?? 0), |
| 1030 |
'unique_cohorts' => (int) ($stats->unique_cohorts ?? 0), |
| 1031 |
'unique_periods' => (int) ($stats->unique_periods ?? 0), |
| 1032 |
'unique_products' => (int) (($stats->unique_products ?? 0) + 1), // Include 'all' |
| 1033 |
]; |
| 1034 |
} |
| 1035 |
} |
| 1036 |
|
